Elsevier

Journal of Hydrology

Volume 289, Issues 1–4, 20 April 2004, Pages 145-164
Journal of Hydrology

Spatial variability of evaporation and moisture storage in the swamps of the upper Nile studied by remote sensing techniques

https://doi.org/10.1016/j.jhydrol.2003.11.038Get rights and content

Abstract

The water balance of the upper Nile swamps, in particular, the Sudd, has been the topic of debate for many years. Surface Energy Balance Algorithm for Land (SEBAL) is a parameterization scheme of surface heat fluxes based on spectral satellite measurements. The SEBAL scheme has been applied to derive the energy balance components from National Oceanic Atmospheric Administration-Advanced Very High Resolution Radiometer (NOAA-AVHRR) images over the extensive swamps in the upper Nile. The study area covers the swamps of the Sudd, Bahr el Ghazal and the Sobat sub-basins. The actual evaporation and soil moisture for an area between 2–12°N and 26–36°E (approximately 1000 km×1000 km) have been investigated. Monthly (actual) evaporation and soil moisture maps for the year 2000 have been generated. The evaporation results were verified against longer-term averaged rainfall and flow data of the three sub-basins. A close resemblance was obtained for the Sudd (1.8% error) and the Sobat Basin (5.7% error), while the balance lacks closure for the Ghazal basin (27%) due to un-gauged or inadequately gauged inflow from the upper catchments. It is concluded that soil moisture availability controls the monthly evaporation rates for all areas that are not saturated with water, and that a significant dry-down in the winter period occurs. It is concluded that the evaporation for the Sudd wetlands is 20% less and the average area occupied by the wetlands is 74% larger than assumed in earlier hydrological studies (e.g. Jonglei canal studies). The derived results can serve as a sound basis to support the widely debated evaporation losses from the Sudd, as well as form an input to regional scale climate models for studying atmospheric circulation patterns over Africa and the Nile Basin.

Introduction

The estimation of spatial variation of evaporation (E) in a catchment is fundamental to many applications in water resources and climate modeling. Evaporation, being the sum of interception, soil evaporation, open water evaporation and transpiration, is a key variable not only in water balance determinations, but also to estimate the moisture, heat and CO2 interactions between land and atmosphere (e.g. Sellers et al., 1996). Several climatic studies have indicated that atmospheric circulation and rainfall are significantly affected by the large-scale variation of soil moisture and evaporation (Savenije, 1996b; Enthekhabi et al., 1999). Three methods can be used to estimate the evaporation at a regional scale: by up-scaling point measurements, by remote sensing techniques (e.g. modeling the energy balance at the land surface) and by hydrological modeling (Savenije, 1997). Each of the three methods has its limitations, and an optimal procedure probably would be a combination of the three approaches. To maximally profit from remote sensing and hydrological modeling, data assimilating is gaining terrain in hydrological studies (Walker et al., 2001; Jhorar et al., 2002; Schuurmans et al., 2003) as well as in climate studies (Dolman et al., 2001). Extensive reviews of remote sensing flux determination methods have been presented by Choudhury (1989), Moran and Jackson (1991) and Kustas and Norman (1996).

Unlike hydrological models, remote sensing techniques compute evaporation directly from the energy balance equation without the need to consider other complex hydrological processes. As a result, the error in the quantification of other hydrological processes is not propagated into evaporation (E), and this is a strong advantage. On the other hand, a major limitation of remote sensing data is that the temporal distribution of satellite-based estimates is poor, and that interpolation techniques are necessary to define evaporation between satellite overpasses. Due to extremely scarce ground hydrometeorological data in the vast study area of the Sudd marshes, utilization of satellite imagery is an attractive study approach. Advanced remote sensing techniques have a high potential to estimate hydrological processes, and are probably far better than the scanty field data available for this large tropical wetland.

The vast wetlands in Southern Sudan are characterized by huge evaporation from the Sudd, Bahr el Ghazal and the Sobat sub-basins. The evaporation from the Sudd alone (Bahr el Jebel swamps) is estimated to be more than 50% of the Nile inflow into the Sudd near Juba, i.e. about 28 Gm3/yr out of the 49 Gm3/yr (Sutcliffe and Parks, 1999). The whole river inflow of the Bahr el Ghazal Basin (12 Gm3/yr) is evaporated before reaching the Nile. About 4 Gm3/yr is claimed to be evaporated from the Machar marches (PJTC, 1960). Therefore, the Sudd became subject of research by planners and engineers to save water and carry more water to the rapidly expanding population living in the downstream areas in North Sudan and Egypt. Their aim is to save water by reducing the evaporation (losses) from the wetlands. Numerous studies and projects were proposed to reduce these losses, of which the (uncompleted) Jonglei canal is the most famous (Jonglei investigation team, 1954; PJTC, 1960).

Most of the past studies to estimate evaporation from the Sudd wetlands rely on the computation of evaporation using meteorological ground station data under the basic assumption that the area is wet throughout the year and moisture is not limiting evaporation rates (Sutcliffe and Parks, 1999; Jonglei investigation team, 1954; Penman, 1963). Although some experiments were made to estimate evaporation from papyrus grown in water tanks (Butcher, 1938), their results were rejected in the subsequent hydrological studies as being too low (1533 mm/yr). Instead, more recent studies used open water evaporation of 2150 mm/yr estimated with the Penman formula (Chan et al., 1980; Sutcliffe and Parks, 1999). Penman (1963) assumed that transpiration from papyrus is similar to the evaporation from open lagoons. However, the computed evaporation by his well-known formula represents the potential open water evaporation, and not the actual evaporation (Ea) from the land surface with a certain heterogeneity in vegetation types and development. The major difficulty of this estimation procedure is that evaporation from point measurements is applied to the total Sudd area. One of the complexities in determining the regional scale evaporation is the variability of the boundaries of the Sudd swamps and the soil moisture behavior throughout the year.

Some of the earlier studies intended to define the extent of the Sudd wetlands from satellite imagery. Travaglia et al. (1995) applied a methodology based on the thermal inertia difference between dry and wetlands from the National Oceanic Atmospheric Administration-Advanced Very High Resolution Radiometer (NOAA-AVHRR). NOAA-AVHRR thermal data were complemented with maps of the Normalized Difference Vegetation Index (NDVI). They defined the seasonal and inter-annual variation of the Sudd area between 28.0 and 48.0 Gm2 during 1991–1993. Mason et al. (1992) used the thermal channel of Meteosat to estimate the Sudd area during 1985–1990 between 8.0 Gm2 at its minimum and 40.0 Gm2 at its maximum. These methods showed a successful application of NOAA-AVHRR and Meteosat thermal data to define the area of wetlands, but such methods fail in the rainy season (cloudy conditions) and do not determine the evaporation.

The approach adopted in this study is based on the derivation of the actual evaporation using the SEBAL techniques (Bastiaanssen et al., 1998a). All components of evaporation (evaporation from open water, from vegetative surface and from bare soil) are calculated based on the surface reflectance and emittance in different parts of the spectrum. This forms basic data that can be used in hydrological and meteorological studies. SEBAL was selected because it does not require any input data, except routine weather station data.

Moisture evaporated into the atmosphere may again precipitate at a different location, but in the same basin (Savenije, 1995, Savenije, 1996a; Trenberth, 1999). If this holds true, the need for reclamation efforts loose ground. Estimation of evaporation by remote sensing over the Sudd is a first step in a larger research project to model moisture recycling in the Nile basin. The derived evaporation results from remote sensing will be used subsequently to verify moisture exchanges between land surface and atmosphere in a regional climate model.

Section snippets

The study area: Sudd, Bahr el Ghazal and Sobat basins

NOAA-AVHRR images have been selected to cover three sub-basins: the Sudd, Bahr el Ghazal and the Sobat basins. These basins stretch from the outfalls on the White Nile to the upstream side near the discharge measuring stations. The names of the most important places are provided in Fig. 1. Fig. 1 shows the natural drainage network within the image boundary, superimposed over the digital elevation model.

The largest wetlands of the Sudd are located at the center of the image, extending from

Major SEBAL model principles

The Surface Energy Balance Algorithm for Land (SEBAL) is an energy partitioning algorithm over the land surface, which has been developed to estimate actual evaporation from satellite images (Bastiaanssen et al., 1998a, Bastiaanssen, 2000). The scheme has found applications in different basins of the world, e.g. Snake River basin in Idaho, USA (Allen et al., 2002), the lake Naivasha drainage basin in Kenya (Farah, 2001), all river basins in Sri Lanka (Bastiaanssen and Chandrapala, 2003) and the

SEBAL results

Eq. (1) has been applied to compute the energy balance components and hence to derive the latent heat flux on a pixel-by-pixel basis over an area of 1000 km×1000 km, covering the Sudd and the neighboring swamps of the Ghazal and the Sobat basins. SEBAL has been applied for 37 images acquired during individual days of the year 2000, Table 1 for the date of acquisition. The daily integrated Ea values have been obtained through the evaporative fraction as defined in Eq. (2) and applied in Eq. (4).

Monthly water balances

A monthly water balance has been computed for the three sub-basins to verify the plausibility of the new evaporation results: Sudd, Ghazal and Sobat. These sub-basins have different land use patterns, precipitation rates and accuracy of available hydrological data. Table 2 shows that the Sudd basin has ponding water at the surface throughout most of the area, and that apart from some higher located spots, the entire sub-basin is flooded in the wet season, or has very shallow groundwater table.

Conclusions

Monthly actual evaporation and soil moisture maps during the year 2000 have been computed for an area of 1000 km×1000 km of the Nile basin covering the wetlands of the Sudd, the Bahr Ghazal, and the Sobat. The SEBAL algorithm has been used to derive the energy balance components at the land surface from the NOAA-AVHRR images because only routine weather data need to be known. The evaporation estimates were verified against water balance results, and close resemblance was derived for two of the

References (45)

  • J.P. Walker et al.

    One-dimensional soil moisture profile retrieval by assimilation of near-surface observations: a comparison of retrieval algorithms

    Adv. Water Resour.

    (2001)
  • M.D. Ahmed et al.

    Retrieving soil moisture storage in the unsaturated zone from satellite imagery and bi-annual phreatic surface fluctuations

    Irrig. Drain. Syst.

    (2003)
  • R.G. Allen et al.

    Crop evapotranspiration: guidelines for computing crop water requirements

    (1998)
  • R.G. Allen et al.

    Evapotranspiration on the watershed scale using the SEBAL model and Landsat images

    (2001)
  • R.G. Allen et al.

    Evapotranspiration from a satellite-BASED surface energy balance for the Snake Plain Aquifer in Idaho

    (2002)
  • W.G.M. Bastiaanssen et al.

    Satellite surveillance of evaporative depletion across the Indus Basin

    Water Resour. Res.

    (2002)
  • G. Boni et al.

    Land data assimilation with satellite measurements for the estimation of surface energy balance components and surface control on evaporation

    Water Resour. Res.

    (2001)
  • W.H. Brutsaert et al.

    Regional surface fluxes from satellite-derived surface temperatures (AVHRR) and radiosonde profiles

    Bound.-Layer Meteorol.

    (1992)
  • A.D. Butcher

    The Sadd Hydraulics

    (1938)
  • S.-O. Chan et al.

    Water balance studies of the Bahr el Ghazal swamp

    (1980)
  • B.J. Choudhury

    Estimating evaporation and carbon assimilation using infrared temperature data: vistas in modeling

  • P. Denny

    Permanent swamp vegetation of the Upper Nile

    Hydrobiologia

    (1984)
  • Cited by (105)

    View all citing articles on Scopus
    View full text