JPCE 970 No. of Pages 10, DTD = 5.0.1
ARTICLE IN PRESS
17 September 2004 Disk Used
Physics and Chemistry of the Earth xxx (2004) xxx–xxx
www.elsevier.com/locate/pce
2 Evaluation of the impact of climate change on hydrology and
3 water resources in Swaziland: Part II
OF
a,*
4 Jonathan I. Matondo , Graciana Peter a, Kenneth M. Msibi b
a
5 Department of Geography, Environmental Science and Planning, University of Swaziland, Private Bag 4, Kwaluseni, Swaziland
b
6 Water Resources Branch, Swaziland
RO
8 Abstract
9
10
11
12
13
14
15
16
DP
The enhanced greenhouse gas effect is expected to cause high temperature increase globally (1–3.5 °C) and this will lead to an
increase in precipitation in some regions while other regions will experience reduced precipitation (±20%). The impact of expected
climate change will affect almost all the sectors of the human endeavor. However, the major purpose of this paper is to evaluate the
impact of climate change on hydrology and water resources for Swaziland. The impact of climate change on hydrology and water
resources has been evaluated using General Circulation Model results (rainfall, potential evapotranspiration, air temperature etc.) as
inputs to a rainfall runoff model. The evaluation of the effect of climate change on hydrology and water resources in Swaziland has
been carried out in three catchments namely: Mbuluzi, Komati and Ngwavuma.
The missing rainfall data in the original raw data files was estimated using regression analysis from the long-term daily averages
CTE
17 rearranged in columns. Pairs of stations were analyzed to establish the ones that correlate the best in terms of the r2. Based on the
18 relationship established in each pair, the linear formulae were applied on the larger original data files to estimate and fill the missing
19 data gaps. The gaps in stream flow series were filled in using rainfall-runoff modeling technique.
20 MAGICC-model has been used to simulate the climate parameters for Swaziland given the baseline conditions. Eleven GCMs
21 were used and three of them were found to simulate very well the observed precipitation for Swaziland. These GCMs are: the Geo-
22 physical Fluid Dynamics Laboratory (GFDL), the United Kingdom Transient Resalient (UKTR), and the Canadian Climate
23 Change Equilibrium (CCC-EQ). The three GCMs were used to project the temperature and precipitation changes for Swaziland
24 for year 2075. This information was used to generate the temperature, precipitation and potential evapotranspiration values for
RE
25 the three catchments for year 2075. This information was used as input data to a calibrated WatBall rainfall runoff model. Simu-
26 lation results show that there will be an annual runoff change of ±5% in the Komati catchment and ±2% in the Mbuluzi catchment
27 given climate change conditions. Simulation results show a negative annual runoff change ranging from 4% to 23% in the Ngwa-
28 vuma catchment under climate change scenarios.
29 Ó 2004 Published by Elsevier Ltd.
30
OR
31 1. Introduction considered rare would occur more frequently in certain 36
regions while drought related and competing water use 37
32 The green house gases effect is expected to cause glo- issues will intensify in other regions (Miller, 1989, Shaa- 38
33 bal warming up which in turn will cause changes in aver- kee, 1989; IPCC, 1990). Therefore, there is a need to 39
UNC
34 age precipitation in the order of ±20% (WMO/ICSU/ evaluate the impact of expected climate change on 40
35 UNEP, 1989). Generally it is expected that floods now hydrology and water resources at the local level. The 41
assessment of the impact of the expected climate change 42
on hydrology and water resources involves the use of 43
*
Corresponding author. Tel.: +268 604 3070 (Cell); fax: +268 518 GCM models coupled with hydrologic models (Kunz, 44
5276. 1993). 45
E-mail address: (J.I. Matondo).
1474-7065/$ - see front matter Ó 2004 Published by Elsevier Ltd.
doi:10.1016/j.pce.2004.09.035
, JPCE 970 No. of Pages 10, DTD = 5.0.1
ARTICLE IN PRESS
17 September 2004 Disk Used
2 J.I. Matondo et al. / Physics and Chemistry of the Earth xxx (2004) xxx–xxx
160
140
Precipitation (mm)
120
100
80
60
40
20
0
Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug
OF
Months
Climate UKTR GFDL CCC-EQ
Fig. 1. Comparison of model performance with 1961–1990 average monthly precipitation.
PRO
46 2. Methodology very well the observed precipitation values. Therefore, 85
they have been used to predict the temperatures and pre- 86
47 The expected climatic changes due to anthropogenic cipitation for Swaziland for year 2075 for low, medium 87
48 activities will cause global warming. The effects of global and high climate change scenarios. 88
49 warming will bring changes in annual average precipita- Table 1 shows the projected temperature increase 89
50 tion values in the order of ±20% (IPCC, 1990). Extreme (°C) and precipitation increase or decrease (%) to year 90
51 events (droughts and floods) now considered rare will 2075 by the various models. It can be seen from Table 91
52 occur more frequently in certain regions. General circu- 1 that, all models are predicting a temperature increase 92
53 lation models (GCMs) provide physically based predic- and a precipitation decrease. This information is used 93
54 tions of the way climate might change as a result of to derive the potential evapotranspiration and rainfall 94
55 increasing concentrations of atmospheric carbon diox-
TED values at year 2075 using regression analysis technique. 95
56 ide and other trace gases. The GCMs are mathemati- This information is then used as input data to a rainfall 96
57 cally representatives of the earthÕs climate system and runoff model. Table 5 shows an example of the proc- 97
58 they simulate atmospheric processes at a field of grid essed information (27 such tables for each catchment). 98
59 points that cover the surface of the earth (IPCC, 1996).
60 The climate scenario generators (CSG) that has been
61 used in this study is a combination of a simple climate 3. Application of WatBall model to the Komati, Mbuluzi 99
REC
62 model ‘‘MAGICC’’ and a climate scenario database and Ngwavuma catchments 100
63 SCENGEN. MAGICC—Model for the Assessment of
64 Greenhouse-gas Induced Climate Change—is a set of The WatBall model has been applied to three river 101
65 linked reduced form simple models that emulate the basins of Swaziland for the evaluation of the effect of cli- 102
66 behaviour of fully three-dimensional dynamic GCMs. mate change on the water resources. There are two 103
67 SCENGEN—Scenario Generator on the other hand is stages in the use of this model, that is calibration and 104
68 a global and regional database containing results of a application. During the calibration stage the model 105
69 large number of GCM experiments as well as the ob- parameters are adjusted by trial and error process till 106
OR
70 served global and four regionsÕ climate data sets. the model closely reproduces the observed stream flow. 107
71 The first step in the application of GCMs is the selec- Tables 2–4 shows the optimal model parameters and 108
72 tion of the GCM models suitable for Swaziland. MA- the correlation coefficient between observed and simu- 109
73 GICC-model was run for the baseline conditions with lated stream flow for the Komati, Mbuluzi and Ngwa- 110
74 observed meteorological information from 1960 to vuma river basins during calibration for wet, dry and 111
75 1994. The criteria for GCM model selection are: model average year conditions. 112
UNC
76 vintage, model resolution and model representativeness
77 as discussed in phase I report. However, model repre-
78 sentativeness is given more weight and that is the model 4. Results and discussion 113
79 should be able to fairly reproduce the observed informa-
80 tion (temperature or precipitation). The response of the catchments in Swaziland (Koma- 114
81 Fig. 1 shows a comparison of model performance ti, Mbuluzi and Ngwavuma) due to climate change has 115
82 with 1961 to 1990 average mean precipitation of Swazi- been evaluated using GCM models which are; the Geo- 116
83 land. What can be seen from Fig. 1 is that all the three physical Fluid Dynamics Laboratory (GFDL), the Uni- 117
84 models (UKTR, GFDL, and CCC-EQ) are simulating ted Kingdom Transient Resalient (UKTR), and the 118
ARTICLE IN PRESS
17 September 2004 Disk Used
Physics and Chemistry of the Earth xxx (2004) xxx–xxx
www.elsevier.com/locate/pce
2 Evaluation of the impact of climate change on hydrology and
3 water resources in Swaziland: Part II
OF
a,*
4 Jonathan I. Matondo , Graciana Peter a, Kenneth M. Msibi b
a
5 Department of Geography, Environmental Science and Planning, University of Swaziland, Private Bag 4, Kwaluseni, Swaziland
b
6 Water Resources Branch, Swaziland
RO
8 Abstract
9
10
11
12
13
14
15
16
DP
The enhanced greenhouse gas effect is expected to cause high temperature increase globally (1–3.5 °C) and this will lead to an
increase in precipitation in some regions while other regions will experience reduced precipitation (±20%). The impact of expected
climate change will affect almost all the sectors of the human endeavor. However, the major purpose of this paper is to evaluate the
impact of climate change on hydrology and water resources for Swaziland. The impact of climate change on hydrology and water
resources has been evaluated using General Circulation Model results (rainfall, potential evapotranspiration, air temperature etc.) as
inputs to a rainfall runoff model. The evaluation of the effect of climate change on hydrology and water resources in Swaziland has
been carried out in three catchments namely: Mbuluzi, Komati and Ngwavuma.
The missing rainfall data in the original raw data files was estimated using regression analysis from the long-term daily averages
CTE
17 rearranged in columns. Pairs of stations were analyzed to establish the ones that correlate the best in terms of the r2. Based on the
18 relationship established in each pair, the linear formulae were applied on the larger original data files to estimate and fill the missing
19 data gaps. The gaps in stream flow series were filled in using rainfall-runoff modeling technique.
20 MAGICC-model has been used to simulate the climate parameters for Swaziland given the baseline conditions. Eleven GCMs
21 were used and three of them were found to simulate very well the observed precipitation for Swaziland. These GCMs are: the Geo-
22 physical Fluid Dynamics Laboratory (GFDL), the United Kingdom Transient Resalient (UKTR), and the Canadian Climate
23 Change Equilibrium (CCC-EQ). The three GCMs were used to project the temperature and precipitation changes for Swaziland
24 for year 2075. This information was used to generate the temperature, precipitation and potential evapotranspiration values for
RE
25 the three catchments for year 2075. This information was used as input data to a calibrated WatBall rainfall runoff model. Simu-
26 lation results show that there will be an annual runoff change of ±5% in the Komati catchment and ±2% in the Mbuluzi catchment
27 given climate change conditions. Simulation results show a negative annual runoff change ranging from 4% to 23% in the Ngwa-
28 vuma catchment under climate change scenarios.
29 Ó 2004 Published by Elsevier Ltd.
30
OR
31 1. Introduction considered rare would occur more frequently in certain 36
regions while drought related and competing water use 37
32 The green house gases effect is expected to cause glo- issues will intensify in other regions (Miller, 1989, Shaa- 38
33 bal warming up which in turn will cause changes in aver- kee, 1989; IPCC, 1990). Therefore, there is a need to 39
UNC
34 age precipitation in the order of ±20% (WMO/ICSU/ evaluate the impact of expected climate change on 40
35 UNEP, 1989). Generally it is expected that floods now hydrology and water resources at the local level. The 41
assessment of the impact of the expected climate change 42
on hydrology and water resources involves the use of 43
*
Corresponding author. Tel.: +268 604 3070 (Cell); fax: +268 518 GCM models coupled with hydrologic models (Kunz, 44
5276. 1993). 45
E-mail address: (J.I. Matondo).
1474-7065/$ - see front matter Ó 2004 Published by Elsevier Ltd.
doi:10.1016/j.pce.2004.09.035
, JPCE 970 No. of Pages 10, DTD = 5.0.1
ARTICLE IN PRESS
17 September 2004 Disk Used
2 J.I. Matondo et al. / Physics and Chemistry of the Earth xxx (2004) xxx–xxx
160
140
Precipitation (mm)
120
100
80
60
40
20
0
Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug
OF
Months
Climate UKTR GFDL CCC-EQ
Fig. 1. Comparison of model performance with 1961–1990 average monthly precipitation.
PRO
46 2. Methodology very well the observed precipitation values. Therefore, 85
they have been used to predict the temperatures and pre- 86
47 The expected climatic changes due to anthropogenic cipitation for Swaziland for year 2075 for low, medium 87
48 activities will cause global warming. The effects of global and high climate change scenarios. 88
49 warming will bring changes in annual average precipita- Table 1 shows the projected temperature increase 89
50 tion values in the order of ±20% (IPCC, 1990). Extreme (°C) and precipitation increase or decrease (%) to year 90
51 events (droughts and floods) now considered rare will 2075 by the various models. It can be seen from Table 91
52 occur more frequently in certain regions. General circu- 1 that, all models are predicting a temperature increase 92
53 lation models (GCMs) provide physically based predic- and a precipitation decrease. This information is used 93
54 tions of the way climate might change as a result of to derive the potential evapotranspiration and rainfall 94
55 increasing concentrations of atmospheric carbon diox-
TED values at year 2075 using regression analysis technique. 95
56 ide and other trace gases. The GCMs are mathemati- This information is then used as input data to a rainfall 96
57 cally representatives of the earthÕs climate system and runoff model. Table 5 shows an example of the proc- 97
58 they simulate atmospheric processes at a field of grid essed information (27 such tables for each catchment). 98
59 points that cover the surface of the earth (IPCC, 1996).
60 The climate scenario generators (CSG) that has been
61 used in this study is a combination of a simple climate 3. Application of WatBall model to the Komati, Mbuluzi 99
REC
62 model ‘‘MAGICC’’ and a climate scenario database and Ngwavuma catchments 100
63 SCENGEN. MAGICC—Model for the Assessment of
64 Greenhouse-gas Induced Climate Change—is a set of The WatBall model has been applied to three river 101
65 linked reduced form simple models that emulate the basins of Swaziland for the evaluation of the effect of cli- 102
66 behaviour of fully three-dimensional dynamic GCMs. mate change on the water resources. There are two 103
67 SCENGEN—Scenario Generator on the other hand is stages in the use of this model, that is calibration and 104
68 a global and regional database containing results of a application. During the calibration stage the model 105
69 large number of GCM experiments as well as the ob- parameters are adjusted by trial and error process till 106
OR
70 served global and four regionsÕ climate data sets. the model closely reproduces the observed stream flow. 107
71 The first step in the application of GCMs is the selec- Tables 2–4 shows the optimal model parameters and 108
72 tion of the GCM models suitable for Swaziland. MA- the correlation coefficient between observed and simu- 109
73 GICC-model was run for the baseline conditions with lated stream flow for the Komati, Mbuluzi and Ngwa- 110
74 observed meteorological information from 1960 to vuma river basins during calibration for wet, dry and 111
75 1994. The criteria for GCM model selection are: model average year conditions. 112
UNC
76 vintage, model resolution and model representativeness
77 as discussed in phase I report. However, model repre-
78 sentativeness is given more weight and that is the model 4. Results and discussion 113
79 should be able to fairly reproduce the observed informa-
80 tion (temperature or precipitation). The response of the catchments in Swaziland (Koma- 114
81 Fig. 1 shows a comparison of model performance ti, Mbuluzi and Ngwavuma) due to climate change has 115
82 with 1961 to 1990 average mean precipitation of Swazi- been evaluated using GCM models which are; the Geo- 116
83 land. What can be seen from Fig. 1 is that all the three physical Fluid Dynamics Laboratory (GFDL), the Uni- 117
84 models (UKTR, GFDL, and CCC-EQ) are simulating ted Kingdom Transient Resalient (UKTR), and the 118
