TDA Basin Climate Change
Okavango TDA study
Assessment of hydrological effects of
climate change in the Okavango Basin
Piotr Wolski
October 2009
1
TDA Basin Climate Change
Okavango TDA study
Assessment of hydrological effects of climate
change in the Okavango Basin
Piotr Wolski, PhD
Contributions on combined climate change development impacts by H Beuster
Cape Town, 29 October 2009
2
TDA Basin Climate Change
Table of contents
1
BACKGROUND AND OBJECTIVES ...................................................................................................... 8
2
INTRODUCTION ....................................................................................................................................... 9
2.1.1
Greenhouse gas emissions .............................................................................................................. 9
2.1.2
Differences in climates simulated by various GCMs .................................................................... 11
2.1.3
Disparity of spatial and temporal scales ...................................................................................... 12
2.1.4
Natural climate variability ............................................................................................................ 13
2.2 METHODS OF INCORPORATING CLIMATE CHANGE SIGNAL IN HYDROLOGICAL PROJECTIONS ................... 13
3
PREVIOUS WORK IN THE OKAVANGO .......................................................................................... 14
4
MATERIALS AND METHODS.............................................................................................................. 15
4.1 GENERAL APPROACH ............................................................................................................................... 15
4.2 CLIMATE AND CLIMATE MODELS DATA ................................................................................................... 16
4.3 QUALITY CONTROL OF DATASETS ........................................................................................................... 17
4.4 ANALYSES OF CONSISTENCY OF DATASETS ............................................................................................. 18
4.5 ANALYSES OF PAST CHANGE AND VARIABILITY ...................................................................................... 18
4.6 GCM DATA PROCESSING ......................................................................................................................... 19
4.7 STATISTICAL DOWNSCALING OF RAINFALL AND TEMPERATURE DATA .................................................... 20
4.7.1
Analyses of applicability of FEWS rainfall as input to statistical downscaling ............................ 20
4.8 HYDROLOGICAL MODELING .................................................................................................................... 21
4.8.1
Okavango catchment ..................................................................................................................... 21
4.8.2
Okavango Delta ............................................................................................................................ 21
5
EXPLORATORY ANALYSES ............................................................................................................... 23
5.1 COMPARISON OF RAINFALL DATASETS .................................................................................................... 23
5.2 CHANGE AND VARIABILITY IN THE PAST CLIMATE .................................................................................. 25
5.2.1
Temperatures ................................................................................................................................ 25
5.2.2
Rainfall ......................................................................................................................................... 26
5.3 ANALYSIS OF APPLICABILITY OF FEWS DAILY RAINFALL DATASET FOR STATISTICAL DOWNSCALING ... 29
5.4 BIASES OF RAW GCMS AND SD .............................................................................................................. 31
6
RESULTS .................................................................................................................................................. 34
6.1 CLIMATE CHANGE SIGNAL FROM GCM ENSEMBLE ................................................................................. 34
6.2 CLIMATE CHANGE SIGNAL FROM THE SD ENSEMBLE .............................................................................. 35
6.2.1
Okavango Delta ............................................................................................................................ 35
6.2.2
Okavango Catchment .................................................................................................................... 38
6.3 DERIVATION OF CHANGE SIGNAL TO USE IN HYDROLOGICAL MODELING ENVELOPE OF CHANGE .......... 39
6.3.1
Comparison between GCM-derived and SD-derived change signal ............................................ 39
6.3.2
Climate scenarios for hydrological modelling .............................................................................. 42
6.4 RESULTS OF OKAVANGO CATCHMENT MODELLING ................................................................................. 44
6.5 RESULTS OF OKAVANGO DELTA MODELING ........................................................................................... 46
7
COMBINED CLIMATE-DEVELOPMENT SCENARIOS .................................................................. 49
7.1 SUPERIMPOSITION OF CLIMATE CHANGE SCENARIOS ON DEVELOPMENT SCENARIOS ............................ 49
7.2 RESULTS OF OKAVANGO CATCHMENT MODELING ................................................................................. 50
7.3 RESULTS OF OKAVANGO DELTA MODELING ........................................................................................... 54
8
SUMMARY AND DISCUSSION ............................................................................................................ 57
9
REFERENCES .......................................................................................................................................... 60
3
TDA Basin Climate Change
List of figures
Fig. 1 Global warming projected under anthropogenic gas emission scenarios (source IPCC,
2007) ................................................................................................................................ 10
Fig. 2 Monthly rainfall climatologies derived from CRU2.0 and GHCNv2 datasets for
stations in Angola for 1950-1974 period. ........................................................................ 23
Fig. 3 Monthly rainfall climatologies derived from observed data (EPSMO dataset) and
CRU2.0 dataset for stations in Angola (overlapping years in 1961-1999 period)........... 24
Fig. 4 Comparison of monthly rainfall climatologies derived from FEWS (1998-2007) and
GHCN (observed, 1950-1974) datasets for stations in Angola ....................................... 24
Fig. 5 30-year moving average of mean annual air temperature for the three regions of the
Okavango system, CRU2.0 reanalysis data ..................................................................... 26
Fig. 6 30-year moving average of mean annual precipitation for three regions of the
Okavango system (CRU2.0 reanalysis data) and observed rainfall at Maun. ................. 27
Fig. 7 Annual values and 30-year moving average (blue line) of rainfall indices for Maun and
Shakawe (tot total annual rainfall [mm], d_2 number of rain days with rain>2
mm/day, p_50 median daily rainfall [mm], pd_max maximum daily rainfall [mm],
day_5 day after 1 July when 5% of given year's rainfall has fallen) ............................ 28
Fig. 8 30-year moving averages of rainfall indices (d_2 number of raindays with rainfall>2
mm, pd_max maximum daily rainfall, p_50 median daily rainfall, day_5 day when
5% of annual total has fallen) for Maun .......................................................................... 29
Fig. 9 Differences between indices of future rainfall obtained from downscaling based on
observed data and FEWS data, and their statistical significance. Results of downscaling
nine GCMs based on Maun and Shakawe data. ............................................................... 30
Fig. 10 Differences between change factors obtained from downscaled rainfall based on
observed data and FEWS data, and their statistical significance (p-values of t-test).
Results of downscaling nine GCMs based on Maun and Shakawe data. ........................ 31
Fig. 11 Skill of statistical downscaling in replicating rainfall indices. Boxplots show p-values
of t-test performed on mean rainfall indices (tot-total annual rainfall, d_2 number of
days with >2mm.day, p_50 median daily rainfall, pd_max maximum daily rainfall,
day_5 day when 5% of annual total is exceeded) calculated from observed data and
from rainfall data obtained from downscaling of NCEP synoptics. 1979-2007 period,
data for 18 stations in Botswana. ..................................................................................... 32
Fig. 12 Biases in mean annual rainfall for SD (1979-2007, Maun, relative to observed
rainfall) and raw GCM (1961-1999, Delta region, relative to CRU2.0) .......................... 33
Fig. 13 Change (2046-2065 as compared to 1960-1990) in mean temperature and mean
rainfall for three basin zones, on annual and seasonal basis, determined from 21 GCMs.
SRES A2 scenario. Bars denote +/- 1 standard error of difference or ratio of means. .... 34
Fig. 14 Change (2046-2065 as compared to 1960-1990) in standard deviation of monthly
temperature and rainfall for three basin zones, on annual and seasonal basis, determined
from 21 GCMs. SRES A2 scenario. Bars denote +/- 1 standard error. ........................... 35
Fig. 15 Change (2046-2065 as compared to 1960-1990) in mean temperature and mean
rainfall for Maun and Shakawe, on annual and seasonal basis, determined from
downscaled climate. SRES A2 scenario. Bars denote +/- 1 standard error of difference or
ratio of means. .................................................................................................................. 36
Fig. 16 Change (2046-2065 as compared to 1960-1990) in standard deviation of temperature
and rainfall for Maun and Shakawe, on annual and seasonal basis, determined from SD,
SRES A2 scenario. Bars denote +/- 1 standard error. ...................................................... 36
Fig. 17 Changes in rainfall indices (future-past) derived from SD for Maun and Shakawe, and
their statistical significance. ............................................................................................. 37
4
TDA Basin Climate Change
Fig. 18 Changes in rainfall indices derived from SD for locations within lower Okavango
catchment (lat: 15-18 S), and their statistical significance. ............................................. 38
Fig. 19 Changes in rainfall indices derived from SD for locations within upper Okavango
catchment (lat: 12-15 S), and their statistical significance. ............................................. 39
Fig. 20 Comparison of GCM-derived and SD-derived change signal for mean temperature on
annual and seasonal basis. GCM data are for Delta region, SD data are for Maun only.
GCM data only for models used in SD. ........................................................................... 40
Fig. 21 Comparison of GCM-derived and SD-derived change signal for mean annual and
seasonal rainfall. GCM data are for Delta region, SD data are for Maun only. GCM data
only for models used in SD. ............................................................................................. 40
Fig. 22 Boxplots of annual and seasonal change in rainfall obtained from GCMs and SD (9
models) for Maun and Shakawe ...................................................................................... 41
Fig. 23 Boxplots of annual and seasonal change in temperatures obtained from GCMs and SD
(9 models) for Maun and Shakawe .................................................................................. 41
Fig. 24 Boxplots of annual and seasonal change in rainfall for Delta (D), lower Okavango
catchment (L) and upper Okavango catchment (U), SD results, 9 models. Data include
downscaled rainfall for all locations within each zone. ................................................... 42
Fig. 25 Boxplots of annual and seasonal change in temperatures for Delta (D), Lower
catchment (L) and upper catchment (U), GCM results, 21 models ................................. 43
Fig. 26 Increase in PET per 1 deg C increase in temperature, obtained using various ET
calculation methods with temperature, humidity, wind and radiation data from seven
GCMs. .............................................................................................................................. 44
Fig. 27 Okavango flow hydrograph, reference run, "dry", "moderate" and "wet" climate
scenarios ........................................................................................................................... 44
Fig. 28 Flow duration curves of Okavango River at Mohembo, reference, "dry", "moderate"
and "wet" climate scenarios. ............................................................................................ 45
Fig. 29 Mean monthly flows of the Okavango River at Mohembo under reference conditions
and climate scenarios. ...................................................................................................... 45
Fig. 30 Average duration of inundation in the Okavango Delta under a) reference conditions,
b) "dry", c)"moderate" and d) "wet" climate change scenarios ...................................... 47
Fig. 31 Thamalakane flows under reference, "dry", "moderate" and "wet" climate scenarios
.......................................................................................................................................... 47
Fig. 32 Thamalakane flow duration curves under reference, "dry", "moderate" and "wet"
climate scenarios .............................................................................................................. 47
Fig. 33 Mean monthly flows of Thamalakane River at Maun, under reference, "dry",
"moderate" and "wet" climate scenarios ......................................................................... 48
Fig. 34 Mohembo flows under reference conditions and four combined climate change and
development scenarios ..................................................................................................... 53
Fig. 35 Flow duration curves for the Okavango River at Mohembo, under reference and four
combined climate change and development scenarios. ................................................... 53
Fig. 36 Mean monthly hydrograph for the Okavango River at Mohembo, under reference and
four combined climate change and development scenarios ............................................. 53
Fig. 37 Average duration of inundation under a) reference, b) low development, dry climate,
c) low development, wet climate, d) medium development, dry climate and e) medium
development, wet climate scenarios. ............................................................................... 55
Fig. 38 Thamalakane River flows under reference conditions and four combined climate
change and development scenarios .................................................................................. 55
Fig. 39 Flow duration curves of Thamalakane River under reference conditions and four
combined climate change and development scenarios .................................................... 56
5
TDA Basin Climate Change
Fig. 40 Mean monthly flows of Thamalakane River under reference conditions and four
combined climate change and development scenarios .................................................... 56
6
TDA Basin Climate Change
List of tables
Table 1 Summary of greenhouse gas emission scenarios (Nakienovi et al., 2000) ............. 10
Table 2 Rainfall stations used in the analyses ......................................................................... 17
Table 3 Global Climate Models used in the analyses .............................................................. 19
Table 4 GCM data averaging zones ......................................................................................... 20
Table 5 Differences between mean values for rainfall indices calculated based on observed
and FEWS data (period of overlap 1998-2007). P-values of t-test in brackets ............... 24
Table 6 Results of analyses of trend and homogeneity of time series of mean annual
temperature (CRU2.0 reanalysis) for the three regions of the Okavango system ........... 25
Table 7 Results of analysis of trend and homogeneity of various indices of Maun (upper
value in each table row) and Shakawe (lower value in each table row) rainfall series
(1922-2008)...................................................................................................................... 26
Table 8 : Matrix of Climate Change and Development Scenarios .......................................... 50
Table 9 Median values of the ecologically-relevant summary statistics for each climate-
change scenario at Mohembo. PD = Present Day. CC = climate change. CCD = driest
climate change prediction. CCW = wettest climate change prediction. ......................... 52
7
TDA Basin Climate Change
1 Background and objectives
Consideration whether or not human greenhouse-gas emission-related climate change is
happening or will happen in the future, is generally accepted in the mainstream science
(IPCC, 2007), and its discussion is well beyond the scope of this report.
The objective of the study reported here is to determine the effects of climate change on the
Okavango River flow and flooding in the Okavango Delta, based on methods commensurate
with the current state of knowledge and technology in prediction of future global and regional
climates and availability of data for the Okavango region. The hydrological focus causes that
climatological analyses are simplified, and methodologies concentrate on derivation of inputs
for hydrological models (i.e. rainfall and evaporation) that reflect a range of possible future
conditions. These are derived from results of General Circulation Models (GCM) and
statistical downscaling (SD).
This report is organized as follows:
- introduction gives an overview of methods used in climate change impact studies, in
particular addressing uncertainty of the process of derivation of climate change
signal
- review section summarized results of earlier work on climate change in the
Okavango
- materials and methods section describes available datasets and methods used in the
analyses
- exploratory analyses section deals in particular with several issues identified earlier:
o comparison of various datasets
o applicability of satellite-derived rainfall for SD
o analyses of past changes and trends
- results section deals with results of analyses of GCM and SD, as well as results of
hydrological models an includes the following:
o climate change projections based on raw GCM
o climate change projections based on SD
o summary of the above derivation of change signal to be applied in
hydrological models
o results of hydrological modeling
present levels of water use in the basin
climate change superimposed on future water use projections
- summary and discussion
8
TDA Basin Climate Change
2 Introduction
Climate projections are generally provided by physically-based Global or General Circulation
Models (GCMs). These are computer models that numerically describe physical processes
of mass and energy transport in Earth's climate system. State-of-the-art GCMs are coupled
atmosphere-ocean models (CGCMs) that simulate both atmospheric and surface and deep
ocean circulations, with transfer of energy and water (evaporation and precipitation) and
momentum occurring at the sea surface. Space and time is discretized within the GCMs to
form distinct computational blocks i.e. units of space and time, within which properties and
fluxes are considered to be uniform. These models are run with observed energy inputs and
parameters such as surface albedo or concentration of greenhouse gases and aerosols to
reflect past conditions. Projections of greenhouse gas emissions and other factors are then
used to simulate future climate.
GCMs are run by several computational centres in the world, and currently there exists over
20 GCMs, output of which is publicly available through World Climate Research Program
(WCRP) Climate Model Intercomparison Project (CMIP3) at www-pcmdi.llnl.gov and several
other internet data outlets.
GCMs differ in such aspects as size of computational grid and processes that are included in
computations and their parametrization. There are therefore differences between climates
simulated by various GCMs, and by virtue of GCMs simplicity as compared to the real world,
differences between modelled and observed climate. These differences vary in magnitude
with spatial and temporal scales, and for various regions and periods. Climate models are
based on well-established physical principles and have been demonstrated to reproduce
observed features of recent climate and past climate changes. There is considerable
confidence that GCMs provide credible quantitative estimates of future climate change,
particularly at continental scales and above. Confidence in these estimates is higher for
some climate variables (e.g., temperature) than for others (e.g., precipitation) (IPCC, 2007).
Translation of outputs of GCMs to quantitative characteristic of future climate at a given
location is affected by a range of uncertainties related to:
- future greenhouse gas emissions and land use changes.
- differences in future climates simulated by various GCMs
- disparity of spatial and temporal scales at which GCMs are working and at which
information on changes in climate is needed for impact studies.
- natural climate variability that is not simulated (unresolved) by GCMs
2.1.1 Greenhouse gas emissions
Future conditions of the climate system depend not only of internal processes within that
system, but also on anthropogenic emissions of greenhouse gases and aerosols. These
relate to the level of population and economic growth and technological development, which
in the long term respond to environmental, economic or institutional constraints. Due to their
relative unpredictability, in the context of the analyses of climate change these variables are
dealt with by emission scenarios several plausible trajectories of socio-economic and
technological development in the future, detailed in the IPCC Special Report on Emission
Scenarios (SRES, Nakienovi et al., 2000).
9
TDA Basin Climate Change
QuickTimeTM and a
decompressor
are needed to see this picture.
Fig. 1 Global warming projected under anthropogenic gas emission scenarios
(source IPCC, 2007)
Table 1 Summary of greenhouse gas emission scenarios (Nakienovi et al.,
2000)
A1 - more integrated world
* Rapid economic growth.
* A global population that reaches 9 billion in 2050 and then gradually declines.
* The quick spread of new and efficient technologies.
* A convergent world - income and way of life converge between regions. Extensive social
and cultural interactions worldwide.
A2 - a more divided world.
* A world of independently operating, self-reliant nations.
* Continuously increasing population.
* Regionally oriented economic development.
* Slower and more fragmented technological changes and improvements to per capita
income.
B1 - a world more integrated, and more ecologically friendly.
* Rapid economic growth as in A1, but with rapid changes towards a service and
information economy.
* Population rising to 9 billion in 2050 and then declining as in A1.
* Reductions in material intensity and the introduction of clean and resource efficient
technologies.
* An emphasis on global solutions to economic, social and environmental stability.
B2 - a world more divided, but more ecologically friendly.
* Continuously increasing population, but at a slower rate than in A2.
* Emphasis on local rather than global solutions to economic, social and environmental
stability.
* Intermediate levels of economic development.
10
TDA Basin Climate Change
* Less rapid and more fragmented technological change than in A1 and B1.
GHG scenarios differ considerably in magnitude of projected change in climate, particularly
for longer time scales. If treated as equiprobable, they give a large range of possible future
climatic conditions, although major differences are noticeable only towards the end of the
21st century.
2.1.2 Differences in climates simulated by various GCMs
It is well recognized that various GCMs differ in how well they simulate past climate, and in
magnitude of change they simulate for future (IPCC, 2007; Giorgi and Francisco, 2000). The
differences arise because GCMs differ in number of processes they represent, in process
parameterization, resolution of spatial grid and initial conditions.
In early climate impact studies, however, outputs of a single, or only a few GCMs were used
to determine change signal. This was based on an assumption that the selected model is
appropriate for a given location and purpose, but the choice was obviously limited by access
to model data. Such an approach was used in earlier climate work in the Okavango region
(e.g. Wolski et al., 2002; ODMP, 2006). However, since the advance of Climate Model
Intercomparison Project, (CMIP2, Meehl et al., 2000), data from over 20 GCMs are freely
available in a uniform format. This has generated a considerable body of work on multi-
model ensembles and their use in climate change impact studies.
The multi-model methods are predicated on the fundamental belief that no model is the true
model, and there is value in synthesizing projections from an ensemble, even when the
individual models seem to disagree with one another. Predictions for the El Nino Southern
Oscillation (ENSO) and seasonal forecasts from multi-model ensembles are generally found
to be better than single-model forecasts (e.g. Palmer et al., 2005), and allow for
formalization of uncertainty related to the modeling errors.
Formalization of range of uncertainty can be done using several methods. The most robust
method is to derive a minimum and maximum values of possible future climatic parameteres
from the ensemble. However, different models often disagree even on the sign of rainfall
changes expected in particular regions (Giorgi and Francisco, 2000), thus simple range
information is not particularly useful in impact studies, as, particularly in case of rainfall, it
often encompasses both increase and decrease. Determination of consistency between
ensemble members in projecting given direction of change (i.e. information on how many
models project given direction of change) as in regional climate projections of the 4th IPCC
report (IPCC, 2007), provides information on likelihood of that projection.
Other methods are based on model averaging, and provide the most probable scenario with
quantified uncertainty range. These methods can either be deterministic or stochastic and
treat various GCMs as equiprobable (e.g. Palmer and Räisänen, 2002). More recent
approaches take the principle that non-uniform weighting may be more appropriate as
models may have unequal skill or simulation capability regional scale temperature change or
change in other climatic variables such as precipitation (Allen and Ingram, 2002). Giorgi and
Mearns, 2002, developed averaging method based on metrics of model bias and
convergence, while Tebaldi et al., 2005 extended it to the Bayesian approach.
The uncertainty across GCMs is generally larger than that across GHG scenarios (e.g.
Prudhomme et al., 2003) thus using one GHG scenario and multimodel GCM ensemble will
encompass a large proportion of the overall uncertainty
11
TDA Basin Climate Change
2.1.3 Disparity of spatial and temporal scales
A clear mismatch exists between climate and hydrologic modelling in terms of the spatial
and temporal scales, and between GCM accuracy and the hydrological importance of the
variables. In spite of computational development since the advance of GCMs in the 1970s
they remain relatively coarse in resolution and are unable to resolve significant subgrid scale
features such as topography, clouds and land use. For example, the Hadley Centre's
HadCM3 model is resolved at a spatial resolution of 2.5° latitude by 3.75° longitude. Impact
applications require the equivalent of point climate observations and are usually highly
sensitive to fine-scale climate variations that are not explicit in GCMs. Additioanlly, the
reproduction of observed spatial patterns of precipitation (Salathe, 2003) and daily
precipitation variability (Burger and Chen, 2005) is not sufficient. However improved results
can be obtained by the application of even simple downscaling methods (Wilby et al., 1999)
The simplest method is to apply GCM-scale projections in the form of change factors (CFs)
the `perturbation method' or `delta-change' approach (Prudhomme et al., 2002). More
sophisticated approaches to downscaling of large-scale GCM output to a finer spatial
resolution include dynamical and statistical downscaling. The dynamical approach uses a
higher- resolution climate model embedded within a GCM. The statistical approach uses
statistical methods to establish empirical relationships between GCM-resolution climate
variables and local climate.
In the dynamical downscaling, regional climate models (RCMs) use large-scale and lateral
boundary conditions from GCMs to produce higher resolution outputs. RCMs are typically
resolved at the 0.5 ° latitude and longitude scale and parameterize physical atmospheric
processes. Thus, they are able to realistically simulate regional climate features such as
orographic precipitation, extreme climate events and regional scale climate anomalies, or
non-linear effects, such as those associated with the El Nino Southern Oscillation (Leung et
al., 2003). However, model skill depends strongly on biases inherited from the driving GCM
and the presence and strength of regional scale forcings such as orography, land-sea
contrast and vegetation cover. Variability in internal parameterizations also provides
considerable uncertainty.
Statistical downscaling methods rely on the fundamental concept that regional climates are
conditioned by two factors: the large-scale atmospheric state and regional/local
physiographic features (e.g. topography and land use). This relationship may be expressed
as a stochastic and/or deterministic function between large-scale atmospheric variables
(predictors) and local or regional climate variables (predictands) that implicitly parameterizes
the physiographic features influence on regional/local climate. Predictor variables useful for
downscaling typically represent the large-scale circulation, e.g. sea-level pressure and
geopotential heights, but can also include measures of humidity and simulated surface
climate variables such as GCM precipitation and temperature. After the local climate- large
scale atmospheric state relationship is established based on observed data, the large-scale
output from GCMs is fed into this statistical model to estimate corresponding local and
regional climate characteristics.
Statistical methods are more straightforward than dynamical downscaling but tend to
underestimate variance and poorly represent extreme events.
Reviews of methods and applicability of dynamical and statistical downscaling can be found
among others in: Fowler et al., 2007; Hewitson and Crane, 1996; IPCC, 2007; Leung et al.,
2003; Prudhomme et al., 2002; Wilby et al., 1998; Wilby et al., 2004; )
12
TDA Basin Climate Change
2.1.4 Natural climate variability
Climatic phenomena are characterized by natural intrinsic variability, that results form nature
of processes of turbulent mass and energy transport. The ability to represent these
processes and thus the variability in climate models is, obviously, very limited. By they
nature (spatial and temporal scales, set of physical processes represented), GCMs are not
able to project climatic conditions for any particular year, but are considered appropriate to
represent average conditions within a period of time, and 30-year averaging period is usually
taken as representative to these prevalent conditions. However, such an assumption does
not take into consideration that 30-year climates vary naturally. Such variation can be
"random", or forced by internal feedbacks involving processes acting at longer time scales
thermohaline circulation, sea surface temperature anomalies, land use/land cover changes
(Ansell et al., 2000; Chelliah and Bell, 2004; Goosse et al., 2005; IPCC, 2007; Meinke et al.,
2005; Parker et al., 2007; Power et al., 1999). However, only a few studies of climate
change impact formally account for the long-term variation (Arnell, 2003; Hulme et al., 1999,
Sorteberg and Kvamstø, 2006). This is either done by derivation of range of natural
variability from unforced long runs of climate models under pre-industrial concentrations of
GHG (Arnell, 2003; Hulme et al., 1999), or by the analysis of multiple runs of a single GCM
with perturbed initial conditions (Sorteberg and Kvamstø, 2006). If range of natural variability
is comparable with climate change signal, these might not be distinguishable. This is
particularly important in the first part of the 21st century, when climate change signal is not
strong.
2.2 Methods of incorporating climate change signal in hydrological
projections
The simple perturbation, delta or change factor approach (Prudhomme et al., 2002). for
temperature just adds a projected temperature increase to the observed temperature record
to obtain a future temperature time series. Precipitation is usually perturbed by a fraction.
The use of these simple rules implies that only simple changes in the characteristics of these
variables, such as changes in monthly, seasonal or annual means, are taken into account.
Changes in the interannual variability, in the number of precipitation days, in the
autororrelation and in the correlation between the different variables are usually not
considered, although schemes can be appliied that account for changes in some of these
characterisics (Lenderink et al., 2005).
The alternative is the "direct" approach, where output of the GCMs, RCMs or SD is used to
force the hydrological model (e.g. Elshamy et al., 2009). Because of biases present in output
of the climate models, it is usually necessary to perform some corrections prior to feeding
the data into hydrological models, what introduces additional uncertainty. Direct approach
can represent more complex changes in climate, e.g. correlations between the different
climate variables, frequency and persistency of circulation patterns. However, these
potential advantages might turn into a disadvantage when the quality of the climate model
output is not good, i.e. when it does not adequately represent observed variability in the
variables of interest (Diaz-Nieto and Wilby, 2005; Hay et al., 2000; Lenderink et al., 2005).
13
TDA Basin Climate Change
3 Previous work in the Okavango
Several studies have used predictions of future climate by Global Circulation Models to
obtain hydrological and hydro-ecological effects of climate change on the Okavango Delta.
The primary focus of these studies, however, seems to be on procedure and indices used
for assessing the hydrological or ecological change, and not much attention was put on the
actual derivation and uncertainty of climate change signal. To date, no study has used any
downscaling procedure to derive change signal for the Okavango system.
Earlier studies differ in the choice of a GCM (or GCMs) on which the prognoses are based.
Wolski et al., 2002 and ODMP, 2006 focused on analyses of hydrological effects based on
HADCM3 model, which simulates future conditions to be hotter and drier than the past. As a
result, a reduction of flows of the Okavango River and flooding in the Okavango Delta are
projected. Andersson et al., 2006; Murray-Hudson et al., 2006; Wolski and Murray-Hudson,
2008 and Todd et al., 2008 used output of five GCMs: with two simulating future conditions
as wetter, two as drier, and one as similar to conditions observed in the past. Changes in
mean annual rainfall ranging from 10% increase to 15% decrease were projected in these
studies, associated with increase in termperature in the range of 2.5-3.5 deg C. These
changes were projected to result in reduction of Okavango River flows by 20% (for drier
scenarios), or incrase by 25%, with marked seasonal differences. However, neither of the
analyses revealed significant changes in interannual variability or temporal pattern of the
annual flood.
Similar wide range of possible future wetness was obtained by Chiyapo, 2006 who analysed
results of eight GCMs and Milzow et al., 2008, who analysed results of five. Obviously, there
is a strong divergence between the various GCMs available in terms of magnitude and
direction of simulated future rainfall change.
Analyses carried out in the framework of regional assessment for IPCC, 2007 report were
not specifically targeting the Okavango. However, that report contains information extracted
from the multi-model GCM ensemble for southern Africa. This information suggests lack of
consistency between members of the GCM ensemble (21 models) in projecting the direction
of change in rainfall in the Okavango, with slightly more models simulating drier than wetter
conditions. The balance of the "drier-wetter" models shifts towards domination of wetter
models in the northern part of the Okavango catchment. However, maps visualizing results
of downscaling analyeses suggest increase in precipitation over the southern Africa.
Temperature projections are more consistent with increase projected in the range of 2-4 deg
C.
14
TDA Basin Climate Change
4 Materials and methods
4.1 General approach
Mean values of rainfall and temperature were determined for a reference period in the 20th
century and for a future period in the 21st century, based on SD results analyses. This was
done for three equiprobable scenarios capturing the range of climates simulated by an
ensemble of models used in SD. Subsequently, a ratio (for rainfall) and a difference (for
temperature) of means were calculated for each scenario. These change factors (CF) were
used to modify the observed time series of rainfall and temperatures, and the modified
series were in turn used to drive a suite of hydrological models. Results of modeling with
modified input data series were compared to these obtained with non-modified, observed
data series, to assess hydrological effects of considered change, using impact indicators
such as flow duration, mean inundation duration and shape of the discharge hydrograph.
The choice of the approach for this study has been constrained by the availability of data,
nature of existing hydrological models, and climatic and hydrological variability in the
analysed system. Notably:
- work followed the methodology of interval-based change factor. This approach has
been selected over the direct methodology due to:
o the role multidecadal rainfall variability plays in the processes of formation of
runoff and flooding in the Okavango system. There is a strong multidecadal
variability in the Okavango. It is manifested by sequences of above average
or below average rainfall and flows at 40-60 years timescale. Results of SD
are available only for two 20-year intervals (2046-2065 and 2081-2100),
which is dictated by availability of daily GCM output. Hydrological conditions
within such short-term periods would not be directly comparable with past
conditions without an explicit selection of analogous period in the past.
However, the multidecadal variability, particularly its temporal pattern, is not
well represented by GCMs (Wolski, 2009), what makes selection of such an
analogous reference period difficult if not impossible.
o the use of FEWS rainfall dataset for downscaling of GCM output in the
Angolan part of the basin. As shown below, the FEWS dataset introduces an
error into the results of SD. The error becomes insignificant when relative
projections, i.e. projections of change are used instead of direct projections of
future conditions.
o hydrological model of the Okavango catchment utilizing rainfall and potential
evaporation data that are a compilation of various data sources. Application
of direct method would have to involve reconstruction of the input data and
probably recalibration of the model. Considering that the model in its current
status has a history of applications, it was considered more appropriate to
use it as is, instead of reconstructing it. Additionally, time frame of this project
did not allow for the major undertaking of such a reconstruction.
o hydrological models run on the monthly time step, which cancels the potential
direct method's advantage of providing information on change in structure of
daily rainfall.
- the work adopted a simple approach of defining "dry", "moderate" and "wet" climate
scenario to characterize ranges of conditions projected by the ensemble of climate
models. The scenarios are considered here to be equiprobable. This approach has
been selected over a more sophisticated approaches of stochastic or deterministic
model averaging and further probabilistic simulations of climate and hydrology for the
following reasons:
o the number of models for which SD output is available is nine. This low
number hardly justifies in-depth statistical analyses, particularly because it is
15
TDA Basin Climate Change
an "opportunistic" rather than a systematic ensemble, i.e. selected based on
data availability and not on criterion of systematic representation of factors
affecting uncertainty of GCM output.
- The scenarios were defined based on SD results, and not accounted for the range of
future climates simulated by raw GCMs. This was done in spite of lack of agreement
between raw GCM and SD results in magnitude and direction of change in rainfall.
SD method is generally considered to provide results more suitable for hydrological
modeling than raw GCM output. SD is calibrated based on observed data and thus
more likely to accurately reflect forces driving climate change at a specific location
than "generic" GCMs. However, the assessment of which results (raw GCMs or SD)
are right or more probable or realistic, is a major research question and well beyond
the scope of this report.
- Analyses were done under assumption of static vegetation, i.e. no effects of
increased CO2 and temperature on vegetation type, density and biogeochemistry,
and through that on hydrological cycle, are considered.
- In the study, only one greenhouse gas emissions scenario is considered - SRES A2.
Discussion of which SRES scenario is more likely is beyond the scope of this report.
However, the differences between scenarios are relatively small up to the mid 21st
century, which is the main period of interest of this study (Fig. 1). Greenhouse gas
scenario used in this study, SRESA2, represents "business-as-usual" conditions.
This is the only scenario for which GCM output is currently being downscaled for
southern Africa.
4.2 Climate and climate models data
The following datasets were used in the study:
- daily rainfall and temperature data for Maun and Shakawe, obtained from
Department of Meteorological Services, Botswana (Table 2).
- Monthly rainfall data for several stations in Angola (Table 2), compiled from various
sources by EPSMO project.
- Monthly observed rainfall data from Global Historical Climatology Network, version 2
(GHCNv2) available from http://www.ncdc.noaa.gov (Peterson et al., 1998).
- daily gridded rainfall Famine Early Warning System (FEWS) dataset, obtained from
http://www.cpc.noaa.gov. Data covers entire Africa with 0.1 deg resolution, and is
composed of two parts, differing in algorithms used to derive rainfall values: RFE1.0
covering the period 1998/01/01 to 1999/12/31 (Herman et al., 1997), and RFE2.0
covering the period 2000/01/01 till 2008/12/31 (based on Xie and Arkin, 1996). For
the analyses carried here, both datasets were combined.
- CRU2.0 reanalysis monthly gridded rainfall and temperature data. Data has spatial
resolution of 0.1 deg, and covers period Jan 1901- Dec 2000 (Mitchell and Jones,
2005).
- mean monthly values of rainfall, minimum and maximum temperature simulated for
20th century (20C3M) and for 2000-2100 under SRES A2 scenario by 19 GCMs.
These data were obtained from World Climate Research Program (WCRP) Climate
Model Intercomparison Project (CMIP3) multi-model dataset available from www-
pcmdi.llnl.gov. GCMs used are listed in Table 3.
- Results of climate downscaling using Self Organizing Maps (SOM) method for 18
meteorological stations in Botswana, including Maun and Shakawe. Downscaling
results were obtained for nine GCMs (Table 3). The results were provided by Climate
System Analysis Group, University of Cape Town.
16
TDA Basin Climate Change
Table 2 Rainfall stations used in the analyses
Station Source
Long.
Latitude
Period
Type
Botswana
Maun DMS
1921-2009
Daily
Shakawe DMS
1934-2009
Daily
Angola
Lubango
EPSMO
13.56
-14.90
1961-2002* Monthly
Huambo EPSMO
15.45
-12.48
1961-2002* Monthly
Bie EPSMO
16.95
-12.38
1961-2002* Monthly
Moxico EPSMO
17.70
-14.66
1961-2002* Monthly
CACONDA GHCNv2
15.0
-13.2
1950-1974* Monthly
CHENGA GHCNv2
15.0
-13.0
1950-1974* Monthly
CAFU GHCNv2
15.3
-16.3
1950-1974* Monthly
CUIMA GHCNv2
15.5
-13.3
1950-1974* Monthly
RIO_CHIPIA GHCNv2
15.5
-12.3
1950-1974* Monthly
PEREIRA_DECA GHCNv2
15.7
-17.2
1950-1974* Monthly
NOVA_LISBOA GHCNv2
15.7
-12.8
1950-1974* Monthly
MUPA GHCNv2
15.8
-16.1
1950-1974* Monthly
CHIANGA GHCNv2
15.8
-12.7
1950-1974* Monthly
CHINGUAR GHCNv2
16.3
-12.5
1950-1974* Monthly
CHITEMBO GHCNv2
16.7
-13.5
1950-1974* Monthly
CEILUNGA GHCNv2
16.9
-12.3
1950-1974* Monthly
SILVA_PORTA GHCNv2
17.0
-12.4
1950-1974* Monthly
GENERAL_MACHADO GHCNv2
17.6
-12.0
1950-1974* Monthly
SERPA_PINTO/MENONGUE GHCNv2 17.7 -14.7
1950-1974* Monthly
COEMBA GHCNv2
18.0
-12.1
1950-1974* Monthly
CUANGAR GHCNv2
18.5
-17.5
1950-1974* Monthly
CANGAMBA GHCNv2
19.8
-13.6
1950-1974* Monthly
GAGO_COUTINHO GHCNv2
21.4
-14.0
1950-1974* Monthly
*principal data period, but contains numerous gaps
4.3 Quality control of datasets
Rainfall data were screened for typing errors by visual inspections of rainfall vs. time plots,
and identified erroneous values were corrected or replaced with a missing data value. The
general approach was to correct only obvious typing errors, such as these identified by
values exceptionally deviating from the mean (+-3 standard deviations rule) and for example
by the number of significant digits. If error was not obvious, value was left unchanged, but
flagged as possibly erroneous. These flags were later used during interpretation of results of
analyses. Data for Angolan stations obtained from Angolan Dept of Meteorology, in the
period of 1970-1998 often included monthly rainfall values that were in the order of 1-10% of
these recorded in 1950-1970s. Such values were considered erroneous. Furthermore, lack
of any indications at to number of missing data days in that dataset caused that the whole
post 1970 Angolan dataset is considered questionable, even if values seemed realistic.
Systematic analyses of Maun and Shakawe rainfall using double mass plots were carried
out in earlier studies (Scudder et al., 1993, SMEC, 1990), and in these studies several
correction factors were included to account for small quasi-systematic departures of
individual stations from regional trend. This was done in order to tweak performance of
hydrological models using rainfall as input. It was later shown (Gieske, 1997; Wolski et al.,
2006) that such corrections are not needed for adequate modeling of system's hydrology. It
was therefore decided not to correct any rainfall data in this study.
17
TDA Basin Climate Change
Data from other sources (FEWS, GCM, CRU2.0, GHCNv2) were taken as is, but screened
for obvious typing errors, and such were corrected or replaced with missing data value.
4.4 Analyses of consistency of datasets
The issues related to rainfall data availability affect two aspects of this study: analyses of
past changes and variability of rainfall, and data needs for the procedure of statistical
downscaling of GCM outputs. In both aspects data sources alternative to station data were
used and therefore there was a need to assess consistency of available station data and
these alternative data sources.
Analyses of past changes and variability of rainfall are relatively straightforward for
Botswana data, as long-term rainfall datasets are available and are of good integrity. For
Angola, such data are not available. No daily rainfall data for any station were available,
precluding analyses of change in daily rainfall. Monthly station data were of questionable
quality, and could not be used in change and variability analyses. In view of the above,
reanalysis data had be used for such purpose. Consistency between CRU2.0 reanalysis
dataset and observed data was assessed though rainfall climatologies. These were obtained
directly from observed data (both GHCNv2 and EPSMO). For CRU2.0 gridded dataset,
climatologies were calculated based on values for each station's location obtained from
bilinear interpolation from grid points.
Due to lack of daily data for Angola, and short overlap between the observed and FEWS
datasets, consistency between FEWS rainfall and observed rainfall could be assessed only
in terms of monthly climatologies. Rainfall climatologies for observed station data were
plotted against FEWS rainfall climatologies calculated based on values for each station's
location obtained from bilinear interpolation from grid points.
For Botswana stations, where daily rainfall was available for periods overlapping with FEWS
dataset, consistency of FEWS and observed data was analysed in terms of the following
indices: mean annual rainfall, number of rain days with > 2mm rainfall (d_2), median daily
rainfall (p50), maximum daily rainfall (pd_max), and an index representing timing of rainy
season. The last index was taken as number of days since the beginning of climatological
year (1 July) when cumulative rainfall exceeded 5% of total for that rainy season. All the
indices were calculated for each of years for which data were available. Differences between
the datasets were assessed using t-test on mean value of these indices calculated for years
overlapping in both datasets.
4.5 Analyses of past change and variability
Past changes and ranges of variability in rainfall and temperatures were assessed through
analyses of observed station rainfall data and values obtained from CRU2.0 reanalyses.
Data series were tested for linear trend (t-test on significance of trend coefficient),
homogeneity (Buishand test, Buishand, 1982) and significance of differences in means (F-
test) and variances (Levene's test, Fox, 1997) between three non-overlapping 30-year (or
almost-30-year) periods: 1920-1950, 1951-1980 and 1981-2008.
For datasets for which only monthly data were available tests were conducted only on
annual means and variances. For daily datasets, the following indices were also assessed:
number of rain days with > 2mm rainfall (d_2), median daily rainfall (p50), maximum daily
rainfall (pd_max), and an index representing timing of rainy season. The last index was
taken as number of days since the beginning of climatological year (1 July) when cumulative
rainfall exceeded 5% of total for that rainy season.
18
TDA Basin Climate Change
4.6 GCM data processing
For the purpose of analyses of GCM climate change signal, each GCM dataset was
regridded to a uniform resolution of 0.5 deg, and monthly values were spatially averaged
within (rectangular) zones representing Okavango Delta, Lower and Upper catchment of the
Okavango River (Table 4), subsequently coded D, L and U.
Means and variances were calculated for rainfall and temperature for each of the zones and
for each of the GCMs (or GCM runs if data from more than one run were available) on
annual and seasonal basis for reference period (1961-2000) and for "near-future" (2046-
2065). Climate change signal was then obtained for each of the GCMs and for each of the
zones by calculating differences (for temperature) and ratios (for rainfall and variances)
between future and reference period means. If more than one run was available for a model,
change signal used in further analyses was calculated by averaging signal obtained from
individual runs.
Table 3 Global Climate Models used in the analyses
Modelling centre
Model name
Used
in SD
Bjerknes Centre for Climate Research, Norway
BCCR-BCM2.0
National Center for Atmospheric Research, USA
CCSM3
Canadian Centre for Climate Modelling & Analysis, Canada
CGCM3.1(T63)
Y
Météo-France / Centre National de Recherches
CNRM-CM3 Y
Météorologiques, France
CSIRO Atmospheric Research, Australia
CSIRO-Mk3.0
CSIRO-Mk3.5
Y
Max Planck Institute for Meteorology, Germany
ECHAM5/MPI-
Y
OM
Meteorological Institute of the University of Bonn, ECHO-G Y
Meteorological Research Institute of KMA, and Model and Data
group. Germany/Korea
US Dept. of Commerce / NOAA / Geophysical Fluid Dynamics GFDL-CM2.0 Y
Laboratory, USA
US Dept. of Commerce / NOAA / Geophysical Fluid Dynamics GFDL-CM2.1
Laboratory, USA
NASA / Goddard Institute for Space Studies, USA
GISS-ER
Y
Instituto Nazionale di Geofisica e Vulcanologia, Italy
INGV-SXG
Institute for Numerical Mathematics, Russia
INM-CM3.0
Institut Pierre Simon Laplace, France
IPSL-CM4
Y
Center for Climate System Research, National Institute for MIROC3.2(medr
Environmental Studies, and Frontier Research Center for es)
Global Change (JAMSTEC), Japan
Meteorological Research Institute, Japan
MRI-
Y
CGCM2.3.2
National Center for Atmospheric Research, USA
PCM
Hadley Centre for Climate Prediction and Research / Met Office, UKMO-HadCM3
UK
Hadley Centre for Climate Prediction and Research / Met Office, UKMO-
UK
HadGEM1
19
TDA Basin Climate Change
Table 4 GCM data averaging zones
Block Latitude
Longitude
Okavango Delta
18.0-20.5 S 21.0-25.0 E
Lower catchment
15.0-18.0 S 16.5-26.0 E
Upper catchment
12.0-15.0 S 16.5-26.0 E
4.7 Statistical downscaling of rainfall and temperature data
Statistical downscaling (SD) of rainfall and temperature data was obtained using the method
developed at Climate System Analysis Group, University of Cape Town, described in details
by Hewitson and Crane, 2006b. In the method, unique combinations of wind vectors, specific
and relative humidities, and surface temperature are determined for a localized domain
surrounding each location of interest on the basis of NCEP 6-hourly reanalysis data from
1979 to 2007. This is done with the Self Organizing Maps clustering technique. Importantly,
the clustering takes into account not only values of variables, but also their spatial
(circulation) pattern. For each unique atmospheric state, precipitation and temperature
probability density functions (PDF) are derived based on observed precipitation and
temperature data. Future climate states are derived from GCM data and temperature and
precipitation values are drawn at random from the associated respective PDFs to produce
downscaled series at given location.
SD has been done using synoptic variables derived from nine GCMs (Table 3). SD results
are obtained in the form of daily rainfall series for reference period (1979-2000) and for
"near-future" period (2046-2065) for each downscaled station/location.
The downscaling procedure requires at least 10-year long sequence of daily rainfall and
temperature measurements overlapping with the period for which observed (NCEP
reanalysis) synoptic variables are available, i.e. 1979-2007. Such data are available for
Maun and Shakawe (and for several other stations in Botswana), but are not available for
Angola. In order to derive rainfall change signal for Angola, FEWS rainfall series has been
used.
As for change in temperature, analyses for Maun and Shakawe (and other Botswana
stations) show that downscaling procedure gives results that are not significantly different
from results derived from raw GCM output. Station observed temperature data are not
available for Angola, however, it was considered unnecessary to attempt to substitute these
data with an alternative dataset. Instead, temperature change in the Angolan part of the
system was derived from raw GCM output.
Downscaling of GCM rainfall projections was done for locations corresponding to centroids
of subcatchments used in hydrological modeling of the Okavango catchment. Since the
hydrological model treats each of the subcatchments as lumped (i.e. no within-unit spatial
heterogeneity is introduced in inputs and processes), this allows for straightforward
implementation of results of climate change analyses into hydrological modeling. Rainfall
data for each of considered locations have been derived from FEWS dataset and used as
input to downscaling procedure.
4.7.1 Analyses of applicability of FEWS rainfall as input to statistical
downscaling
Due to unavailability of observed daily rainfall time series for Angolan part of the Okavango
catchment, FEWS rainfall dataset was used in the downscaling procedure. FEWS rainfall
carries systematic and non-systematic biases compared to observed station rainfall.
20
TDA Basin Climate Change
Potential errors involved in using FEWS data instead of observed station data as input to
downscaling procedure have been assessed by comparing results of downscaling for Maun
and Shakawe (and other stations in Botswana) based both on station data and FEWS data.
That comparison was carried out through:
a) analyses of differences in downscaled future rainfall indices between FEWS-derived and
observed station-derived downscaled rainfall and
b) analyses of differences in relative change (ratio of future to past) in rainfall indices
between FEWS-derived and observed station-derived downscaled rainfall.
The t-test on means has been used to assess statistical significance of these differences.
4.8 Hydrological modeling
Assessment of hydrological effects of change in climate characteristics was done using two
hydrological models, each simulating a separate part of the Okavango system.
4.8.1 Okavango catchment
The Okavango catchment upstream of Mohembo was simulated using Pitman model
embedded in SPATSIM modeling engine, developed by Institute for Water Research (IWR),
Rhodes University, South Africa. It is a semi-distributed model that represents
subcatchments as s set of storages linked by functions designed to represent the main
hydrological processes: interception, evaporation, infiltration, surface runoff, groundwater
recharge and discharge and flow attenuation in river channels. The model runs on a monthly
time-step. A detailed description of the model can be found in Hughes et al., 2006 and
Hughes, 2004.
Hydrological conditions under projected climate were simulated by modifying monthly rainfall
and evaporation values used in the reference run (1960-2003) for each of the
subcatchments. The modifications were applied separately to each of the subcatchments
and seasons: Dec-Feb (DJF), Mar-May (MAM), Jun-Aug (JJA) and Sep-Nov (SON). In case
of rainfall, change factor was derived directly as a ratio of future to reference period rainfall.
In case of evaporation, change factor was derived as difference of future and reference
period temperatures, separately for minimum and maximum temperatures. The observed
(CRU2.0) temperatures were then modified (location-wise and seasonally) and used for
calculating potential evaporation from Hargreaves (Hargreaves and Samani, 1985) formula.
The modified time series of potential evaporation was used as input to the catchment model.
4.8.2 Okavango Delta
Okavango Delta and Boteti were simulated with a hybrid linear reservoir-GIS model
developed at Harry Oppenheimer Okavango Research Centre, University of Botswana. The
reservoir model is a monthly semi-conceptual, semi-distributed model, where Okavango
Delta is represented by nine large units. Within each of the units a monthly water balance is
calculated accounting for upstream inflow, rainfall, evaporation, surface water-groundwater
interactions and downstream outflow. GIS model allows inundated area obtained as a
lumped value for each of the units of the reservoir model to be represented in the form of a
distributed inundation map. The GIS model is based on the analysis of the time series of
satellite-derived flood maps in a probabilistic setting. Details of the models are presented in
Wolski et al., 2006.
21
TDA Basin Climate Change
Hydrological conditions in the Okavango Delta under projected climate were simulated using
the output form the catchment model and rainfall and potential evaporation data modified in
a similar way to these for the Okavango River catchment.
22
TDA Basin Climate Change
5 Exploratory analyses
5.1 Comparison of rainfall datasets
For stations in Angola there is a relatively good consistency between the observed and
reanalysis data. CRU2.0 dataset has no systematic biases compared to GHCNv2 dataset in
terms of mean monthly rainfall, although there occur non-systematic differences in minimum
and maximum monthly rainfall (Fig. 2 and Fig. 3). For FEWS rainfall dataset mean monthly
rainfall generally agree with observations. However, this dataset seems to underestimate
maximum monthly rainfall, and overestimate minimum monthly rainfall. Part of the
differences can, however, result from the fact that comparison presented in Fig. 4 contains
data from non-overlapping periods. FEWS dataset is known to be systematically biased, and
methods are available for unbiasing it (). The standard method, however, reduces
systematic bias in overall climatology and does not allow for selective correction of lower
and higher than average values. Lack of overlapping observed dataset from Angola,
prevents development of a tailored unbiasing algorithm.
For stations in Botswana part of the system, differences were assessed in terms of indices
derived from daily rainfall (Table 5) obtained from observed and FEWS dataset for identical
period. Although biases do exist, these are mostly statistically insignificant considering
variability of data. The only significant difference is that in terms of number of rain days for
Shakawe station. Since such a difference occurs only in 2 out of 18 Botswana stations (data
not presented), it does not seem to indicate a systematic bias that would require correction.
Fig. 2 Monthly rainfall climatologies derived from CRU2.0 and GHCNv2
datasets for stations in Angola for 1950-1974 period.
23
TDA Basin Climate Change
Fig. 3 Monthly rainfall climatologies derived from observed data (EPSMO
dataset) and CRU2.0 dataset for stations in Angola (overlapping years in 1961-
1999 period)
Fig. 4 Comparison of monthly rainfall climatologies derived from FEWS (1998-
2007) and GHCN (observed, 1950-1974) datasets for stations in Angola
Table 5 Differences between mean values for rainfall indices calculated based
on observed and FEWS data (period of overlap 1998-2007). P-values of t-test in
brackets
Annual
total
Days>2mm
Median
Max daily
Onset of
daily
rains
Maun
-121 (0.21)
-5 (0.28)
0.63 (0.25)
-7 (0.47)
0.9 (0.91)
Shakawe -122
(0.35)
-20(<0.01) 0.14
(0.49)
-0.08 (0.99) 3.9 (0.64)
24
TDA Basin Climate Change
5.2 Change and variability in the past climate
5.2.1 Temperatures
Since no consistent time series of temperature measurements were available that would
allow for assessment of 20th century changes and variability, reanalysis data is used instead.
These data indicate that mean 30-year temperature fluctuated in the 20th century within +-
0.2 deg C of the long-term mean, and that there was no clear overall trend (Fig. 1Fig. 5).
There is an increase in temperature recorded since 1960s that could be attributed to
changing climate. However, a similar increase is present during the period of 1920-1940,
although both the rate of increase and range were smaller than within the post 1960 period.
Pattern of variability is similar in the three regions of the Okavango system.
Statistical analyses of homogeneity of temperature time series show that although there is
no significant overall trend, the series is non-homogeneous in mean and variance (Table 6).
This suggests that factors other than natural short-term variability influenced long-term
characteristics of temperatures in the Okavango region. Determination of causes of this
long-term variability is beyond the scope of this report.
Table 6 Results of analyses of trend and homogeneity of time series of mean
annual temperature (CRU2.0 reanalysis) for the three regions of the Okavango
system
Delta
Lower
Upper
Okavango
Okavango
Mean value
295.8
295.3
294.1
Standard deviation
0.46
0.34
0.33
Trend coefficient (per
0.006 0.005
0.004
year)
Trend significance
0.87 0.89
0.90
(p-value)
Homog. in mean
<0.01 <0.01
<0.01
Buishand test (p-value)
Homog. in variance
<0.01 <0.01
<0.01
Levene test* (p-value)
Homog. in mean, (AOV) <0.01 0.02
0.04
F-test* (p-value)
*test performed splitting the series into three independent groups: 1921-1950, 1951-
1980, 1981-2008
25
TDA Basin Climate Change
Fig. 5 30-year moving average of mean annual air temperature for the three
regions of the Okavango system, CRU2.0 reanalysis data
5.2.2 Rainfall
The general characteristics of Okavango river runoff time series during the period of record
(1920-onward) is such that the first part of the 20th century is relatively dry, mid-century
years are relatively wet, and last years of the century are relatively dry again (Wolski et al.,
2002, Mazvimavi and Wolski, 2006, McCarthy et al., 2000). Similar pattern in observed in
annual rainfall (Fig. 6). A consequence of such variability is that it is erroneous to analyse
past climatic changes from data sets starting in mid 20th century. Such analyses would
reveal trends that do not reflect long-term conditions in the system. When the entire
available series is analysed, no significant (linear) trends are revealed in neither of the
analysed indices (Table 7). The indices mostly appear to be homogeneous in means and
variances, and the only statistically significant departure from homogeneity is revealed for
median daily rainfall for Shakawe. The differences in rainfall indices between various periods
of the 20th century can, therefore, result purely from short-term variability. Temporal patterns
of the indices are not consistent, and do not suggest presence of systematic trends (Fig. 7,
Fig. 8). As a consequence, there is no basis for stating that changes in rainfall observed
since mid 20th century are an expression of anthropogenic change in climate.
Table 7 Results of analysis of trend and homogeneity of various indices of
Maun (upper value in each table row) and Shakawe (lower value in each table
row) rainfall series (1922-2008)
Total
Number
Median
Maximum
Onset of
annual
of rain
daily
daily
rainy season
rainfall
days
rainfall
rainfall
[day after 1
[mm/year]
>2mm
[mm/day] [mm/day]
July]
[days]
26
TDA Basin Climate Change
Mean value
448.6
36
5.4
54.1
102
522.2
38
5.8
57.4
101
Standard deviation 168.6
10
2.1
25.7
16
205.2
12
2.1
19.9
25.9
Trend coefficient
-0.3
-0.008
-0.02
-0.13
-0.05
(per year)
0.85
0.1
-0.01
-0.14
0.36
Trend significance 0.70
0.97
0.86
0.93
0.85
(p-value)
0.43
0.69
0.91
0.67
0.37
Homog. in mean
>0.1
>0.1
>0.1
>0.1
>0.1
Buishand test (p-
>0.1
0.01
0.02
0.07
>0.1
value)
Homog. in
0.32
0.15
0.48
0.80
0.54
variance
0.36
0.41
0.05
0.72
0.67
Levene test* (p-
value)
Homog. in mean,
0.41
0.78
0.05
0.89
0.43
(AOV) F-test* (p-
0.73
0.44
0.001
0.08
0.67
value)
*test performed by splitting the series into three independent groups: 1921-1950,
1951-1980, 1981-2008
Fig. 6 30-year moving average of mean annual precipitation for three regions
of the Okavango system (CRU2.0 reanalysis data) and observed rainfall at
Maun.
27
TDA Basin Climate Change
Fig. 7 Annual values and 30-year moving average (blue line) of rainfall indices
for Maun and Shakawe (tot total annual rainfall [mm], d_2 number of rain
days with rain>2 mm/day, p_50 median daily rainfall [mm], pd_max
maximum daily rainfall [mm], day_5 day after 1 July when 5% of given year's
rainfall has fallen)
28
TDA Basin Climate Change
Fig. 8 30-year moving averages of rainfall indices (d_2 number of raindays
with rainfall>2 mm, pd_max maximum daily rainfall, p_50 median daily
rainfall, day_5 day when 5% of annual total has fallen) for Maun
5.3 Analysis of applicability of FEWS daily rainfall dataset for
statistical downscaling
In order to assess errors involved in the use of FEWS rainfall dataset in the downscaling
procedure, downscaling was done using observed data and FEWS data for the same
locations.
Future rainfall obtained from the downscaling procedure based on FEWS data does not
correspond to that obtained from downscaling based on observed data (Fig. 9). Significant
differences (p-values of t-test predominantly lower than 0.05) were obtained for each
analysed index (Fig. 9). Additionally, the differences are systematic, and their magnitude
changes for various locations (nine left-hand side points in each panel in Fig. 9 are for Maun,
nine right-hand side points are for Shakawe)
Relative change (future-past) obtained from downscaling appears to be more robust with
respect to the input data. Differences between change factors for various indices obtained
from the downscaling procedure based on FEWS and these obtained from downscaling
based on observed data are still present (Fig. 10). However, these differences appear to low
compared to the natural variability, and thus have low statistical significance (p-values in the
order of 0.1 to 1 for most of the cases). Additionally, the differences are not systematic i.e.
there is no consistent over- or underestimation of change factors when FEWS data are used
instead of observed data. Similar results were obtained for 16 stations in Botswana (data not
presented here).
In view of the above, the application of FEWS dataset in downscaling is possible,
however only when relative change in rainfall is derived. If future rainfall obtained
based on FEWS dataset were to be used directly - it is likely that such rainfall would
contain errors that would hinder detection/estimation of climate change effects.
29
TDA Basin Climate Change
Fig. 9 Differences between indices of future rainfall obtained from downscaling
based on observed data and FEWS data, and their statistical significance.
Results of downscaling nine GCMs based on Maun and Shakawe data.
30
TDA Basin Climate Change
Fig. 10 Differences between change factors obtained from downscaled rainfall
based on observed data and FEWS data, and their statistical significance (p-
values of t-test). Results of downscaling nine GCMs based on Maun and
Shakawe data.
5.4 Biases of raw GCMs and SD
The performance of the downscaling procedure was assessed by comparing observed
rainfall with that generated by SD based on the observed synoptic variables. Statistically
insignificant differences suggest good skill of SD method, i.e. that the main large-scale
drivers of local rainfall have been incorporated into the SD procedure. Results of such an
analysis carried out for data from 18 stations in Botswana show that SD reproduces
relatively well timing of rainy season (day_5), number of rain days and median daily rainfall
(Fig. 11). Maximum daily rainfall and total annual rainfall are not well reproduced
downscaling with synoptic variables used during training, produces significantly different
annual total and maximum daily rainfall values than the observed ones. However, the SD
procedure gives lower biases than these present in raw GCMs (Fig. 12).
31
TDA Basin Climate Change
Fig. 11 Skill of statistical downscaling in replicating rainfall indices. Boxplots
show p-values of t-test performed on mean rainfall indices (tot-total annual
rainfall, d_2 number of days with >2mm.day, p_50 median daily rainfall,
pd_max maximum daily rainfall, day_5 day when 5% of annual total is
exceeded) calculated from observed data and from rainfall data obtained from
downscaling of NCEP synoptics. 1979-2007 period, data for 18 stations in
Botswana.
32
TDA Basin Climate Change
Fig. 12 Biases in mean annual rainfall for SD (1979-2007, Maun, relative to
observed rainfall) and raw GCM (1961-1999, Delta region, relative to CRU2.0)
33
TDA Basin Climate Change
6 Results
6.1 Climate change signal from GCM ensemble
Rainfall
GCM-based projections of change in mean rainfall vary broadly within -25% to + 15 %
range, depending on a GCM and season (Fig. 13).
In general there is a tendency of members of the GCM ensemble to project decrease in total
annual rainfall majority of models give change factor smaller than one.
In terms of differences between various parts of the Okavango system, there seems to be a
slight tendency towards "less drying" conditions towards the north i.e. the number of
models projecting decrease in rainfall reduces towards the north.
There is again a lack of consistency between members of the GCM ensemble in terms of
changes in interannual variability of rainfall (taken as changes in standard deviations of
annual or seasonal totals) (Fig. 14).
These results are broadly consistent with the regional-scale assessment presented in IPCC
2007 report (IPCC, 2007). Fig. 11.2 of that report suggests that there is a lack of consistency
in between ensemble GCMs in terms of projected direction of change in the northern part of
the Okavango basin, while in the southern part of the basin, majority of GC models project
decrease in rainfall.
Fig. 13 Change (2046-2065 as compared to 1960-1990) in mean temperature
and mean rainfall for three basin zones, on annual and seasonal basis,
determined from 21 GCMs. SRES A2 scenario. Bars denote +/- 1 standard error
of difference or ratio of means.
34
TDA Basin Climate Change
Fig. 14 Change (2046-2065 as compared to 1960-1990) in standard deviation of
monthly temperature and rainfall for three basin zones, on annual and
seasonal basis, determined from 21 GCMs. SRES A2 scenario. Bars denote +/-
1 standard error.
Temperature
There is a relative consistency in between members of the GCM ensemble in terms of
projections of future temperatures (Fig. 13). Projected is an increase by 1.1-3.5 deg C in
average annual temperatures. Mean seasonal temperatures are projected to increase within
the same range and differences between seasons are minor. Importantly, there is relatively
little spatial variation i.e. projected change in northern part of the basin is similar to that in
the southern part.
There is a consistent increase in standard deviation of annual and seasonal mean
temperature (Fig. 14). This, however, probably reflects an increase in range of temperatures
in the future resulting from systematic trend within the averaging period rather than an
increase in year-to-year variability.
6.2 Climate change signal from the SD ensemble
Statistical downscaling has been done for Maun and Shakawe stations based on observed
daily rainfall and temperatures. As mentioned in earlier, daily rainfall data were not available
for any station in Angolan part of the Okavango system. Daily rainfall derived from FEWS
dataset has been used instead, and downscaling has been done for locations corresponding
to centroids of subcatchments used in the hydrological model of the catchment.
6.2.1 Okavango Delta
Rainfall
35
TDA Basin Climate Change
The results obtained for Maun and Shakawe indicate that the majority of SD ensemble
members project an increase in mean annual rainfall (Fig. 15). There is a marked difference
between seasons. Projections of the ensemble appear to be centered on zero for DJF and
SON, while there is a consistent increase in MAM and JJA. The change in JJA, although
relatively strong, is of little absolute significance because total rainfall during that period is
minimal.
Change in interannual variability of rainfall is not consistent between the members of the
ensemble (Fig. 16). Additionally, the changes are of little or no statistical significance (p-
values of F test on difference in variances are in the range of 0.2-1.0, with only 2 out of 18
cases (nine models and two stations) having p-value of less than 0.05).
Fig. 15 Change (2046-2065 as compared to 1960-1990) in mean temperature
and mean rainfall for Maun and Shakawe, on annual and seasonal basis,
determined from downscaled climate. SRES A2 scenario. Bars denote +/- 1
standard error of difference or ratio of means.
Fig. 16 Change (2046-2065 as compared to 1960-1990) in standard deviation of
temperature and rainfall for Maun and Shakawe, on annual and seasonal basis,
determined from SD, SRES A2 scenario. Bars denote +/- 1 standard error.
36
TDA Basin Climate Change
SD projects consistently an increase in number of rain days, but projections of changes of
other indices, including total annual rainfall, are not consistent between models, and the
changes are not significant (Fig. 17).
Rainfall projections obtained here from SD are broadly consistent with these presented in
the IPCC 2007 report (IPCC, 2007). Fig. 11.3 of that report shows that downscaled GCMs
project increase in rainfall in southern Africa in general and in northern Botswana in
particular.
Fig. 17 Changes in rainfall indices (future-past) derived from SD for Maun and
Shakawe, and their statistical significance.
Temperatures
Changes in mean annual temperature projected by members of the SD ensemble for Maun
and Shakawe fall within the range of 2-3.2 deg C, and there is a difference between seasons
with DJF projected to have increases between 1.5 and 2.8 deg C, while MAM is projected
to have an increase between 2 and 3.5 deg C (Fig. 15).
There is a consistent increase in standard deviation of annual and seasonal means (Fig. 16).
Similarly to the signal obtained from the GCM ensemble, this seems to signify an increase in
range of temperatures related to the trend rather than year-to-year variability.
37
TDA Basin Climate Change
6.2.2 Okavango Catchment
Rainfall
Fig. 18 Changes in rainfall indices derived from SD for locations within lower
Okavango catchment (lat: 15-18 S), and their statistical significance.
38
TDA Basin Climate Change
Fig. 19 Changes in rainfall indices derived from SD for locations within upper
Okavango catchment (lat: 12-15 S), and their statistical significance.
Downscaling results for the Okavango catchment indicate statistically significant increase in
mean annual rainfall (Fig. 18 and Fig. 19) Larger increases are projected to take place in the
north. This increase is projected to be due to increase in number of rain events, change of
which is also statistically significant. Projections of change in other indices are not consistent
in direction and between locations and models, and are not statistically significant.
Temperature
Due to lack of consistent time series of observed daily temperatures for locations in the
Okavango catchment, no downscaling of temperature signal for that region has been
attempted. However, as comparison of differences between temperature signal derived from
SD and from raw GCM data shows these differences are minimal (Fig. 20). Thus, for the
purpose of hydrological modeling it was decided to use in the temperature change signal
derived directly from GCM data.
6.3 Derivation of change signal to use in hydrological modeling
envelope of change
6.3.1 Comparison between GCM-derived and SD-derived change signal
To avoid possible misinterpretation of possible artifacts resulting from the use of FEWS
dataset in downscaling procedure, only downscaling results based on observed data for
Maun and Shakawe is interpreted here. There are rather strong differences between
39
TDA Basin Climate Change
projections of change in rainfall between GCMs and SD methods (Fig. 20, Fig. 21, Fig. 22).
Projections of change in temperatures are, however, similar between methods (Fig. 20, Fig.
21, Fig. 23). Similar results are obtained for 16 stations in Botswana (data not presented
here).
Fig. 20 Comparison of GCM-derived and SD-derived change signal for mean
temperature on annual and seasonal basis. GCM data are for Delta region, SD
data are for Maun only. GCM data only for models used in SD.
Fig. 21 Comparison of GCM-derived and SD-derived change signal for mean
annual and seasonal rainfall. GCM data are for Delta region, SD data are for
Maun only. GCM data only for models used in SD.
40
TDA Basin Climate Change
Fig. 22 Boxplots of annual and seasonal change in rainfall obtained from
GCMs and SD (9 models) for Maun and Shakawe
Fig. 23 Boxplots of annual and seasonal change in temperatures obtained from
GCMs and SD (9 models) for Maun and Shakawe
In general, while GCMs project general decrease in rainfall, SD produces an increase in
rainfall. This is a surprising result as SD procedure is "nested" within the respective GCM.
The set of boundary characteristics of the atmosphere that drives generation of rainfall in
GCMs and in SD are, therefore, identical. Preliminary work suggests, however, that there
are some differences in the set of synoptic variables that GCMs and SD rainfall respond to,
what would explain the discrepancy in direction and magnitude of change. Similar
discrepancy has been obtained in earlier work for South Africa (Hewitson and Crane,
2006a), and in the regional-scale assessment presented in IPCC 2007 report (IPCC, 2007).
41
TDA Basin Climate Change
SD is one of the accepted methods to derive local-scale responses from GCMs. The SD
method used here is a well-established one and has been applied for climate change
assessment in southern Africa (Hewitson and Crane, 2006a; Schulze, 2005). It has been
therefore decided to base assessment of future climatic and hydrological conditions in the
Okavango system on results of SD method.
6.3.2 Climate scenarios for hydrological modelling
Considering the above, it was decided to derive three basic scenarios:
- "dry" - that corresponds to driest conditions (i.e. bottom of the envelope of change in
rainfall and top of the envelope of change in temperature) projected by SD
- "moderate" that corresponds to median conditions (median change in rainfall and
tmedian increase in temperatures) projected by SD
- "wet" that corresponds to wettest conditions (top of the envelope of change in
rainfall and minimum of the envelope of change in temperatures) projected by SD
In further assessment it has been decided to use 25th and 75th percentiles of ensemble
values to define top and bottom of the change envelopes. Considering that SD dataset
comprises results based on nine GCMs, this effectively screens out two lowest and two
highest projections. In this way, both spread and consistency of projections from various
models are taken into account, and the final results are not affected by extreme projections
from a single model. Boxplots showing ranges of change factors for rainfall and temperature
for the three blocks/zones of the Okavango system are presented in Fig. 24 and Fig. 25.
Fig. 24 Boxplots of annual and seasonal change in rainfall for Delta (D), lower
Okavango catchment (L) and upper Okavango catchment (U), SD results, 9
models. Data include downscaled rainfall for all locations within each zone.
42
TDA Basin Climate Change
Fig. 25 Boxplots of annual and seasonal change in temperatures for Delta (D),
Lower catchment (L) and upper catchment (U), GCM results, 21 models
Change in temperatures was recalculated into change in potential evaporation using
Hargreaves (Hargreaves and Samani, 1985) formula. This formula was used because it was
used earlier in the Okavango Delta and the Okavango catchment models to derive potential
evaporation values. Additionally, this formula does not require any other climate data apart
from temperatures to derive potential evaporation. Its application is, therefore,
straightforward. Because of its empirical nature, the formula should implicitly take into
account changes in other climatic factors that are associated with temperature, such as
humidity. Comparison was carried out between the PET increase rate obtained from
Hargreaves formula and from Penman (Doorenbos and Pruitt, 1977) and Penman-Monteith
(Allen et al., 1998) formulas. The two latter ones incorporated changes in relative humidity,
incoming solar radiation, and wind speed, derived from GCMs. The comparison was carried
out only for five GCMs for which the above mentioned variables were available from PCMDI.
Results indicate that Hargreaves formula gives a slight underestimation (in the order of 0.5%
per 1 deg increase in temperature) of potential evaporation compared to the Penman and
Penman-Monteith methods (Fig. 26). Considering data requirements for the two latter
methods, it was decided to use the Hargreaves method.
43
TDA Basin Climate Change
Fig. 26 Increase in PET per 1 deg C increase in temperature, obtained using
various ET calculation methods with temperature, humidity, wind and radiation
data from seven GCMs.
6.4 Results of Okavango catchment modelling
Results of modeling of the Okavango catchment with the Pitman model under climate
scenarios are summarized in Fig. 27 (flow hydrograph), Fig. 28 (flow duration curves) and
Fig. 29 (mean monthly hydrograph). "Dry" climate scenario results in minimal changes to the
Okavango River discharges notably, a slight increase in peak flows. Obviously, change
(increase) in evaporation considered in this scenario is compensated by change (increase)
in rainfall. Under "wet" and "moderate" climate scenarios, discharges of the Okavango River
increase throughout the whole range i.e. both low flows and peak flows are increased.
There are no noticeable effects on timing of flood hydrograph.
Fig. 27 Okavango flow hydrograph, reference run, "dry", "moderate" and "wet"
climate scenarios
44
TDA Basin Climate Change
Fig. 28 Flow duration curves of Okavango River at Mohembo, reference, "dry",
"moderate" and "wet" climate scenarios.
Fig. 29 Mean monthly flows of the Okavango River at Mohembo under
reference conditions and climate scenarios.
45
TDA Basin Climate Change
6.5 Results of Okavango Delta modeling
Results of modeling of the Okavango Delta under climate scenarios are summarized in Fig.
31 through to Fig. 33. "Dry" scenario results in a reduction in average duration of inundation
in the mid and distal parts of the Okavango Delta. This is also associated with a small
reduction in frequency of inundation. Permanently inundated zone reduces and so does the
area subject to seasonal and occasional inundation. These generally drier than reference
conditions occur in spite of a slight increase in inflow and rainfall. In the evaporation-
dominated Delta, unlike in the catchment, the increase in inflow and rainfall does not
compensate for the increase in evaporation, and this results in drier conditions.
Under "moderate" and "wet" increases of inflow and rainfall exceed increase in evaporation,
thus, ultimately, conditions that are wetter than reference occur. Under these scenarios
expansion of the permanently inundated areas and areas subject to long inundation is
observed. There is a relative reduction in areas subject to short inundation.
Thamalakane flows reduce under "dry" climate with no-flow conditions increasing from 29%
of time (under reference conditions) to 39 % of time, but peak flows remain relatively
unaffected. Under "wet" climate, both peak flows and low flows increase, and no-flow
conditions reduce to approximately 13% of time.
There is no change in temporal distribution of Thamalakane flows.
a)
b)
c)
d)
46
TDA Basin Climate Change
Fig. 30 Average duration of inundation in the Okavango Delta under a)
reference conditions, b) "dry", c)"moderate" and d) "wet" climate change
scenarios
Fig. 31 Thamalakane flows under reference, "dry", "moderate" and "wet"
climate scenarios
Fig. 32 Thamalakane flow duration curves under reference, "dry", "moderate"
and "wet" climate scenarios
47
TDA Basin Climate Change
Fig. 33 Mean monthly flows of Thamalakane River at Maun, under reference,
"dry", "moderate" and "wet" climate scenarios
48
TDA Basin Climate Change
7 Combined climate-development scenarios
7.1 Superimposition of Climate Change Scenarios on Development
Scenarios
The preceding analyses have shown that, subject to present levels of water use in the
Okavango catchment, the projected increase in rainfall overcompensates for the projected
increase in evaporation caused by higher temperatures. As a result, an increase in runoff is
projected. In the Delta, for "dry" scenarios, the increase in evaporation and transpiration
may exceed the increase in local rainfall and inflow from the catchment, with an overall
decrease in system's wetness. However, for "wetter" scenarios, overall increase in system's
wetness is projected. These projected changes will only hold if current levels of water
resource development and water use continue into the future. To assess the situation under
likely future water use conditions, the climate modified runoff sequences produced by the
Pitman catchment model were configured in the WEAP systems model to assess the
combined effects of water resource development and climate change futures on the water
resources of the Okavango system.
The water resource development scenarios that were considered include a limited number of
dams, irrigation and hydropower schemes and a low increase in urban and other water
demands in a Low Scenario, approximately representing the present 5-7 year plans of the
three governments. All of these interventions, plus more that represented possible 10-15
year plans were included in the Medium Scenario. The High Scenario added a further
layer of interventions, some of which are probably not realistic. The main purpose of this
final scenario was to `push' the ecosystem as far as possible in terms of development
interventions, to assess if there would be significant ecological and social impacts.
The development scenarios are summarised as follows:
· The Present Day scenario includes all existing water resource developments,
notably:
o About 2 700 ha of irrigation in Namibia
o The urban water demands of Menongue and Cuito Cuanavale (Angola),
Rundu (Namibia), and Maun (Botswana)
· A low water use scenario which is based on the continuation of historical growth in
water demands in the three countries. Growth rates in Angola reflect the recent
acceleration associated with resettlement in de-mined areas. Increased water
consumption is mainly due to growth in urban and rural domestic, livestock and
irrigation water demands. The largest water demands are represented by:
o About 3 100 ha of irrigation in Namibia
o About 18 000 ha of irrigation along the Cuebe River in Angola
o One storage based and three run-of-river hydropower stations in Angola
· A medium scenario which includes
o About 8 400 ha of irrigation in Namibia
o Development of a first phase of the Eastern National Carrier (17 Mm3/a) for
water supply from the Kavango to Grootfontein and Windhoek,
o About 198 000 ha of irrigation at various locations in Angola
o One storage based and four run-of-river hydropower stations in Angola
· A high scenario which includes:
o About 15 000 ha of irrigation in Namibia
o About 338 000 ha of irrigation at various locations in Angola
o Completion of all planned hydropower stations in Angola, i.e. one storage
based and nine run-of-river hydropower stations in Angola ,
49
TDA Basin Climate Change
o Completion of a second phase of the Eastern National Carrier (total capacity
100 Mm3/a),
o Development of a storage based water supply scheme for urban and
industrial water supply from a dam in the Boteti River to Maun.
o At these levels of demand, it was necessary to introduce a hypothetical dam
in the upper basin (Cuchi River) with a capacity of about 500 million m3 to
provide for shortfalls in irrigation water supply and inter-basin transfers.
As can be seen from the above, there are many possible combinations climate-development
water futures. For this assessment, the unlikely High water use scenario and the Moderate
climate change scenarios were not considered. In total, seven combinations of climate
change and water resource development scenarios were assessed. These are shown in
Table 8.
Table 8 : Matrix of Climate Change and Development Scenarios
WATER FUTURES
High
Water
Use
(M)
MD
MW
Medium
water use
Water Use
LD
LW
use (L)
Low water
(R)
RR
Reference
Reference
Dryer (D)
Moderate (M)
Wetter (W)
(R)
Climate Change
7.2 Results of Okavango Catchment Modeling
The WEAP modelling system was used to simulate the combined effects of future water
resource developments and climate futures on flow sequences at points of interest in the
Okavango catchment upstream of Mohembo. WEAP operates on a monthly time step, and
is capable of simulating the operation of water resource developments such as irrigation
scheme and urban abstractions, in channel dams for irrigation water supply, inter-basin
transfers, run-of-river and storage based hydropower schemes. Naturalised (undeveloped)
runoff sequences resulting from climate modified rainfall and temperature were exported
from the Pitman catchment model and used as inflow time series in the WEAP model.
50
TDA Basin Climate Change
Present day (reference) and future water resource developments (Low and Medium water
use scenarios) were then configured in the WEAP model for use in the scenario simulations.
Results of modeling of the Okavango catchment under combined climate-development
scenarios at Mohembo are summarized in Table 9 (flow changes1), Fig. 2734 (flow
hydrograph), Fig. 28 (flow duration curves) and Fig. 29 (mean monthly hydrograph).
In summary, climate change under the driest scenario reverses the loss of mean annual
runoff brought about by the Low and Medium water use scenarios and under the wettest
scenario increases mean annual runoff by up to 20% above present day levels even under
the Medium water use scenario.
The flood season starts about the same time with climate change, except in the Cuito2
where it is up to two months earlier. It also lasts up to three months longer, with the most
extreme case again being the Cuito. At Mohembo there is a 12-16% reduction in peak
under the driest scenario and a very low increase under the wettest. Flood volumes move
back toward present day values in the driest scenario, ameliorating development, and
greatly exceed present day by up to 50% in the wettest. The overall picture is of the flood
season starting a little earlier, lasting longer, having higher flood peaks and providing more
water than present day, particularly in the wettest scenario and the upper basin. The Cuito
shows the most extreme response, with flooding starting up to two months earlier, peaks up
to 20% higher, flood volumes up to 75% higher and the flood season being up to 62%
longer.
The dry season is predicted to begin at about the same time as present day or slightly later
and to become shorter. Climate change partially or completely returns minimum flows to
present day levels even under Medium development. Again, the most dramatic changes are
for the Cuito, where the dry season could be up to 19 weeks shorter with minimum flows up
to 18% higher.
1 A detailed description of the ecological relevance and selection of the flow change indicators is given in
EPSMO/Biokavango Report Number 2; Process Report.
2 Flow change indicators for the IFA sites upstream of Mohembo are provided in EPSMO/Biokavango Report
Number 8; Final IFA Project Report.
51
TDA Basin Climate Change
Table 9
Median values of the ecologically-relevant summary statistics for
each climate-change scenario at Mohembo. PD = Present Day. CC = climate
change. CCD = driest climate change prediction. CCW = wettest climate
change prediction.
Low Medium
Flow Change
PD
No
No
Comment
CCD CCW
CCD CCW
CC
CC
CCD mitigates
development and CCW
Mean Annual
270 261 287 341 245 270 324 goes further, increasing
Runoff (Mcm)
MAR to 20-26% more than
PD
CC mitigates
Dry season
Aug July Aug Aug July July Aug development, with onset
onset
as PD.
CCD partially (Medium) or
completely (Low) mitigates
Dry season
development and CCW
115 130 110 71 145 133 92
duration (days)
goes further, shortening
the dry season by up to
38% of PD
Dry season
CCD and CCW partially or
minimum flow
114 101 113 125 93 107 122 completely mitigate
(m3s-1)
development
Flood season
Jan Jan Jan Dec Jan Jan Jan No
change.
onset
Reduction to 84% of PD in
Flood season
620 618 528 649 611 519 635 CCD and increase by up
peak (m3s-1)
to 4% in CCW.
Under CCD volume moves
back toward or just above
Flood season
5269 4980 5587 7882 4450 5038 7236 PD, and in CCW show a
volume (Mcm)
large increase to 37-50%
above PD
Under CCD duration
moves back toward or
Flood season
150 143 158 190 129 141 178 slightly longer than PD,
duration (days)
and in CCW is longer than
PD by 19-27%
52
TDA Basin Climate Change
Fig. 34 Mohembo flows under reference conditions and four combined climate
change and development scenarios
Fig. 35 Flow duration curves for the Okavango River at Mohembo, under
reference and four combined climate change and development scenarios.
Fig. 36 Mean monthly hydrograph for the Okavango River at Mohembo, under
reference and four combined climate change and development scenarios
53
TDA Basin Climate Change
7.3 Results of Okavango Delta Modeling
Results of modeling of the Okavango Delta under combined climate-development scenarios
are summarized in Fig. 31 through to Fig. 33.
Under the wet climate change and low water use scenario, inundation patterns revert back
to present day levels, with an increase in the average duration of inundation in the mid and
distal parts of the Delta. If the medium water use scenario is superimposed on wet climate
change, the effects of a wetter climate compensate for increased water use, and inundation
patterns closely resemble those associated with present day conditions.
Under the dry climate change scenarios, the drying out of the Delta is exacerbated by
increasing levels of water use, with a moderate shift from permanent swamps to seasonal
swamps and savanna under the Low CC Scenario and a more severe shift under the
Medium CC Scenario to the same conditions predicted for the original High Scenario with no
climate change. The impact on the permanent swamps is limited to some extent by the
location of future water use developments, which are mostly located in the Cubango sub-
basin. Were there to be a shift of development into the Cuito sub-basin (as is the case for
the High water use scenario) which provides the bulk of dry-season inflows into the Delta,
the impact on the permanent swamps would be much more pronounced.
Under the wet climate change scenario, some of the impacts of development on
Thamalakane flows are reversed. Under these climate conditions, flow conditions resemble
the present day situation even with Low and Medium development included. Under the drier
conditions, drying out of the Delta has a corresponding impact of outflows into the
Boteti/Thamalakane system. In the outflow system, the low water use / dry climate change
scenario would resemble the medium water use with no climate change, while the medium
water use / dry climate change represent a condition half way between the medium and high
water use scenarios with no climate change.
a)
54
TDA Basin Climate Change
b)
c)
d) e)
Fig. 37 Average duration of inundation under a) reference, b) low development,
dry climate, c) low development, wet climate, d) medium development, dry
climate and e) medium development, wet climate scenarios.
Fig. 38 Thamalakane River flows under reference conditions and four
combined climate change and development scenarios
55
TDA Basin Climate Change
Fig. 39 Flow duration curves of Thamalakane River under reference conditions
and four combined climate change and development scenarios
Fig. 40 Mean monthly flows of Thamalakane River under reference conditions
and four combined climate change and development scenarios
56
TDA Basin Climate Change
8 Summary and discussion
Analyses carried out in this project, based on data available for the Okavango system,
reveal the following:
Past climate variability and change in the Okavango
- Past rainfall time series (1920-to date) from the Okavango region bears no clear
signatures of changes in total annual rainfall and interannual variability exceeding
ranges of natural variability expected within a 30-year period. For the Okavango
Delta, this statement can be extended to several other rainfall indices such as mean
and maximum daily rainfall and number of rain days, and onset of rainy season.
These statements do not exclude the possibilty of anthropogenic, GHG-driven
change, but recognize that such changes were not stronger than, and thus not
distinguishable from the natural rainfall variabiltiy.
- Temperature records exhibit long-term variability that exceeds ranges of natural
variabiilty expected within a 30-year period. It is beyond the scope of this report to
attempt attrubution of this long-term variaibility.
Projected climate change signal
- Projections of future climate accepted in this report give a general increase in future
temperatures and increase in rainfall. Temperatures are projected to increase by 2.3-
3 deg C, with stronger increase in the south, weaker in the north. Rainfall is projected
to increase by 0-20%. Relative change in rainfall (expressed in %) is projected to be
weaker in the north, and stronger in the south. However, due to north-south rainfall
gradient, these translate to absolute change (in mm) being stronger in the in the
north and weaker in the south.
- Projections accepted here give significant changes only in the total rainfall and in the
number of raindays. No significant shift in timing of rains, rainfall intensity and
interannual variability is projected.
- There are marked differences in magnitude of change between seasons. Strongest
temperature change is projected to occur in Sep-Nov, the weakest in Mar-May. For
rainfall, the strongest increase is projected for Mar-May, the weakest for Sep-Nov.
Projected hydrological change (present levels of water use)
- In the Okavango catchment, the projected increase in rainfall overcompensates for
the projected increase in evaporation caused by higher temperatures. As as result,
an increase in runoff (total and monthly) is projected. The increase in peak flows is
projected to be proportionately stronger than that in low flows.
- In the Delta, for "dry" scenarios, the increase in evaporation and transpiration may
exceed the increase in local rainfall and inflow from the catchment, with an overall
decrease in system's wetness. However, for "wetter" scenarios, overall increase in
system's wetness is projected. The "dry" conditions will manifest by decrease in
frequency and duration of inundation throughout the Delta and in reduction of low
flows in the rivers draining the system, as exemplified by the Thamalakane. The
wetter conditions will show through the increase in duration and frequency of
inundation throughout the Delta, and the increase of high and low flows in the rivers
draining the system.
Projected hydrological change with increased water use
57
TDA Basin Climate Change
- In the Okavango catchment, the projected increase in rainfall overcompensates for
the projected increase in evaporation caused by higher temperatures. Under the dry
climate change scenario the loss of additional water use is reversed for the low and
medium water use scenario. Under the wet scenario mean annual runoff is
increased by up to 20% above present day levels even under the Medium water use
scenario.
- In the Delta, for "dry" scenarios, the increase in evaporation and transpiration may
exceed the increase in local rainfall and inflow from the catchment, with an overall
decrease in system's wetness. This is exacerbated by increased levels of water use
but the impacts would be limited to some extent if future water use development is
concentrated in the Cubango sub-basin. For "wetter" scenarios, the projected overall
increase in system's wetness compensates for increased water use to the extent that
inundation patterns under a medium water use scenario closely resemble those
associated with present day conditions. is projected. Under the wet climate change
scenario, some of the impacts of development on Thamalakane flows are reversed.
Under these climate conditions, flow conditions resemble the present day situation
even with Low and Medium development included. Under the drier conditions, drying
out of the Delta has a corresponding impact of outflows into the Boteti/Thamalakane
system with the medium water use / dry climate scenario representing conditions half
way between the medium and high water use scenarios with no climate change.
The above conclusions are based on datasets and procedures used in the analyses, and are
subject to a range of uncertainties, most importantly:
- Uncertainty in future GHG emissions. The study is based exclusively on the SRES
A2 GHG scenario that represents "business-as-usual" emissions. Other scenarios
are viable, with higher and lower emission levels, however, differences in between
them in the period of interest (near future up to 2060) are rather low. Analyses
carried for GCM data (not presented here) indicate that the use of a different GHG
scenario would affect quantitative results, but will have little bearing on the overall
direction and magnitude of climatic and hydrological change.
- Long term natural variability in climate. The Okavango system is experiencing rather
strong variability on the decadal to multidecadal time scales. Analyses (not presented
here) indicate that GCMs simulate ranges of long term variability comparable to
observations, but fail to represent its temporal pattern. Thus, the effects of long-term
variabilty can potentially be confused with effects of GHG-driven climate change.
However, the effects of long-term variability affect results of a single GCM run, but
will average out when multiple runs are used, and when an ensemble of GCMs is
used. The long-term climate variability is not explicitely accounted for in the results
presented above. However, because of the results being based on the ensemble of
GCM/SD outputs, possible effects of long-term variability are expected to be low, and
affect quantiative results with little influence on the overall magnitude and direction of
projected climatic and hydrological change.
- Data limitations. In this project, no continuous series of daily rainfall and
temperatures were available from Angola, and very limited data were available from
Botswana. The use of substitute data undobtedly introduces uncertainty into final
results.
o Conclusions about past changes are based on the observed station data for
Botswana, and reanalysis data for Angola. Results from Botswana are,
therefore, realistic, while these for Angola can be questioned. Analyses of
consistency between datasets used show that there are no major biases
58
TDA Basin Climate Change
between observed and reanalysis datasets, and that temporal patterns are
consistent. No conclusions are drawn here with respect of rainfall indices that
could not be derived from reanalysis data for Angola.
o In the downscaling procedure, FEWS satallite dataset was used instead of
observations. Inacuracies of that datset caused that only relative change
could be determined, as this was shown to be robust with respect to the
quality of input dataset. Absolute values of future rainfall were likely to be
erroneous. As shown in the analyses carried out, errors involved in the use of
FEWS dataset in the downscaling were not systematic. The use of FEWS
dataset introduces therefore a potential quantitative error into the final results,
but is expected to have no influence on the overall magnitude and direction of
projected climatic and hydrological change.
- Accuracy and adequacy of GCMs and modelling error. This report is based on the
unquestioned assumption that GCMs are suitable tools for modelling future climate. It
also recognizes that there is an error involved in each model's results. In line with
current methodologies of climate change assessment, it utilizes an ensemble of
GCMs/SD results. Such an ensemble defines a range of possible future climates that
encompass modelling errors. Techniques exist to stratify members of the ensemble
according to their reliability, thus narrowing the range of possible climate scenarios.
However, this was not done due to the few models available, and other errors
involved, which the relatively large range of scenarios adopted here is believed to
encompass.
- Accurracy and adequacy of dowscaling procedure.
o The most problematic issue revealed in this study is the disparity between
direction of rainfall change projected by raw GCM data and that projected by
SD. Earlier work with this method revealed similar differences in change
between SD and GCMs to these revealed here, with preservation of spatial
pattern of change. The disparity is not, therefore, unique to the Okavango,
and cannot be an effect of input data artefacts or procedural errors.
Nonetheless, the results of SD method in that matter should be verified,
preferably against independent downscaling method. At the time of writing
this report, CSAG SD method used in this study was the only operational
method available for Africa, and such a verification was beyond the scope of
this project.
o Statistical downscaling procedures have their inherent uncertainties related to
their empirical nature they tend to underestimate change in extremes and
interannual variability. Thus, conclusions related to these indices are to be
treated with reservations.
59
TDA Basin Climate Change
9 References
Allen, M.R. and Ingram, W.J., 2002. Constraints on future changes in climate and the
hydrologic cycle. Nature, 419: 224-232.
Allen, R.G., Pereira, L.S., Raes, D. and Smith, M., 1998. Crop evapotranspiration:
Guidelines for computing crop requirements. Irrigation and Drainage Paper No. 56.,
FAO, Rome.
Andersson, L., Wilk, J., Todd, M., Hughes, D., Earle, A., Kniveton, D., Layberry, R. and
Savenije, H., 2006. Impact of climate change and development scenarios on flow
patterns in the Okavango River. J. Hydrol., 331: 43-57.
Ansell, T.J., Reason, C.J.C., Smith, I.N. and Keay, K., 2000. Evidence for decadal variabitliy
in souther Australian rainfall and relationships with regional pressure and sea surface
temperature. International Journal of Climatology, 20: 1113-1129.
Arnell, N.E., 2003. Relative effects of multi-decadal climatic variability and changes in the
mean and variability of climate due to global warming: future streamflows in Britain.
Journal of Hydrology, 270: 195-213.
Buishand, T.A., 1982. Some methods for testing the homogeneity of rainfall records. Journal
of Hydrology, 58: 11-27.
Chelliah, M. and Bell, G.D., 2004. Tropical multidecadal and interannual climate variability in
the NCEP-NCAR reanalysis. Journal of Climate, 17: 1777-1803.
Chiyapo, D., 2006. Development and application of integrated surface-groundwater model
(ISGM) for assessing impacts of possible climate change on the flooding patterns of
the Okavango Delta wetland system, University of Botswana, Gaborone, 127 pp.
Diaz-Nieto, J. and Wilby, R.L., 2005. A comparison of statistical downscaling and climate
change factor methods: impacts on low flows in the river Thames, United Kingdom.
Climate Change, 69: 245-268.
Doorenbos, J. and Pruitt, W.O., 1977. Crop water requirements. Irrigation and Drainage
Paper No. 24., FAO, Rome, Italy.
Elshamy, M.E., Seierstad, I.A. and Sorteberg, A., 2009. Impacts of climate change on Blue
Nile flows using bias-corrected GCM scenarios. Hydrology and Earth System
Sciences, 13: 551-565.
Fowler, H.J., Blenkinsop, S. and Tebaldi, C., 2007. Linking climate change modelling to
impact studies: recent advaces in downscaling techniques for hydrological
modellling. International Journal of Climatology, 27: 1547-1578.
Fox, J., 1997. Applied Regression, Linear Models, and Related Methods. Sage.
Gieske, A., 1997. Modelling outflow from the Jao/Boro River system in the Okavango Delta,
Botswana. J. Hydrol., 193: 214-239.
Giorgi, F. and Francisco, R., 2000. Evaluating uncertainties in the prediction of regional
climate change. Geophysical Research Letters, 27: 1295-1298.
Giorgi, F. and Mearns, L.O., 2002. Calculation of average, uncertainty range and reliability of
regional climate changes from AOGCM simulations via the "Reliability Ensemble
Averaging" (REA) method. Journa of Climate, 15: 1141-1158.
Goosse, H., Renssen, H., Timmermann, A. and Bradley, R.S., 2005. Internal and forced
climate variability during the last millenium: a model-data comparison using
ensemble simulations. Quarternary Science Reviews, 24: 1345-1360.
Hargreaves, G.H. and Samani, Z.A., 1985. Reference crop evapotranspiration from
temperature. Applied Eng. in Agriculture, 1(2): 96-99.
Hay, L., Wilby, R.L. and Leavesley, G.H., 2000. A comparison of delta change and
downscaled GCM scenarios fo three mountainous basins in United States. J. Am.
Water Res. Assoc., 362: 387-397.
Herman, A., Kumar, V.B., Arkin, P.A. and Kousky, J.V., 1997. Objectively determined 10-day
African rainfall estimates created for famine early warning systems. International
Journal of Remote Sensing, 18(10): 2147-2159.
Hewitson, B.C. and Crane, R.G., 1996. Climate downscaling techniques and applications.
Climate Research, 7: 85-95.
60
TDA Basin Climate Change
Hewitson, B.C. and Crane, R.G., 2006a. Consensus between GCM climate change
projections with empirical downscaling: precipitation downscaling over South Africa.
International Journal of Climatology, 26: 1315-1337.
Hewitson, B.C. and Crane, R.G., 2006b. Consensus between GCM climate change
projections with emprical downscaling: precipitation downscaling over South Africa.
Int. J. Climatology, 26: 1315-1337.
Hughes, D., Andersson, L., Wilk, J. and Savenije, H., 2006. Regional calibration of the
Pitman model for the Okavango River. J. Hydrol., 331: 30-42.
Hughes, D.A., 2004. Incorporating groundwater recharge and discharge functions into an
existing monthly rainfallrunoff model. Hydrological Sciences Journal/Journal des
Sciences Hydrologiques, 49(2).
Hulme, M., Barrow, E.M., Arnell, N.E., Harrison, P.A., Johns, T.C. and Downing, T.E., 1999.
Relative impacts of human-induced climate change and natural climate variability.
Nature, 397: 688-691.
IPCC, 2007. Climate Change 2007: The Physical Science Basis. Contribution of Working
Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate
Change. Cambridge University Press, Cambridge, United Kingdom and New York,
USA, 996 pp.
Lenderink, G., Buishand, A. and van Deursen, W., 2005. Estimates of future discharge of the
river Rhine using two scenario methodologies: direct versus delta approach.
Hydrology and Earth System Sciences.
Leung, L.R., Mearns, L.O., Giorgi, F. and Wilby, R.L., 2003. Regional climate research:
needs and opportunities. Bull. Am. Meteorol. Soc., 84: 89-95.
Mazvimavi, D. and Wolski, P., 2006. Long-term variations of annual flows of the Okavango
and Zambezi Rivers. Physics and Chemistry of the Earth, 31: 944-951.
McCarthy, T.S., Cooper, G.R.J., Tyson, P.D. and Ellery, W.N., 2000. Seasonal flooding in
the Okavango Delta, Botswana - Recent history and future prospects. South African
Journal of Science, 96(1): 25-33.
Meehl, G., Boer, G.J., Covey, C., Latif, M. and Stouffer, R.J., 2000. The Coupled Model
Intercomparison Project (CMIP). Bull. Am. Meteorol. Soc., 81.
Meinke, H., deVoil, P., Hammer, G.L., Power, S., Allan, R., Stone, R.C., Folland, C. and
Potgieter, A., 2005. Rainfall variability at decadal and longer time scales: signal or
noise? Journal of Climate, 18: 89-96.
Milzow, C., Burg, V. and Kinzelbach, W., 2008. Simulating climate change and water
abstractions impacts on the hydrology and ecology of the Okavango Delta, The 2nd
IASTED International Conference on Water Resource Management. IASTED,
Gaborone, Botswana.
Mitchell, T.D. and Jones, P.D., 2005. An improved method of constructing a database of
monthly climate observations and associated high-resolution grids. International
Journal of Climatology, 25: 693-712.
Murray-Hudson, M., Wolski, P. and Ringrose, S., 2006. Scenarios of the impact of local and
upstream changes of climate and water use on hydro-ecology in the Okavango
Delta. J. Hydrol., 331: 73-84.
Nakienovi, N., Alcamo, J., Davis, G., de Vries, B., Fenhann, J., Gaffin, S., Gregory, K.,
Grübler, A., Jung, T.Y., Kram, T., La Rovere, E.L., Michaelis, L., Mori, S., Morita, T.,
Pepper, W., Pitcher, H., Price, L., Raihi, K., Roehrl, A., Rogner, H.-H., Sankovski, A.,
Schlesinger, M., Shukla, P., Smith, S., Swart, R., van Rooijen, S., Victor, N. and
Dadi, Z., 2000. Emissions Scenarios. A Special Report of Working Group III of the
Intergovernmental Panel on Climate Change. Cambridge University Press,
Cambridge, UK and New York, NY, USA, 599 pp.
ODMP, 2006. Okavango Delta Management Plan Project Draft Managment Plan, Okavango
Delta Management Plan Project, Maun, Botswana.
Palmer, T.N., Doblas-Reyes, F.J., Hagedorn, R. and Weisheimer, A., 2005. Probabilistic
prediction of climate using multi-model ensembles: from basics to application. Phil.
Trans. R. Soc. B, 360: 1991-1998.
61
TDA Basin Climate Change
Palmer, T.N. and Räisänen, J., 2002. Quantifying the risk of extreme seasonal precipitation
events in a changing
climate. Nature, 415: 512-514.
Parker, D., Folland, C., Scaife, A., Knight, J., Colman, A., Baines, P. and Dong, B., 2007.
Decadal and multidecadal variability and climate change background. Journal of
Geophysical Research, 112: 1-18.
Peterson, T.C., Vose, R., Schmoyer, R. and Razuvaev, V., 1998. Global Historical
Climatology Network (GHCN) quality control of monthly temperature data.
International Journal of Climatology, 18: 1169-1179.
Power, S., Tseitkin, F., Mehta, V., Lavery, B., Torok, S. and Holbrook, N., 1999. Decadal
climate variability in Australia during the twentieth century. International Journal of
Climatology, 19: 169-184.
Prudhomme, C., Jakob, D. and Svensson, C., 2003. Uncertainty and climte change impact
on the flood regime of small UK catchments. Journal of Hydrology, 277(1-23).
Prudhomme, C., Reynard, N. and Crooks, S., 2002. Downscaling of global climate models
for flood frequency analysis. Where are we now? Hydrological Processes, 16: 1137-
1150.
Schulze, R.E., 2005. Climate change and water resources in southern Africa. Studies on
scenarios, impacts, vulnerabilities and adaptation. WRC Report No. 1430/1/05,
Water Resources Commission, Pretoria, South Africa.
Scudder, T., Manley, R.E., Coley, R.W., Davis, R.K., Green, J., Howard, G.W., Lawry, S.W.,
Martz, D., Rogers, P.P., Taylor, A.R.D., Turner, S.D., White, G.F. and Wright, E.P.,
1993. The IUCN review of the Southern Okavango Integrated Water Development
Project. IUCN, Gland, Switzerland.
SMEC, 1990. Southern Okavango Integrated Water Development Project, Department of
Water Affairs, Gaborone, Botswana.
Sorteberg, A. and Kvamstø, N.G., 2006. Tellus. The effect of internal variability on
anthropogenic climate projections, 58A: 565-574.
Tebaldi, C., Smith, R.L., Nychka, D. and Mearns, L.O., 2005. Quantifying Uncertainty in
Projections of Regional Climate Change: A Bayesian
Approach to the Analysis of Multimodel Ensembles. Journal of Climate, 18: 1524-1540.
Todd, M., Andersson, L., Hughes, D., Kniveton, D., Layberry, R., Murray-Hudson, M.,
Savenije, H.G.H., Wilk, J. and Wolski, P., 2008. Simulating climate impacts on water
resources: Experience from the Okavango River, southern Africa. In: S.e. al. (Editor),
Hydrological modelling and the Water Cycle: coupling atmospheric and hydrological
models. Springer, pp. 243-266.
Tyson, P.D., Cooper, G.R.J. and McCarthy, T.S., 2002. Millennial to multi-decadal variability
in the climate of southern Africa. International Journal of Climatology, 22: 1105-1117.
Wilby, R.L., Charles, S.P., Zorita, E., Timbisi, B., Whetton, P. and Mearns, L.O., 2004.
Guidlines for use of climate scenarios developed from statistical downscaling
methods, International Panel on Climate Change.
Wilby, R.L., Wigley, T.M.L., Conway, D., Jones, P.D., Hewitson, B.C., Main, J. and Wilks,
D.S., 1998. Statistical downscaling of general circulation model output: a comparison
of methods. Water Resources Research, 34(2995-3008).
Wolski, P., 2009. Climate change and multidecadal variability of rainfall in the Okavango
Basin, 25th South African Society for Atmospheric Sciences (SASAS) Conference, 9-
10 September 2009, Cape Town, South Africa.
Wolski, P., Gumbricht, T. and McCarthy, T.S., 2002. Assessing future change in the
Okavango Delta: the use of a regression model of the maximum annual flood in a
Monte Carlo simulatioin., Conference on Environmental Monitoring of Tropical and
Subtropical Wetlands, 4-8 December 2002, Maun, Botswana.
Wolski, P. and Murray-Hudson, M., 2008. "Alternative futures" of the Okavango Delta
simulated by a suite of global climate and hydro-ecological models. Water SA, 34(5):
605-610.
62
TDA Basin Climate Change
Wolski, P., Savenije, H., Murray-Hudson, M. and Gumbricht, T., 2006. Modelling the
hydrology of the Okavango Delta, Botswana using a hybrid GIS-reservoir model. J.
Hydrol., 331: 58-72.
Xie, P. and Arkin, P.A., 1996. Analysis of global monthly precipitation using gauge
observations, satellite estimates, and numerical model predictions. Journal of
Climate, 9: 840-858.
63
TDA Basin Climate Change
The Okavango River Basin Transboundary Diagnostic Analysis Technical
Reports
In 1994, the three riparian countries of the Okavango River Basin Angola, Botswana and Namibia agreed to
plan for collaborative management of the natural resources of the Okavango, forming the Permanent Okavango
River Basin Water Commission (OKACOM). In 2003, with funding from the Global Environment Facility,
OKACOM launched the Environmental Protection and Sustainable Management of the Okavango River Basin
(EPSMO) Project to coordinate development and to anticipate and address threats to the river and the
associated communities and environment. Implemented by the United Nations Development Program and
executed by the United Nations Food and Agriculture Organization, the project produced the Transboundary
Diagnostic Analysis to establish a base of available scientific evidence to guide future decision making. The
study, created from inputs from multi-disciplinary teams in each country, with specialists in hydrology, hydraulics,
channel form, water quality, vegetation, aquatic invertebrates, fish, birds, river-dependent terrestrial wildlife,
resource economics and socio-cultural issues, was coordinated and managed by a group of specialists from the
southern African region in 2008 and 2009.
The following specialist technical reports were produced as part of this process and form substantive background
content for the Okavango River Basin Transboundary Diagnostic Analysis.
Final Study
Reports integrating findings from all country and background reports, and covering the entire
Reports
basin.
Aylward, B.
Economic Valuation of Basin Resources: Final Report to
EPSMO Project of the UN Food & Agriculture Organization as
an Input to the Okavango River Basin Transboundary
Diagnostic Analysis
Barnes, J. et al.
Okavango River Basin Transboundary Diagnostic Analysis:
Socio-Economic Assessment Final Report
King, J.M. and Brown,
Okavango River Basin Environmental Flow Assessment Project
C.A.
Initiation Report (Report No: 01/2009)
King, J.M. and Brown,
Okavango River Basin Environmental Flow Assessment EFA
C.A.
Process Report (Report No: 02/2009)
King, J.M. and Brown,
Okavango River Basin Environmental Flow Assessment
C.A.
Guidelines for Data Collection, Analysis and Scenario Creation
(Report No: 03/2009)
Bethune, S. Mazvimavi,
Okavango River Basin Environmental Flow Assessment
D. and Quintino, M.
Delineation Report (Report No: 04/2009)
Beuster, H.
Okavango River Basin Environmental Flow Assessment
Hydrology Report: Data And Models(Report No: 05/2009)
Beuster,
H. Okavango River Basin Environmental Flow Assessment
Scenario Report : Hydrology (Report No: 06/2009)
Jones, M.J.
The Groundwater Hydrology of The Okavango Basin (FAO
Internal Report, April 2010)
King, J.M. and Brown,
Okavango River Basin Environmental Flow Assessment
C.A.
Scenario Report: Ecological and Social Predictions (Volume 1
of 4)(Report No. 07/2009)
King, J.M. and Brown,
Okavango River Basin Environmental Flow Assessment
C.A.
Scenario Report: Ecological and Social Predictions (Volume 2
of 4: Indicator results) (Report No. 07/2009)
King, J.M. and Brown,
Okavango River Basin Environmental Flow Assessment
C.A.
Scenario Report: Ecological and Social Predictions: Climate
Change Scenarios (Volume 3 of 4) (Report No. 07/2009)
King, J., Brown, C.A.,
Okavango River Basin Environmental Flow Assessment
Joubert, A.R. and
Scenario Report: Biophysical Predictions (Volume 4 of 4:
Barnes, J.
Climate Change Indicator Results) (Report No: 07/2009)
King, J., Brown, C.A.
Okavango River Basin Environmental Flow Assessment Project
and Barnes, J.
Final Report (Report No: 08/2009)
Malzbender, D.
Environmental Protection And Sustainable Management Of The
Okavango River Basin (EPSMO): Governance Review
Vanderpost, C. and
Database and GIS design for an expanded Okavango Basin
Dhliwayo, M.
Information System (OBIS)
Veríssimo, Luis
GIS Database for the Environment Protection and Sustainable
Management of the Okavango River Basin Project
Wolski,
P.
Assessment of hydrological effects of climate change in the
Okavango Basin
Country Reports
Angola
Andrade e Sousa,
Análise Diagnóstica Transfronteiriça da Bacia do Rio
Biophysical Series
Helder André de
Okavango: Módulo do Caudal Ambiental: Relatório do
Especialista: País: Angola: Disciplina: Sedimentologia &
Geomorfologia
64
TDA Basin Climate Change
Gomes, Amândio
Análise Diagnóstica Transfronteiriça da Bacia do Rio
Okavango: Módulo do Caudal Ambiental: Relatório do
Especialista: País: Angola: Disciplina: Vegetação
Gomes,
Amândio
Análise Técnica, Biofísica e Socio-Económica do Lado
Angolano da Bacia Hidrográfica do Rio Cubango: Relatório
Final:Vegetação da Parte Angolana da Bacia Hidrográfica Do
Rio Cubango
Livramento, Filomena
Análise Diagnóstica Transfronteiriça da Bacia do Rio
Okavango: Módulo do Caudal Ambiental: Relatório do
Especialista: País: Angola: Disciplina:Macroinvertebrados
Miguel, Gabriel Luís
Análise Técnica, Biofísica E Sócio-Económica do Lado
Angolano da Bacia Hidrográfica do Rio Cubango:
Subsídio Para o Conhecimento Hidrogeológico
Relatório de Hidrogeologia
Morais, Miguel
Análise Diagnóstica Transfronteiriça da Bacia do Análise Rio
Cubango (Okavango): Módulo da Avaliação do Caudal
Ambiental: Relatório do Especialista País: Angola Disciplina:
Ictiofauna
Morais,
Miguel
Análise Técnica, Biófisica e Sócio-Económica do Lado
Angolano da Bacia Hidrográfica do Rio Cubango: Relatório
Final: Peixes e Pesca Fluvial da Bacia do Okavango em Angola
Pereira, Maria João
Qualidade da Água, no Lado Angolano da Bacia Hidrográfica
do Rio Cubango
Santos,
Carmen
Ivelize
Análise Diagnóstica Transfronteiriça da Bacia do Rio
Van-Dúnem S. N.
Okavango: Módulo do Caudal Ambiental: Relatório de
Especialidade: Angola: Vida Selvagem
Santos, Carmen Ivelize
Análise Diagnóstica Transfronteiriça da Bacia do Rio
Van-Dúnem S.N.
Okavango:Módulo Avaliação do Caudal Ambiental: Relatório de
Especialidade: Angola: Aves
Botswana Bonyongo, M.C.
Okavango River Basin Technical Diagnostic Analysis:
Environmental Flow Module: Specialist Report: Country:
Botswana: Discipline: Wildlife
Hancock, P.
Okavango River Basin Technical Diagnostic Analysis:
Environmental Flow Module : Specialist Report: Country:
Botswana: Discipline: Birds
Mosepele,
K. Okavango River Basin Technical Diagnostic Analysis:
Environmental Flow Module: Specialist Report: Country:
Botswana: Discipline: Fish
Mosepele, B. and
Okavango River Basin Technical Diagnostic Analysis:
Dallas, Helen
Environmental Flow Module: Specialist Report: Country:
Botswana: Discipline: Aquatic Macro Invertebrates
Namibia
Collin Christian &
Okavango River Basin: Transboundary Diagnostic Analysis
Associates CC
Project: Environmental Flow Assessment Module:
Geomorphology
Curtis, B.A.
Okavango River Basin Technical Diagnostic Analysis:
Environmental Flow Module: Specialist Report Country:
Namibia Discipline: Vegetation
Bethune, S.
Environmental Protection and Sustainable Management of the
Okavango River Basin (EPSMO): Transboundary Diagnostic
Analysis: Basin Ecosystems Report
Nakanwe, S.N.
Okavango River Basin Technical Diagnostic Analysis:
Environmental Flow Module: Specialist Report: Country:
Namibia: Discipline: Aquatic Macro Invertebrates
Paxton,
M. Okavango River Basin Transboundary Diagnostic Analysis:
Environmental Flow Module: Specialist
Report:Country:Namibia: Discipline: Birds (Avifauna)
Roberts, K.
Okavango River Basin Technical Diagnostic Analysis:
Environmental Flow Module: Specialist Report: Country:
Namibia: Discipline: Wildlife
Waal,
B.V. Okavango River Basin Technical Diagnostic Analysis:
Environmental Flow Module: Specialist Report: Country:
Namibia:Discipline: Fish Life
Country Reports
Angola
Gomes, Joaquim
Análise Técnica dos Aspectos Relacionados com o Potencial
Socioeconomic
Duarte
de Irrigação no Lado Angolano da Bacia Hidrográfica do Rio
Series
Cubango: Relatório Final
Mendelsohn,
.J.
Land use in Kavango: Past, Present and Future
Pereira, Maria João
Análise Diagnóstica Transfronteiriça da Bacia do Rio
Okavango: Módulo do Caudal Ambiental: Relatório do
Especialista: País: Angola: Disciplina: Qualidade da Água
Saraiva, Rute et al.
Diagnóstico Transfronteiriço Bacia do Okavango: Análise
Socioeconómica Angola
Botswana Chimbari, M. and
Okavango River Basin Trans-Boundary Diagnostic Assessment
Magole, Lapologang
(TDA): Botswana Component: Partial Report: Key Public Health
65
TDA Basin Climate Change
Issues in the Okavango Basin, Botswana
Magole,
Lapologang
Transboundary Diagnostic Analysis of the Botswana Portion of
the Okavango River Basin: Land Use Planning
Magole, Lapologang
Transboundary Diagnostic Analysis (TDA) of the Botswana p
Portion of the Okavango River Basin: Stakeholder Involvement
in the ODMP and its Relevance to the TDA Process
Masamba,
W.R.
Transboundary Diagnostic Analysis of the Botswana Portion of
the Okavango River Basin: Output 4: Water Supply and
Sanitation
Masamba,W.R.
Transboundary Diagnostic Analysis of the Botswana Portion of
the Okavango River Basin: Irrigation Development
Mbaiwa.J.E. Transboundary Diagnostic Analysis of the Okavango River
Basin: the Status of Tourism Development in the Okavango
Delta: Botswana
Mbaiwa.J.E. &
Assessing the Impact of Climate Change on Tourism Activities
Mmopelwa, G.
and their Economic Benefits in the Okavango Delta
Mmopelwa,
G.
Okavango River Basin Trans-boundary Diagnostic Assessment:
Botswana Component: Output 5: Socio-Economic Profile
Ngwenya, B.N.
Final Report: A Socio-Economic Profile of River Resources and
HIV and AIDS in the Okavango Basin: Botswana
Vanderpost,
C.
Assessment of Existing Social Services and Projected Growth
in the Context of the Transboundary Diagnostic Analysis of the
Botswana Portion of the Okavango River Basin
Namibia
Barnes, J and
Okavango River Basin Technical Diagnostic Analysis:
Wamunyima, D
Environmental Flow Module: Specialist Report:
Country: Namibia: Discipline: Socio-economics
Collin Christian &
Technical Report on Hydro-electric Power Development in the
Associates CC
Namibian Section of the Okavango River Basin
Liebenberg, J.P.
Technical Report on Irrigation Development in the Namibia
Section of the Okavango River Basin
Ortmann, Cynthia L.
Okavango River Basin Technical Diagnostic Analysis:
Environmental Flow Module : Specialist Report Country:
Namibia: discipline: Water Quality
Nashipili,
Okavango River Basin Technical Diagnostic Analysis: Specialist
Ndinomwaameni
Report: Country: Namibia: Discipline: Water Supply and
Sanitation
Paxton,
C.
Transboundary Diagnostic Analysis: Specialist Report:
Discipline: Water Quality Requirements For Human Health in
the Okavango River Basin: Country: Namibia
66
TDA Basin Climate Change
67
Document Outline
- þÿ
- þÿ
- þÿ
- þÿ
- þÿ
- þÿ
- þÿ
- þÿ
- þÿ