05Mobin-ud-Din- estimating...-Physics and Chemistry of the Earth

CATCHMENT IN THE LIMPOPO BASIN, SOUTH AFRICA

Mobin-ud-Din Ahmad1, Thulani F. Magagula2, David Love3,4,, Victor Kongo5, Marloes L. Mul6, 7 and

Jeniffer Kinoti2

1

International Water Management Institute (IWMI), PO Box 2075, Colombo, Sri Lanka 2

International Water Management Institute, 41 Creswell St, Weavind Park, 0184, Pretoria, South

Africa

3

WaterNet, PO Box MP600, Mount Pleasant, Harare, Zimbabwe

4

ICRISAT Bulawayo, Matopos Research Station, PO Box 776 Bulawayo, Zimbabwe

5

School of Bioresources Engineering and Environmental Hydrology, University of KwaZulu-Natal,

PB X01, Scottsville, 3209 Pietermaritzburg, South Africa

6

Department of Civil Engineering, University of Zimbabwe, PO Box MP167, Mount Pleasant, Harare,

Zimbabwe

7

UNESCO-IHE, Institute for Water Education, PO Box 3015, 2601 DA Delft, the Netherlands

ABSTRACT

This paper describes a case study that uses a remote sensing technique, the Surface Energy Balance Algorithm for Land (SEBAL) to assess actual evapotranspiration across a range of land uses in the middle part of the Olifants Basin in South Africa.. SEBAL enables the estimation of pixel scale ETa using red, near infrared and thermal bands from satellite sensors supported by ground-based measurements of wind speed, humidity, solar radiation and air temperature.

The Olifants River system, although supplying downstream users in Mpumalanga Province (South Africa) and Chókwè District (Mozambique), is over-committed, principally for irrigation, in the upper reaches. Therefore, quantification of evapotranspiration from irrigated lands is very useful to monitor respect of compliance in water allocations and sharing of benefits among different users.

A Landsat7 ETM+ image, path 170 row 077, was acquired on 7 January 2002, during the rainy season and was used for this analysis. The target area contains diverse land uses, including rainfed agriculture, irrigated agriculture (centre pivot, sprinkler and drip irrigation systems), orchards and rangelands. Commercial farming (rainfed and irrigated agriculture) is one of the main economic activities in the area. SEBAL ETa estimates vary from 0 to 10 mm/day over the image. Lowest ETa was observed for barren/fallow fields and highest for open water bodies. ETa for vegetative areas ranges 3 to 9 mm/day but irrigated areas, using central pivot, drip and sprinkler systems, appear to evaporate with a higher rate: 6 and 9 mm/day. Penman-Monteith reference crop evapotranspiration ET0 on the same day was found to be 7 mm/day. This indicates that these irrigated areas have no water stress and potential yields can be achieved provided there is no nutrient deficiency. The major finding is that SEBAL results showed that 24% of ETa is from agricultural use, compared to 75% from non-agricultural land use classes(predominantly forest) and only 1% from water bodies. Although irrigation accounts for roughly half of diverted streamflow in the basin, it contributes only about 4% of basin-scale daily ETa at the time of assessment.

Keywords: agricultural water management, evapotranspiration, SEBAL, remote sensing

1

INTRODUCTION

Actual Evapotranspiration ETa is one of the most useful indicators to explain whether the water is used as “intended” or not. ETa variations, both in space and time, and from different land use classes (particularly from irrigated lands) are thought to be highly indicative for the adequacy, reliability and equity in water use; the knowledge of these terms is essential for judicious water resources management. Unfortunately, ETa estimation under actual field conditions is still a very challenging task for scientists and water managers. The complexity associated with the estimation of ET has lead to the development of various methods for estimating this parameter over time Doorenbos and Pruitt (1977); Allen et al. (1998).

The methods for estimating ET can generally be grouped into 4 categories i.e. the hydrological methods (water balance), direct measurement (lysimeters), micrometeorological (energy balance) and empirical or combination methods (Thornthwaite), based on energy balance or climatic factors Thomthwaite and Mather (1955). Most of these methods can only provide point estimates of ET which are not sufficient for system-level water management. Distributed physically-based hydrological models can compute ET patterns but require enormous amounts of field data which are often unavailable in many river basins in the world.

During the last two to three decades, significant progress has been made to estimate actual evapotranspiration (ETa) using satellite remote sensing Engman and Gurney (1992), Kustas and Norman (1996), Bastiaanssen et al. (1998, 2002) and Kustas et al. (2003). These methods provide a powerful means to compute ETa from the scale of an individual pixel right up to an entire raster image.

2

This study demonstrates the application of a remote sensing method, the Surface Energy Balance Algorithm for Land (SEBAL) Bastiaanssen et al. (1998) & Bastiaanssen (2000) in a catchment in the middle reach of Olifants basin in South Africa. The Olifants River system, although supplying downstream users in Mpumalanga Province (South Africa) and Chókwè District (Mozambique), is over-committed in the middle reaches by 94 Mm3/year. The commitments are mainly for irrigation, which accounts for 86% of the abstracted water demand in the middle reaches, and 57% of the total water requirements in the South African part of the basin Basson and Rossouw (2003). Therefore, quantification of evapotranspiration from irrigated lands is very useful in cross-checking actual water use against water allocation and in understanding its implications for the specification and management of water rights in a basin.

MATERIALS AND METHOD Description of the study area Location

The Olifants River passes through three provinces of South Africa (Gauteng, Mpumalanga and Limpopo Province), through the Kruger National Park, into Mozambique, where it joins the Limpopo. It is a major tributary to the Limpopo River, located in the north east of South Africa (see figure 1). Its catchment area spans over ,672Km2. The topography of the basin varies widely with altitudes ranging between 2,300m at highest point in the upper part of the catchment and 300m at the Mozambique border.

3

N

Figure 1: Location of the Olifants Basin (Source: DWAF, 2002)

Rainfall and Runoff

On average, the Olifants catchment receives an annual precipitation of 631 mm, which varies spatially over the basin (see figure2). The mean annual runoff is estimated at 2,040 million m3 and the demand for human purposes is estimated at 965 million m3 including hydropower. 460 million m3 is estimated to be the annual reserve requirement. To honour international commitments, about 1,137 million m3 annually still flows to Mozambique. This means that of the annual runoff of 2040 million m3 the basin has to meet its demands from the 903 million m3 left after meeting international commitments. It is in this light that the National Water Resources Strategy NWRS (2004) recognizes that judicious assessment of the Reserve together with careful implementation planning to minimise possible social disruption will be required. The South African Water Act requires that a portion of the available water resources

OfansRive4

be reserved for ecological purposes; this is what is termed the reserve. Estimated future demand for human purposes by 2010 is projected at 1,356.5 million m3, without the reserve and international commitments DWAF (2002). The river has however been known to have zero flow during short periods as it enters Kruger Park, making it a water scarce catchment.

Figure 2: Spatial variation of Rainfall over the Olifants basin [Source: Schulze, (1997)]

Agriculture

There are about 1.2 million hectares of cultivated land in the Olifants catchment. Three distinct forms of farming exist in South Africa: commercial irrigated, commercial dryland and subsistence/semi-commercial farming. About 44% of the cultivated area in the catchment is used for the growing of maize, which is South Africa’s staple crop. The area and estimated production of maize are shown in table 1 below.

5

Table 1: Area, production and yields of maize in the Olifants catchment Farming TypeCommercial IrrigatedCommercial DrylandSubsistence/ Semi-Comm.TotalArea(ha)Production (tons)Average Yield(tons/ha)4.52.30.6 16,712 75,912 383,090 8,472 117,953 73,740 517,755 1,039,124

The total estimated value of production from all crops grown in the catchment based on 2002 market prices prepared by the Department of Agriculture Statistical Department is about R5 billion, van-Heerden and Magagula (2003). It is estimated that about 96% of the total value of production is from commercial farming, split 59% and 37% between dryland/rain-fed and irrigated farming respectively van-Heerden and Magagula (2003).

Irrigation water requirement is estimated 557 million m3 according to the National Water Resources Strategy, 2004. This makes irrigation by far the largest user accounting for about 58% of the total demand for human purposes.

The cropping calendar (figure 5) follows the hydrological year, which begins with summer rain around October and ends in September the following year. A dry season starts around March/April.

6

GROUNDNUTS: 150 DaysSWEET POTATOE: 120 DaysTEMPORAL PASTURE / FULLOW: 150 DaysPOTATOE: 115 DaysCITRUSTOMATOES: 139 Days CABBAGE: 115 Days DRY BEANS: 90 Days WHEAT: 140 Days MAIZE (Late crop): 136 Days MAIZE: 136 Dyas AugSepOctNovSepOctNovDecJanFebMarAprMayJunJulAugDecWET SEASONDRY SEASON

Hydrologic Year

Figure 5: Cropping calendar for some of the crops grown in the Olifants basin [Source: van-Heerden and Magagula, (2003)]

Data collection Satellite imagery

A LANDSAT ETM+ image (Path 170 Row 077), covering nearly the entire middle Olifants, was acquired on 7th January 2002 and was downloaded from Global Land Cover Facility of the University of Maryland website (http://glcf.umiacs.umd.edu/data/landsat/).

Weather data

Meteorological data for a representative station was obtained from the Weather Service Department. Hourly and daily data were used in SEBAL processing. The weather conditions prevailing on 7th January 2002 are shown in table 2.

7

Table 2: Weather conditions at the time of satellite overpass on the 7 January 2002

Satellite Overpass time GMT Local TimeDate Jan 7, 2002hr9hr10Average* / Daily**Temp (K)o7.849.84Temp (C)24.426.425.4*298.55Humidity (%)513744*Wind speed 10m above the ground (m/s)2.12.42.25*12.7**0**Actual Daily Sunshine (hours) Daily Precipitation (mm)

Methodology: Surface Energy Balance Algorithm (SEBAL)

SEBAL computes a complete radiation and energy balance along with the resistances for momentum, heat and water vapour transport for each pixel Bastiaanssen et al. (1998) & Bastiaanssen (2000). The key input data for SEBAL consists of spectral radiance in the visible, near-infrared and thermal infrared part of the spectrum. In addition to satellite images, the SEBAL model requires the following routine weather data parameters (wind speed, humidity, solar radiation, air temperature).

Evaporation is calculated from the instantaneous evaporative fraction , and the daily averaged net radiation, Rn24. The evaporative fraction  is computed from the instantaneous surface energy balance at satellite overpass on a pixel-by-pixel basis:

ERnG0H

(1)

Where: E is the latent heat flux, Rn is the net radiation, G0 is the soil heat flux and H is the sensible heat flux (see Figure 6).

The latent heat flux describes the amount of energy consumed to maintain a certain crop evaporation rate. The surface albedo, surface temperature and vegetation index are derived

8

from satellite measurements, and are used together to solve Rn, G0 and H. The instantaneous latent heat flux, E, is the calculated residual term of the energy budget, and it is then used to compute the instantaneous evaporative fraction :

The instantaneous evaporative fraction  expresses the ratio of the actual to the crop evaporative demand when the atmospheric moisture conditions are in equilibrium with the soil moisture conditions. The instantaneous value can be used to calculate the daily value because evaporative fraction tends to be constant during daytime hours, although the H and

ΛEEHERn-G0 (2)

E fluxes vary considerably. The difference between the instantaneous evaporative fraction at satellite overpass and the evaporative fraction derived from the 24-hour integrated energy balance is marginal and may be neglected Brutsaert and Sugita (1992), Crago (1996), Farah (2001 and 2004)). For time scales of 1 day or longer, G0 can be ignored and net available energy (Rn - G0) reduces to net radiation (Rn). At daily timescales, ET24 (mm/day) can be computed as:

Where: Rn24 (W/m2) is the 24-h averaged net radiation,  (J/kg) is the latent heat of vaporization, and w (kg/m3) is the density of water.

ET24800103ΛRn24w (3)

9

G0GG0RnG0HEERn-G0EEHRnRnRnLEEHERn-G0HEHE Figure 6: Various components of Energy Balance & main equations to compute latent heat

flux

RESULTS AND DISCUSSION

Actual evapotranspiration (ETa) in mm/day for the 7 January 2002 was computed by solving the surface energy balance using equation 1, 2 and 3. The spatial variation of ETa is shown in figure 7. It ranges from 0 mm/day for bare soil and fallow land to 8 mm/day or more for water bodies. Non-agricultural land classes, particularly forest and woodlands (including degraded forest and woodlands) make up about 56% of the study area and have average ETa values of 4.27 mm/day and 2.74 mm/day respectively and an average of 3.51 mm/day for the entire land cover class.

10

Figure 7: Actual Evapotranspiration (ETa) estimates using SEBAL for Landsat7 ETM+ imagery for part of the middle Olifants water management area. 7 January 2002.

Actual Evapotranspiration and Land Cover

Statistics have been extracted from the ETa map using an overlay of land cover/use map Thomson (1999) and are shown in table 3, as mean ETa for each land cover/use. Water bodies have an average ETa of 7.92 mm/day, inclusive of large and small water bodies that can consist of multiple mixed pixels falling both on land and inside water bodies as well as averaging differences in the water surface temperature due to turbidity.

11

Table 3: Mean ETa of different land cover types and the percentage ETa from each land cover type. Land CoverCultivated Commercial Dryland-Temporal-PermanentCultivated Commecial Irrigated-Temporal-PermanentCultivated Semi-commercial / subsistence drylandForest and WoodlandsForest PlantationsGrasslandsThicket & bushlandsMines and QuarriesUrban / built-up landDongas & sheet erosion scarsWaterbodies560692773276847321333812737390052197146877843.67 2,057,0923. 179,1863.04 2,225,5933.51 29,729,6853.82 34,8223.48 2,238,4213.49 4,439,4282.49 22,4362.88 1,503,7044.57 67,0887.92 616,1494.050.3.3858.470.074.408.730.042.960.131.21 27502627812.78 7,5,8413.19 88,62215.040.17Area (ha)Mean ETa (mm/day)ETa Volume (m)3% ETa

Forest and woodlands account for about 58% of the ETa in this particular day. About 24% is beneficial/agricultural field ETa, but inferences based on these statistics are not accurate unless the contributions of each land use classes to livelihoods and productive use such as livestock feeding are known. January is usually a wet month and it is a month of lots of activity across all farming types as farmers are planting or have planted summer crops. It also means that forest ETa will be higher than at other times of the year, due to minimum water stress. It is observed in the ETa map that a greater part of the commercial temporal dryland farming area has low ETa, with values of 2mm/day or less. This could be an indication that most of the land has just been prepared. Under dryland or rainfed conditions, planting depends on rainfall events that provide sufficient moisture for land preparation and planting. It is for that reason that most of the cultivated land would still be fallow, or just been prepared hence the close to zero values of ETa at this time of the year (January).

12

We now focus on the quaternary catchment B31J, shown in figure 8, where four main types characterize land cover/use: natural vegetation (forest and woodlands), cultivated land, water bodies and a bit of built up area. Water bodies had the highest ETa (see figure 9). Forest and woodland, which dominate the upper part of the catchment have higher average ETa than commercial dryland cropping. The upper part of B31J is an endoreic area Van Vuuren et al (2003), described as the portion of a hydrological catchment that does not contribute towards local river flow nor to river flow in downstream catchments. In such catchments, the water generally drains to pans where much of the water is lost through evaporation. In other places, concentrated surface run-off recharges groundwater. The WARMS database of registered water users reveal that about 96% of irrigators in quaternary catchment B31J use boreholes for irrigation and centre pivot is the predominant irrigation system.

Figure 8: Actual Evapotranspiration ETa, a focus on the quaternary catchment B31J.

13

B31J7.006.005.004.003.002.001.000.00Max of ETa_MEANQUATERNARYB31J Degraded: forest andwoodlandForestUrban / built-up land:industrial / transportForest and WoodlandUnimproved grasslandUrban / built-up land:residentialCultivated: temporary -semi-commercial/subsistenceImproved grasslandWaterbodiesCultivated: temporary -commercial drylandCultivated: permanent -commercial irrigatedCultivated: temporary -commercial irrigatedForest plantationsMines & quarriesLAND COVERThicket & bushland (etc)(blank)

Figure 9: Max average ETa for the different land cover types in the quaternary catchment B31J in the middle Olifants.

The interpretation of ETa values depends on the knowledge of actual vegetation cover if accurate determinations of water use by vegetation are to be made. The wide spread use of centre pivot was observed in a field trip to the middle reaches of the Olifants, evident in the image as circular areas with high ETa (see figure 8).

14

CONCLUSION

The paper demonstrates one the first applications of the remote sensing method, SEBAL, to determine spatial variation of actual evapotranspiration for the Olifants river basin in South Africa. Data for SEBAL processing can be sourced from Landsat, NOAA AVHRR, MODIS and ASTER at different scales but requires routine meteorological measurement of air temperature, humidity, wind speed and sunshine duration.

In this study, 30 meter spatial resolution, Landsat7 ETM+ image of Jan. 7, 2002 was used to delineate the spatial variation in ETa. The snapshot computed in this study demonstrates that water bodies have highest ETa, forest and woodlands transpire at a higher rate than cultivated land on Jan. 7, 2002. Volumetrically, forest and woodlands account for about 58% of the ETa in this particular day, the highest of all land cover types. Agricultural field ETa is only 24% of the overall ETa from the investigated area. However, in addition to ETa,, knowledge of the contribution of the each land use to livelihoods and productive use is essential 1) to compute beneficial vs. non-beneficial uses of water and 2) to devise strategies to improve water management/productivity. We can see that, although irrigation requires over 50% of the diverted streamflow and groundwater in the basin, it accounts for a much more modest portion of basin evapotranspiration. In this study, we do not know the beneficial values of forest in terms of timber produced and in terms of hydrological services in maintaining base-flows and catchment yield. Therefore, it is not possible to make further comparisons, nor assess the water productivity. Clearly, a snapshot indicates an overall annual trend in spatial ETa in the basin, due to the relative magnitudes of the areas of each type of land use. However, some form of seasonal and annual integration is also desirable to account for, among other things, reduced forest ETa in the dry season and conversely relative increase in irrigated ETa.

15

Temporal integration is currently only feasible using MODIS or AVHRR data at 1km2 resolution, which then loses the ability to define ETa precisely by land use class.

16

ACKNOWLEDGEMENTS

This paper contains research results from a workshop funded by the International Water Management Institute (IWMI). The cooperation of the Departments of Water Affairs and Forestry and Weather Service has been essential and is gratefully acknowledged. Authors are also thankful to Dominique Rollin (IWMI-South Africa office), Steve Twomlow (Global Theme Leader-Agro-Ecosystems Development at ICRISAT) and Hugh Turral (Theme Leader – Basin Water Management at IWMI) for useful discussions.

The opinions and results presented in this paper are those of the authors and do not necessarily represent the donors or participating institutions

17

REFERENCES

Allen, R. G., L. S. Pereira, D. Raes, and M. Smith, 1998. Crop evapotranspiration, guidelines

for computing crop water requirements, FAO Irrig. and Drain. Pap. 56, 300 pp., Food and Agric. Organ. of the U. N. (FAO), Rome, Italy.

Basson, M.S. and Rossouw, J.D. (2003). Olifants Water Management Area: overview of

water resources availability and utilization. Report P WMA 04/000/00/0203, Department of Water Affairs and Forestry, Pretoria, South Africa.

Bastiaanssen, W.G.M. 2000. SEBAL-based sensible and latent heat fluxes in the irrigated

Gediz Basin, Turkey. Journal of Hydrology 229:87-100

Bastiaanssen, W. G. M., Ahmad, M. D., and Chemin, Y., 2002. Satellite surveillance of

evaporative depletion across the Indus. Water Resources Research 38(12): 1273, 1-9. Bastiaanssen, W.G.M., M. Menenti, R.A. Feddes and A.A.M. Holtslag, 1998. A remote

sensing surface energy balance algorithm for land (SEBAL), part 1: formulation, Journal of Hydrology. 212-213: 198-212. Brutsaert, W. and M. Sugita, 1992. Application of self-preservation in the diurnal evolution of

the surface energy budget to determine daily evaporation. Journal of Geophysical Research. 97, D17: 18,322-18,377. Crago, R.D. 1996. Conservation and variability of the evaporative fraction during the day

time. Journal of Hydrology. 180: 173-194.

Doorenbos, J. and W.O. Pruitt, 1977, Crop water requirements. Irrigation and Drainage Paper

no. 24 (revised), FAO, Italy, 144pp. DWAF, 2002. Proposal for the establishment of a Catchment Management Agency for the

Olifants Water Management Area. Department of Water Affairs and Forestry, Pretoria Engman, E.T., and R.J. Gurney, 1991. Remote sensing in hydrology; Chapman and Hall,

London.

Farah, H.O., 2001. Estimation of regional evaporation using a detailed agro-hydrological

model. Journal of Hydrology. 229(1-2): 50-58. Farah, H.O., W.G.M. Bastiaanssen and R.A. Feddes, 2004. Evaluation of the temporal

variability of the evaporative fraction in a tropical watershed, International Journal of Applied Earth Observation and Geoinformation 5: 129-140, Gieske, A. and W. Meijninger 2003. High density NOAA time series of ET in the Gediz

Basin, Turkey, . ICID Workshop on Remote Sensing of ET for Large Regions, Montpellier, France 17 Sept. 2003. Kustas WP, GR Diak and MS Moran 2003. Evapotranspiration, Remote Sensing of.

Encyclopedia of Water Science pp 267 – 274. Marcel Dekker, Inc., New York.

Kustas, W.P. and J.M. Norman, 1996. Use of remote sensing for evapotranspiration

monitoring over land surfaces; Hydr. Sci. J. 41(4): 495-516.

NWRS 2004. National Water Resources Strategy (First Edition), Department of Water Affairs

and Forestry, Pretoria, South Africa.

Schulze, R. E. 1997. South African atlas of agrohydrology and climatology. Pretoria, South

Africa:

18

Thomson, M.W., 1999. South African National Land Cover Database Project: Data Users

Manual, Final Report (Phase 1,2&3). ENV/P/V 98136 Version 3.1, Council for

Scientific & Industrial Research (CSIR), Division of Water, Environment and Forest Science, Pretoria, South Africa Thomthwaite, C.W.; Mather, J.R. 1955. The water balance. Publications in climatology.

Centerton, NJ: Drexel Institute of Technology. Vol. VIII, No. 1. van Vuuren, A.J., H. Jordaan, E. Van Der Walt and S. Van Jaarsveld, 2003. Olifants Water

Management Area: Water Resources Situation Assessment, DWAF Report

P/04000/00/0101, Department of Water Affairs and Forestry, Pretoria, South Africa. van-Heerden P. and Magagula T.F., 2003. Crop Production Systems of the Olifants River

Basin (Drainage Region B) in South Africa. Unpublished report, International Water Management Institute, Pretoria, South Africa..

19