New lessons on the hydrology learned from remote sensing and climate modeling Y. A. Mohamed, H. H. G. Savenije, W. G. M. Bastiaanssen, B. J .J. M. van den Hurk

To cite this version:

Y. A. Mohamed, H. H. G. Savenije, W. G. M. Bastiaanssen, B. J .J. M. van den Hurk. New lessons on the Sudd hydrology learned from remote sensing and climate modeling. Hydrology and Earth System Sciences Discussions, European Geosciences Union, 2006, 10 (4), pp.507-518. ￿hal-00305006￿

HAL Id: hal-00305006 https://hal.archives-ouvertes.fr/hal-00305006 Submitted on 10 Jul 2006

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Hydrol. Earth Syst. Sci., 10, 507–518, 2006 www.hydrol-earth-syst-sci.net/10/507/2006/ Hydrology and © Author(s) 2006. This work is licensed Earth System under a Creative Commons License. Sciences

New lessons on the Sudd hydrology learned from remote sensing and climate modeling

Y. A. Mohamed1,2, H. H. G. Savenije1,3, W. G. M. Bastiaanssen4, and B. J .J. M. van den Hurk5 1UNESCO-IHE, P.O. Box 3015, 2601 DA Delft, The Netherlands 2IWMI Basin and Eastern Africa Sub Region, P.O. Box 5689, Addis Ababa, Ethiopia 3Delft University of Technology, Stevinweg 1, 2628 CN Delft, The Netherlands 4ITC (currently with WaterWatch), Generaal Foulkesweg 28, 6703 BS Wageningen, The Netherlands 5KNMI, P.O. Box 201, 3730 AE De Bilt, The Netherlands Received: 21 June 2005 – Published in Hydrol. Earth Syst. Sci. Discuss.: 16 August 2005 Revised: 24 March 2006 – Accepted: 12 June 2006 – Published: 10 July 2006

Abstract. Despite its local and regional importance, hydro- (with and without the Sudd wetland). The paper presents meteorological data on the Sudd (one of Africa’s largest wet- some of the model results addressing the Sudd’s influence on lands) is very scanty. This is due to the physical and political rainfall, evaporation and runoff of the river Nile, as well as situation of this area of Sudan. The areal size of the wetland, the influence on the microclimate. the evaporation rate, and the influence on the micro and meso The paper presents a case study that confirms the feasibil- climate are still unresolved questions of the Sudd hydrology. ity of using remote sensing data (with good spatial and poor The evaporation flux from the Sudd wetland has been temporal coverage) in conjunction with a regional climate estimated using thermal infrared remote sensing data and model. The combined model provides good temporal and a parameterization of the surface energy balance (SEBAL spatial representation in a region characterized by extremely model). It is concluded that the actual spatially averaged scarce ground data. evaporation from the Sudd wetland over 3 years of differ- ent hydrometeorological characteristics varies between 1460 and 1935 mm/yr. This is substantially less than open wa- ter evaporation. The wetland area appears to be 70% larger 1 Introduction than previously assumed when the Sudd was considered as an open water body. The temporal analysis of the Sudd evap- Wetland development projects (conservations, resource uti- oration demonstrated that the variation of the atmospheric lization, etc.) require an accurate knowledge of the water demand in combination with the inter-annual fluctuation of balance components over the wetland: precipitation, evap- the groundwater table results into a quasi-constant evapora- oration, inflow, outflow and interaction with groundwater. tion rate in the Sudd, while open water evaporation depicts a Similarly, evaporation and biophysical characteristics of the clear seasonal variability. The groundwater table character- wetland are required to better understand its interaction and izes a distinct seasonality, confirming that substantial parts feedback with the atmosphere. Usually evaporation from a of the Sudd are seasonal swamps. wetland is a major component of its water budget, though The new set of spatially distributed evaporation parame- complex to determine (Linacre et al., 1970). ters from remote sensing form an important dataset for cal- Remote sensing techniques are increasingly employed to ibrating a regional climate model enclosing the Nile Basin. estimate land surface evaporation (see a review in Choud- The Regional Atmospheric Climate Model (RACMO) pro- hury, 1989; Kustas and Norman, 1996; Menenti, 2000; Kus- vides an insight not only into the temporal evolution of the tas et al. 2003; Coureault et al., 2005). The technique is even hydro-climatological parameters, but also into the land sur- more attractive to derive evaporation and biophysical char- face climate interactions and embedded feedbacks. The im- acteristics over wetlands, characterized by difficult accessi- pact of the flooding of the Sudd on the Nile hydroclimatology bility. The actual evaporation (Ea) from wetland includes has been analysed by simulating two land surface scenarios all evaporation forms: open water evaporation, plant tran- spiration and wet/dry soil evaporation. A distinctive advan- Correspondence to: Y. A. Mohamed tage of remote sensing measurement is that it provides an ([email protected]) optimal spatial distribution from several kilometers to a few

Published by Copernicus GmbH on behalf of the European Geosciences Union. 508 Y. A. Mohamed et al.: New lessons on the Sudd hydrology meters. On the other hand, a major limitation is that the tem- searchers show no consensus. The JIT (1954) and Howell et poral distribution of satellite-based estimates is poor, and that al. (1988, p. 375) suggest no impact is expected on the re- interpolation techniques are necessary to define evaporation gional climate by draining part of the Sudd by the Jonglei between satellite overpasses. In this respect hydrological and canal. Eltahir (1989), Eagleson (1986) among others sug- climate models once properly calibrated can be very effective gest that the evaporation from the Sudd would surely be felt to fill in the gaps between satellite overpasses. climatically over a wider region. World wide, numerous field experiments have been exe- This paper presents new insight into the Sudd hydrology cuted to measure and model wetland evaporation. However, derived from remote sensing energy balance modeling and results remain site-specific and are difficult to extrapolate in numerical climate modeling. The spatial and temporal vari- space and time. In general, wetland evaporation is estimated ability of evaporation and biophysical properties have been based on either direct measurements or through modeling. determined from NOAA-AVHRR LAC (National Oceanic Examples of direct measurements are: Rijks (1969), Jose` et Atmospheric Administration – Advanced Very High Reso- al. (2001) and Jacobs et al. (2002) using energy balance tech- lution Radiometer Local Area Coverage) satellite images, niques (Bowen ratio and eddy correlation methods). The wa- using the SEBAL algorithm (Surface Energy Balance Algo- ter balance approach to estimate wetland evaporation either rithm for Land). The derived evaporation provides an essen- as the balance of the whole wetland or as a measuring tech- tial input to hydrological and climate models, which provides nique, e.g., Lysimeter and water tank experiments was used further in-depth understanding of the wetland system and its by Butcher (1938), Lott and Hunt (2001). Applications of impact on the surroundings. Section 2 of the paper gives a remote sensing to estimate wetland evaporation exist but are brief description of the Sudd wetland. Section 3 shows the very limited (e.g., Bauer et al., 2002). evaporation result and how it has been calculated. Section 4 Some of the wetland evaporation studies assume that wet- presents some results of the Sudd impacts on local and re- land Ea resembles open water evaporation Ew (Penman, gional climate. Finally a summary of the results and conclu- 1963); others assume that Ea resembles the potential evapo- sions is outlined in Sect. 5. ration Ep, i.e. evaporation from vegetative cover with no wa- Parts of this manuscript have been published in previous ter constraint (e.g., Lott and Hunt, 2001). In such cases Ea articles. Detailed analysis of the spatial variability of evap- is computed from routine meteorological data using formu- oration over the Sudd derived from remote sensing was dis- lae like: Penman 1948, Priestley-Taylor, Penman-Monteith cussed in Mohamed et al. (2004). The regional climate mod- (P-M), among others (see a review in Jacobs et al., 2002). eling part and assessment of the Sudd’s impact on the Nile However, a wetland system is a mixed composition of marsh- hydroclimatology was discussed in Mohamed et al. (2005a) land vegetation types, open water bodies and (un)saturated and Mohamed et al. (2005b). However, the main objective soil. Depending on the vegetation canopy structure, the wet- of this article is to present the piece-wise results published land vegetation may intercept the incoming solar radiation, previously pertinent to the Sudd question into one paper, and can shelter the blowing wind. The question is: Does supported by new computations for 2 additional years. Fur- the transpiration provided by the wetland vegetation offset thermore, new results on the temporal evolution of the Sudd the deficit caused by the vegetation shading or exceeds it evaporation and associated biophysical prosperities provides (Gilman, 1994)? additional knowledge on the seasonality of the Sudd evap- The Sudd is a huge interconnected wetland located on the oration compared to open water evaporation, an assumption Nile. About half of the river flow spills over and persisted for decades on the Sudd studies. evaporates from the Sudd. Water resources planners search- ing for additional Nile water have had the intention to build short cut channels to divert river water from upstream the 2 Study area: the Sudd wetland Sudd (e.g. the Jonglei canal). Despite the intensive studies conducted to understand the Sudd hydrology and assess the The Sudd wetland is one of the biggest swamps in Africa, impacts of water diversion (e.g., JIT, 1954; Howel et al., neighboring the smaller wetlands of the Bahr el Ghazal and 1988) still many questions remain unresolved. The exact the Machar marshes (Fig. 1). The permanent swamps, usu- evolution of the Sudd boundary is unknown. There are at- ally close to the main river course are permanently wet. How- tempts to define its size based on areal surveys (JIT, 1954), ever, substantial parts of the Sudd are seasonal swamps cre- based on hydrological models (Sutcliffe and Parks, 1999), ated by flooding of the Nile or when ponds are filled sea- based on remote sensing (Travaglia et al., 1995), or based on sonally with rainwater. Depending on the definition, the sur- remote sensing and hydrological models (Mohamed et al., face area is approximately 30 000 to 40 000 km2. The area of 2004). Similarly, the literature shows a wide range of evapo- the permanent swamps has tripled after the immense flood- ration estimate over the Sudd, between 1530 to 2400 mm/yr ing of the early 1960’s (Sutcliffe and Parks, 1999). The Sudd (Butcher 1938, Mijahid, 1948, Sutcliffe and Parks 1999). terrain is generally flat, composed of clayish soils, usually An important unresolved question, is how much the mois- poor in nutrients. Rain falls in a single , lasting from ture feedback to the atmosphere is? Here also different re- April to November and varying in the Sudd area from about

Hydrol. Earth Syst. Sci., 10, 507–518, 2006 www.hydrol-earth-syst-sci.net/10/507/2006/ Y. A. Mohamed et al.: New lessons on the Sudd hydrology 509

Fig. 1. Location of the Nile Basin and the Sudd wetland.

900 mm/yr in the south to 800 mm/yr in the north. Tempera- area of approximately 30% will be drained. The additional tures average to 30–33◦C during the dry season, dropping to gained water amounts to about 4 Gm3/yr (JIT, 1954; Howell an average of 26–28◦C in the rainy season. et al., 1988). Due to the war in the southern part of Sudan the The Sudd environment supports a variety of vegetation work on the canal stopped in 1983. species including: Cyperus papyrus, Phragmites (reed), Ty- pha swamps (cattail), Wild rice (Oryza longistaminata). The 3 Estimation of the Sudd evaporation Echinochloa pyramidalis grasslands dominate the seasonally inundated floodplains. Beyond the floodplain, Hyparrhenia Accurate determination of the Sudd evaporation is hindered rufa grasslands cover the rain-fed wetlands. Acacia seyal and by its immense size and difficult accessibility. Earlier at- Balanites aegyptica woodlands border the floodplain ecosys- tempts to measure evaporation in the Sudd started by the tem (Denny, 1991). The Sudd wetland is very important to experiments of Butcher (1938) and Migahid (1948) and the the pastoral economy of the local inhabitants (cattle grazing), calculations of Hurst and Philips (1938). The JIT (1954) and and the swamps support rich biota, including different bird Sutcliffe and Parks (1999) estimated the Sudd evaporation and mammal species. as being similar to open water evaporation. In this study, the The average annual Nile flow in and out of the Sudd for actual evaporation of the Sudd is estimated through the appli- the period 1961–1983 is 49 and 21 Gm3/yr, respectively. The cation of the SEBAL remote sensing algorithm that utilizes difference can be ascribed to evaporation, and an amount of NOAA-AVHRR images. 28 Gm3/yr of evaporative depletion has attracted planners to build short cut channels for bypassing the river water. The 3.1 The SEBAL algorithm Jonglei canal phase 1 is the first phase in a series of pro- posed water conservation projects. The canal (360 km long, The SEBAL algorithm is an energy-partitioning algorithm 2/3 completed) has an average bed width of 38 m, 4 to 8 m over the land surface, which estimates the actual evaporation deep, with a ground slope varying between 7 to 12 cm/km. from satellite images (Bastiaanssen et al., 1998a). The min- If Nile water resources upstream of the Sudd are pushed into imum input requirements are routine meteorological station the Jonglei canal, there will be less flooding and a wetland data. The satellite image provides an excellent spatial cover- www.hydrol-earth-syst-sci.net/10/507/2006/ Hydrol. Earth Syst. Sci., 10, 507–518, 2006 510 Y. A. Mohamed et al.: New lessons on the Sudd hydrology

energy partitioning: λE λE 1 3 = a = a = (2) λEa + H Rn − G0 1 + β

where β is the Bowen ratio (H/λEa). The evaporative frac- tion shows less variation during the daytime than the Bowen ratio as was investigated over the savannah landscape in Kenya by Farah et al. (2004). Detailed description of the SEBAL algorithm including verification results can be found in Bastiaanssen et al. (1998a, b, 2005); Allen et al. (2002) among others. The main assumption to obtain daily evapo- ration from the instantaneous SEBAL results is that the in- stantaneous evaporative fraction is equal to its daily value integrated over a period of 24 h (e.g., Brutsaert and Sugita, 1992), although newer versions of SEBAL allow to make this flexible for the inclusion of intermittent cloud cover and advection processes. The daily soil heat flux is assumed neg- ligible as it balances out during day and night. The daily net radiation is obtained from routine meteorological data at the ground stations. The daily evaporation is calculated as the instantaneous evaporative fraction times the daily net radiation. The monthly evaporation results are obtained by extrapolating daily evaporation data assuming that the daily Fig. 2. Mean annual evaporation over the Sudd (mm/yr), mean of ratio of actual evaporation to reference evaporation is valid 1995, 1999 and 2000. also for a monthly time step (Allen et al., 2002). Daily and monthly reference evaporation were computed by the Penman-Monteith equation based on routine weather data age with a resolution of 1 km. The temporal coverage is lim- measured at ground stations in the Sudd area. It can be seen ited to the time of the satellite overpass. So, the derived pa- that extrapolation from daily to monthly evaporation involves rameters need to be extrapolated to daily and monthly values assumptions which may not be completely satisfied in real- using various techniques. In this paper, the temporal charac- ity. The extrapolation to monthly maps is based on three teristic of the Sudd evaporation has been studied by repetitive daily evaporation maps on day: 5; 15; and 25 (Mohamed calculations for years 1995, 1999 and 2000. et al., 2004). If the soil moisture condition changes immedi- The SEBAL algorithm computes the latent heat flux as the ately after a given day, the ratio may not be representative for residue of the energy balance equation: the coming 10 days. Knowing this limitation, however, we believe that the error introduced is small, and that there is a high probability of positive and negative values to cancel out. λE = R − G − H (1) a n 0 However, correct monthly evaporation maps should be based on daily maps for the entire month. Availability of suitable 2 where Rn is the net radiation over the surface (W/m ), G0 is NOAA-AVHRR images and the very long calculation pro- 2 2 the soil heat flux (W/m ), H is the sensible heat flux (W/m ), cess drove us to adopt this approach. We may conclude that 2 λEa is the latent heat flux (W/m ) and λ is the latent heat compared to the assumption of constant evaporative fraction of vaporization (J/Kg). The major SEBAL steps required to for the whole month, used by some researchers, this approach produce an evaporation map are: (i) Pre-processing of the is considered adequate. satellite image (radiometric correction, geometric correction and removal of cloud pixels), (ii) Computation of the Soil 3.2 Spatial variability of the Sudd evaporation Vegetation Atmosphere Transfer (SVAT) parameters, includ- ing: surface albedo r0, Leaf Area Index ILA, thermal infrared More than 115 satellite images have been processed with SE- emissivity ε0, surface roughness z0m, land surface tempera- BAL to obtain monthly evaporation maps for the years 1995, ture T0, (iii) Computation of Rn and G0, (iv) The sensible 1999 and 2000. The 3 years have different hydrometeorolog- heat H is computed based on an iteration procedure that de- ical conditions. Figure 2 gives the mean annual evaporation scribes buoyancy effects on the aerodynamic resistance of from the 3 years of data. the land surface rah, (v) Computation of instantaneous latent The derived evaporation results over the Sudd can be ver- heat flux λEa and instantaneous evaporative fraction 3. The ified through water balance computations. The areal size of evaporative fraction, is a key parameter in SEBAL to express the Sudd is one of the key problems for assessing the water

Hydrol. Earth Syst. Sci., 10, 507–518, 2006 www.hydrol-earth-syst-sci.net/10/507/2006/ Y. A. Mohamed et al.: New lessons on the Sudd hydrology 511 balance as it varies during the different of the year. 10 1 Ea There is an ongoing debate on the Sudd boundaries, and also Ew Ea/Ew the boundary between the Sudd and the neighboring Bahr el 8 0.8 Ghazal swamps is highly questionable. The annual evapora- tion map can be considered as a suitable indicator for the an-

6 0.6 (-) w nual wetland area. However, seasonally the area can be dif- (mm/day) w /E a ferent (Mohamed et al., 2004). The Sudd boundary based on E , E a

this assumption is shown in Fig. 2. Please note that, no dis- E 4 0.4 tinction is made between seasonal and permanent swamps, however, a clear distinction is seen between the swamp area 2 0.2 (influenced by river flooding) and the surrounding area sub- Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec ject to rainfall alone. The delineated Sudd area is 38.6 Gm2, which is 74% larger than the value used in the past when the Fig. 3. Monthly fluctuations of actual Sudd evaporation Ea, open Sudd was considered an open water body. water evaporation Ew (mm/day), and the ratio (Ea/Ew), averaged The monthly water balance of the Sudd for the years 1995, values over the Sudd (mean of years 1995, 1999, 2000). 1999 and 2000 is computed by: dS created by river flooding and precipitation. So knowledge of = Q + P − E − Q (3) dt in a out vegetation characteristics and their impact on the canopy re- sistance and aerodynamic resistances rc and ra respectively dS dt Q where / is the monthly change of storage volume. in is are key to explaining the range of Ea values in relation to the estimated monthly river inflow at Mangala based on Lake open water evaporation Ew. This can be explained by fur- Victoria outflows (corrected for Lake Kayoga and Lake Al- ther investigation of the biophysical properties of the Sudd bert contributions) and the torrents flow between lake Albert wetland in relation to open water properties. The relation- and Mangala. The monthly rainfall P over the Sudd is taken ship between biophysical properties and the evaporative flux as the average of Juba, Malakal and Wau. Outflow Qout has can be derived from the Penman-Monteith energy balance been derived from the inflow based on the equation of Howell combination equation: et al. (1988, p. 497). Evaporation Ea is estimated by SEBAL over the Sudd area. The results of the monthly water balance 0.4081(R −G) + γ 900 U (e − e ) = n T +273 2 a d are given in Appendix A. The results show acceptable annual ET o (4) 1 + γ (1 + 0.34 U2) closure error of 0.1, −0.6 and −3.0% of E for the 3 years a ◦ 1995, 1999 and 2000, respectively. where 1 (Pa/C ) is the slope of the saturated vapor pressure ◦ Although satisfactory validation results of SEBAL were curve, cp(J/kg/C ) is the specific heat at constant air pressure, 3 obtained for 3 different catchments (Sudd, Bahr El Ghazal, ρa (kg/m ) is the air density, (es−ea) is the vapour pressure ◦ and the Sobat) for year 2000 as discussed in (Mohamed et al., deficit in (Pa), γ (Pa/C ) is the psychrometric constant. rs 2004), the water balance calculation given here confirmed (s/m) is a mixture of rc (s/m) that dictates canopy transpi- the obtained results by data from additional two years over ration, soil resistance that controls soil evaporation and the the Sudd. The three years 1995, 1999, and 2000, were criti- resistance for open water (usually zero when the water body cally selected to represent dry, wet, and medium hydrological is unpolluted). rs is equal to rc if soil and water surfaces are condition over the Sudd, respectively. completely covered by vegetation in a wetland ecosystem. The biophysical parameters of the Sudd were calculated 3.3 Temporal variability of evaporation and biophysical from the AVHRR images using semi-empirical formulae, properties over the Sudd which are part of SEBAL algorithm (Mohamed et al., 2004). The parameters include: r0, ILA, ε0 and z0m. The bulk sur- The seasonal variability of the Sudd Ea is given in Fig. 3 face resistance rs is calculated backward from the SEBAL (mean monthly values of the 3 years). For comparison, the Ea using the inverse Penman-Monteith equation (Eq. 4). The open water evaporation Ew is plotted in the same figure. areal mean values of the biophysical parameters over the en- Ew has been calculated with the Penman-Monteith equation tire Sudd averaged for the years 1995, 1999 and 2000 are (Eq. 4) using the Sudd meteorological data and the physi- presented in Fig. 4a and b. The temporal variability clearly cal properties of water. The data reveals a clear seasonality reflects seasonal climate influence (Rn,(es−ea), 1), and of Ew (high during the dry season and low during the rainy the hydrological condition caused by precipitation and Nile months), whereas Ea is remarkably stable. As mentioned flooding as depicted in Fig. 5a and b. The data of Fig. 5a above, the Sudd wetland is not a pure water body, instead it and b were derived from ground gauging stations around the is a swampy area partially covered with wetland vegetation. Sudd area. Secondly, it is composed of permanent swamps (wet all year Figure 4a shows a clear seasonality of ILA, in accordance around) located close to the river course, and seasonal parts with the rainfall season and river flooding as presented in www.hydrol-earth-syst-sci.net/10/507/2006/ Hydrol. Earth Syst. Sci., 10, 507–518, 2006 512 Y. A. Mohamed et al.: New lessons on the Sudd hydrology

(a) (b) 1 500

ILA rs r r 0.8 0 a 400

) (m) 10xz0m 0M 0.6 300 (s/m)

0.4 200 s , (10xZ 0 , r

0.2 100 a r , ,r LA I 0 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Fig. 4. (a) Monthly fluctuations of Leaf Area Index ILA (–), albedo r0 (–), and (10 times) surface roughness height 10×z0m (m). (b) Monthly fluctuations of surface resistance rs and aerodynamic resistance ra (s/m), averaged values over the Sudd (mean of years 1995, 1999, 2000).

Fig. 5b. High ILA values occur during the peak rainy season. Therefore, the possible explanation for the quasi-steady The roughness height z0m follows the ILA curve (by con- variation of Ea in contrast with Ew, is that net radiation 2 struction) and has a peak value in August and lowest value varies only between 120 to 150 W/m and that (es−ea) and rs in the dry months of February-April. The albedo r0 behaves have cancelling effects due to their natural feedback mecha- fairly stable in time. The surface thermal infrared emissivity nisms as described by Jarvis (1976) and Stewart (1988). This ε0 shows small seasonal variability between 0.92 in the dry is an important conclusion for this tropical wetland in Sudan. season to 0.96 in the rainy season (not shown here).

The lowest (es−ea) is recorded during the rainy sea- son elapsing from July–October, while highest values are 4 Regional climate modeling recorded in the dry season from November-April (Fig. 5a). Due to the higher cloud cover during periods of higher solar An analysis of the Sudd hydrology by a regional climate radiation in August-October, the net radiation remains fairly model (RCM) is essential to understand the land-surface cli- stable over time (Fig. 5a). This has an important impact on mate interaction in the region and assess the impact of future Ea, which usually obeys the temporal pattern of Rn. The scenarios (e.g. draining the swamps). It is widely believed slope of the vapor pressure curve 1 shows small variabil- that large land use changes can impose changes on regional ◦ ity throughout the seasons (0.18 to 0.24 kPa/ C) (data not climate. Similarly, climate change directly influences basin shown), because the temperature is fairly stable throughout hydrology, and subsequently the water resources. Evapo- the year. ration from a large area contributes to atmospheric mois- The hydrological control on evaporation depends on mois- ture through moisture recycling and enhances precipitation ture availability in the root zone, which governs the surface downwind (Savenije, 1995; Schar¨ et al., 1999). While many resistance rs. Figure 4b shows that the surface resistance rs researchers support positive soil moisture atmosphere feed- has a distinct seasonal variability, which is consistent with back, i.e., an increased soil moisture anomaly favors an in- the inter-seasonal variation of (es−ea) and ILA. The lowest crease of precipitation (e.g., Betts et al., 1999), there are rs values are associated with the lowest (es−ea) and the high- researchers who claim a negative soil moisture feedback, est ILA during the wet months July–October, and the reverse which is attributed to increased convective precipitation due occurs during the dry months February–April. The variabil- to enhanced buoyancy over the dried soils (e.g., Ek and Holt- ity of rs also correlates with the river flow regime and the cal- slag, 2004). Soil moisture-atmosphere feedback in a region culated ground water table fluctuations. A qualitative assess- depends on the climate system of the region and how it has ment of the temporal variability of the groundwater level over been modeled. The literature shows no consensus on the im- the Sudd derived from satellite data and water balance calcu- pact of draining the Sudd on the regional water cycle. lations is presented in Fig. 6. It shows a profound seasonal In this section we present the experience of the develop- variation (lowest in May and highest in October), demon- ment of an RCM over the Nile Basin to study the role the strating that the majority of the Sudd is non-inundated and Sudd wetland has on regional atmospheric circulation pro- has a seasonal decaying vegetation system. Only the lower cesses. Two simulations are compared: the present clima- parts near the riverbed are permanently saturated. The lower tology (control run CTL) in which the Sudd is seasonally groundwater table reduces the soil moisture in the root zone flooded, and a drained Sudd scenario in which the Nile in- and increases the leaf water potential, i.e. higher rs values. flow into the Sudd is stopped by means of a by pass that

Hydrol. Earth Syst. Sci., 10, 507–518, 2006 www.hydrol-earth-syst-sci.net/10/507/2006/ Y. A. Mohamed et al.: New lessons on the Sudd hydrology 513

(a) (b) 350 9

Rn P

) (kPa) 300 100x(es-ea) Qin 7.5 a -e s

250 6 /month) 3

200 4.5

150 3 (Gm ), 100x(e in 2

100 1.5 P, Q

(W/m 50 0 n Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec R

2 Fig. 5. (a) Monthly fluctuations of net surface radiation Rn (W/m ), 100 times vapor pressure deficit 100×(es−ea) (kPa). (b) Monthly 3 fluctuations of inflow Qin and rainfall P (Gm /month), averaged values over the Sudd (mean of years 1995, 1999, 2000).

prohibits any Nile water to flow into the marshland (Drained 0 600 GWT run DRA). rs 500 -0.5 The RCM is based on RACMO (Regional Atmospheric 400 Climate Model), described in Mohamed et al. (2005a). It is the main limited area model used by KNMI (The Royal -1 300 (s/m) s GWT (m) Netherlands Meteorological Institute) for climate research. 200 r ◦ ◦ ◦ The model extends between 10 E to 54.4 E and 12 S to -1.5 36◦ N (Fig. 1), and has a horizontal resolution of 50 km, 100 including 31 levels in the vertical. The initial and lateral -2 0 boundary conditions are taken from the ECMWF ERA-40 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec reanalysis data. The model has been adjusted to simulate Fig. 6. Monthly fluctuations of the areal average groundwater table the spilling of the Nile water over the Sudd by routing the GWT (m), and surface resistance rs(s/m), averaged values over the flow generated from upstream catchments to 15 grid points Sudd (mean of years 1995, 1999, 2000). that represent the Sudd wetland. The simulation period ex- tends from 1995 to 2000. The model has been validated against various observational datasets including: radiation, 4.1 Impact of the Sudd wetland on the Nile hydroclimatol- precipitation, runoff, and evaporation. The default RACMO ogy settings were first modified based on a series of a one year simulations aimed at: modifying land-cover representation; Complete diversion of all natural Nile water inflow into the smoothening of orography; reduction of hydraulic conductiv- Sudd may have several implications on hydroclimatology ity; adjustment of aerosols to increase incoming short wave both locally and regionally. During the dry period, December radiation (Sect. 3.2 of Mohamed et al., 2005a). The SEBAL to March, soil moisture and evaporation over the Sudd are re- estimations of evaporative fraction, soil moisture and evap- duced substantially in the DRA run. This causes a reduction oration have been used to adapt the input parameters of the of the screen level humidity (RH) by 30 to 40% compared to RACMO land surface model, notably the (i) the minimum the CTL run, while the difference is small (<10%) in the wet canopy resistance has been reduced, (ii) a higher soil mois- season June to September (Fig. 7a and b). The incoming so- ture content has been established by introduction of flood- lar energy, which is partitioned into latent and sensible heat ing, (iii) the soil layer depth has been increased for larger in the CTL scenario, with minimal soil heat in the wetland, storage capacity and (iv) the hydraulic conductivity has been will be partitioned mainly into sensible heat at low values of decreased to reduce deep percolation. These changes in the soil moisture during the dry season. This results in a rise of model significantly affect the partitioning of net radiation the screen level temperature (T) by 4 to 6◦C on the drained into surface sensible and latent heat fluxes. The detailed parts in the dry season (Fig. 7c), while it shows a small rise of validation results are given in Mohamed et al. (2005a). In 0.5◦C in the wet season (Fig. 7d). There are small changes general, the model provided a sound representation of the of RH and T outside the Nile Basin on the eastern part of hydroclimatological processes over the region. The model the Congo Basin (increased RH and reduced T). A possible has been run for a second scenario with the Sudd completely explanation for this is that air passing over the drained Sudd drained (Mohamed et al., 2005b). (from east to west in the dry season) has higher water uptake www.hydrol-earth-syst-sci.net/10/507/2006/ Hydrol. Earth Syst. Sci., 10, 507–518, 2006

514 Y. A. Mohamed et al.: New lessons on the Sudd hydrology

(a) (b)

(a) (c) 24°E 32°E 24°E 32°E

16°N 16°N N N ° ° 16 16

8°N 8°N N N ° ° 8 8

0° 0°0° 0° 0°

24°E 32°E 24°E 32°E

-0.4 -0.3 -0.1 -0.05 0.05 0.1 0.3 0.4 -0.4 -0.3 -0.1 -0.05 0.05 0.1 0.3 0.4

(c) (d) (a) (c) 24°E 32°E 24°E 32°E

16°N 16°N N N ° ° 16 16

8°N 8°N N N ° ° 8 8

-0.5

0° 0° 0° 0°

-0.5

24°E 32°E 24°E 32°E

-6 -4 -2 -0.5 0.5 2 4 6 -6 -4 -2 -0.5 0.5 2 4 6

Fig. 7. (a) Change of the relative humidity RH in the dry season, (b) change of RH in the wet season (–), (c) change of temperature T in the dry season, (d) change of T in the wet season (◦C). Mean seasonal values of 1995 to 2000. capacity, which enhances water advection from further south, at smaller time steps (6 hourly) reveals that the stability of the while enhanced surface evaporation provides additional RH regional climate could be sensitive to draining the Sudd wet- and subsequently reduces T . During the rainy season, the lands. A possible explanation of the slight increase in P for impact of draining the Sudd is much less influencing since the drained Sudd scenario is due to the enhanced convection both the Sudd itself and the surrounding area are wet from over the dried soils. the rain. Although the evaporation rate over the Sudd is about 3 It appears that the role of the Sudd on the Nile Basin hy- times the average rate in the surrounding area, volume-wise drological budget is negligible. The inter-annual variability the ratio of the Sudd evaporation is very small compared to of: P , E, R, and dS/dt (change of sub-surface moisture stor- regional evaporation. Secondly, the Sudd evaporation consti- age) of the Nile catchment at the outlet is more dis- tutes only around 1% of the volume of the atmospheric mois- tinct than the impact of draining the Sudd wetlands (Fig. 8). ture flux over the Nile region, which is mainly of oceanic The error bars represent one standard deviation (std) around origin. Obviously theses results could have been expected in the mean. The difference in P , E, R and dS/dtfor the DRA view of the small size of the Sudd relative to the Nile Basin run, is much smaller than the standard deviation, indicating (about 1%). So, it can be concluded that in terms of mass an insignificant change compared to the inter-annual variabil- balance, the impact of the Sudd on the Nile’s atmospheric ity in the 6 years record. Detailed inspection of model results budget is negligible.

Hydrol. Earth Syst. Sci., 10, 507–518, 2006 www.hydrol-earth-syst-sci.net/10/507/2006/

Y. A. Mohamed et al.: New lessons on the Sudd hydrology 515

(a) Precipitation P (b) Evaporation E 5 5 CTL DRA 4 4

3 3

2 2 mm/day mm/day

1 1

0 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

(c) Runoff R (d) change of sub-surface storage dS/dt 5 2 4 1 3 0 2 mm/day mm/day -1 1 -2 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Fig. 8. The hydrological budget terms of the Nile at Aswan (mm/day), mean annual cycle 1995–2000: (a) precipitation, (b) evaporation, (c) runoff, (d) change of sub-surface moisture storage. The error bars represent one std around the mean of the CTL data.

Table 1. Annual water budget terms of the area upstream of the White Nile at Malakal in Gm3/yr (mean annual values of 1995 to 2000).

3 3 3 3 3 Qin Gm /yr P Gm /yr E Gm /yr dS/dt Gm /yr Qout Gm /yr Observed 0 – – 0 (assumed) 33 CTL 0 +(53) 899 879.7 −12.5 85.2 DRA 0 964 902.1 −15.5 78.3

The additional Nile water provided at the outlet down- Nile flooding over the wetland. This is slightly higher than stream of the Sudd is literally the whole river runoff gen- the observed Nile flow at Juba of 49 Gm3/yr (mean between erated upstream of the Sudd for the DRA run, while for the 1961 and 1983). Here the observed runoff is used, rather than CTL run, the major part of this upstream runoff is evapo- the computed runoff, since RACMO underestimates runoff at rated over the Sudd, in particular during the dry season. By this location (21 Gm3/yr). In the DRA run no water has been diverting water off the Sudd, evaporation becomes entirely distributed over the Sudd. dependent on rainfall. Part of the rainfall will contribute to runoff and to groundwater recharge, and the remaining part Since the 53 Gm3/yr volume has been supplied externally is evaporated. to simulate river routing and flooding over the wetlands, it is not part of the natural runoff yield. To assess the The White Nile catchment upstream of the outlet at runoff difference between the two runs, the artificially sup- Malakal (includes Sudd, Sobat and Bahr el Ghazal) has been plied water (53 Gm3/yr) is subtracted from the CTL runoff at selected to assess the amount of additionally gained runoff, Malakal, which yields to 32.2 Gm3/yr (85.2–53), Therefore, (catchment area is 1.48×106 km2). Table 1 shows the wa- the runoff difference between CTL and DRA runs amounts ter budget terms of the White Nile at Malakal. Inflow is to 46.1 Gm3/yr (78.3–32.2). This number is the extra outflow zero since the whole catchment upstream of Malakal has that can be expected to flow into the Nile downstream of the been considered and there is no substantial inter-basin trans- Sudd (for no flooding case). However, a correction is needed fer. For the CTL run, 53 Gm3/yr has been distributed over as it was found that RACMO has 10 Gm3/yr more evapo- the 15 grid points of the Sudd (4.1 mm/day) to represent the ration than SEBAL. So, a correction of the CTL runoff by www.hydrol-earth-syst-sci.net/10/507/2006/ Hydrol. Earth Syst. Sci., 10, 507–518, 2006 516 Y. A. Mohamed et al.: New lessons on the Sudd hydrology

10 Gm3/yr seems plausible. It is assumed that RACMO over- droclimatology are unresolved. This paper presents a con- estimation of evaporation is at the expense of the computed tribution towards better understanding the Sudd hydrology runoff. In the DRA run there is no overestimation of evapo- and its impact on the sub-continental atmospheric circulation ration in the dry season because of the absence of river flood- processes. ing. This implies that the “best guess” impact of draining the Remote sensing techniques (SEBAL algorithm) proved to whole Sudd wetlands yields an extra Nile discharge of 46– be instrumental in defining net radiation and actual evapora- 10=36 Gm3/yr. This is somewhat more than the long-term tion over the Sudd (characterized by scanty ground observa- 1961–1983 mean losses over the Sudd of 29 Gm3/yr given in tions). Monthly actual evaporation and soil moisture maps Sutcliffe and Parks (1999). The discrepancy is well under- during 3 years of different hydrometeorological conditions stood from the noted model deficiencies and bias corrections (1995, 1999 and 2000) have been prepared. The area of the applied. Sudd based on an average annual evaporation characteristic The table shows that for this catchment both P and E are is 38.6 Gm2. The annual evaporation rate for 1995, 1999 and relatively higher in the CTL run than the DRA run. As dis- 2000 is 1460, 1935 and 1636 mm/yr, respectively (hence 57, 3 cussed above for the whole Nile catchment (Fig. 8), a possi- 74, 63 Gm /yr). The Sudd actual evaporation Ea doesn’t ble explanation of the small increase in P in the drained Sudd show much seasonal variability, whereas open water body scenario is due to an enhanced convective precipitation over evaporation Ew clearly follows a seasonal climatic variation. the dried soils, and that the relatively higher precipitation The Ea value of the Sudd is quasi-steady state, which can produces an increased evaporation. Over the Sudd wetland be ascribed to the lack of seasonality of net radiation and the alone, evaporation is very much different in the two runs: canceling effect between the vapor pressure deficit and sur- 68 and 21 Gm3/yr in the CTL and DRA runs, respectively, face resistance throughout the season. The distinct seasonal- while precipitation is almost similar 22 and 21 Gm3/yr, re- ity of the surface resistance over the Sudd can be explained spectively. Obviously, less moisture is available for evapora- by the seasonality of the atmospheric vapor pressure deficit tion from the Sudd in the DRA scenario. and availability of soil moisture due to rainfall variability and It is to be mentioned that RACMO computation of catch- Nile flooding. ment runoff over the White Nile shows to be extremely sen- The interaction of the Sudd with the atmosphere has sitive to inaccuracy of either P or E. This is due to the ex- been studied with regional climate model simulations. The ceptionally small runoff coefficient over this vast catchment RACMO model has been calibrated using various datasets (Mohamed et al., 2005a). For both the CTL and DRA runs, including SEBAL outputs on evaporation and related param- RACMO overestimates the runoff at Malakal (Table 1), and eters. Two scenarios (with and without flooding) have been underestimates runoff at Juba (just upstream of the Sudd, simulated for the period 1995–2000. Evaluation of the re- see Fig. 1). The annual runoff computed at Juba is 21 and sults from the two simulations has shown that draining the 24 Gm3/yr for CTL and DRA, respectively, while the ob- entire Sudd will have a significant impact on the microcli- served long-term (1961–1983) mean is 49 Gm3/yr. Although mate. The relative humidity will drop by 30 to 40%, and the ◦ RACMO is not suitable for distributed hydrological analysis local temperature rises by 4 to 6 C during the dry season. – it shows to be very instrumental for estimating the major During the wet season the impact of the Sudd can hardly be changes in the water balance. discerned, because the surrounding area is saturated by rain. Based on the design capacity of the canal, the uncom- The simulated results show that the impact of the Sudd on the pleted Jonglei canal planned to divert 4 Gm3/yr. Although regional hydrological budget of the Nile Basin (precipitation, not modeled in this study, the general perception is that an evaporation, runoff and sub-surface storage) is negligible and amount of 4 Gm3/yr, which is 8% of longer-term inflow into insignificant compared to the inter-annual variability of these the Sudd, will drain about 30% of the Sudd wetland (Howell parameters. The net gain of the Nile water by complete diver- et al., 1988). In view of the atmospheric modeling results ob- sion of the Nile water from the Sudd would be an additional tained, it can be stated safely that 30% reduction of the Sudd ∼36 Gm3/yr, which is more than the observed Nile evapora- area will have no alternation of the regional rainfall patterns, tion over the Sudd (29 Gm3/yr). while the impact on micro-climatic and near-surface weather Further research is desired to confirm the Sudd evapora- conditions during the dry season can be expected (analogous tion estimates against ground observations. Distributed hy- to the obtained results). drological modeling of the Sudd is needed to better under- stand the spatial and temporal evolution of the permanent and seasonal swamps. For the regional climate modeling it 5 Conclusions would be interesting to repeat the numerical experiment for a longer time span (40 yr), and finer model resolution (smaller Despite the importance of the Sudd wetland, both for the lo- than 50 km grid). However, the main conclusions derived cal environment, an international bird paradise and as an ex- here are likely to be confirmed rather than discarded, since pected additional supplier of the Nile water, still many ques- the hydrological fluxes to and from the Sudd are relatively tions related to its hydrology and impact on the regional hy- small compared to the atmospheric fluxes.

Hydrol. Earth Syst. Sci., 10, 507–518, 2006 www.hydrol-earth-syst-sci.net/10/507/2006/ Y. A. Mohamed et al.: New lessons on the Sudd hydrology 517

Appendix A Acknowledgements. The material of this paper is part of a larger research study on moisture recycling over the Nile Basin funded Table A1. Monthly water balance of the Sudd wetland for 1995, by The International Water Management Institute IWMI, ITC units in Gm3/month. Enschede and the UNESCO-IHE Delft. The climate modeling part of the study has been carried out at The Royal Netherlands Month PEa Qin Qout dS/dt Meteorological Institute (KNMI). − Jan 0.00 4.42 2.95 1.50 2.97 Edited by: A. Bardossy Feb 0.12 3.95 2.57 1.40 −2.67 Mar 0.65 5.15 2.73 1.41 −3.18 Apr 3.13 4.72 2.70 1.36 −0.25 May 4.11 5.13 3.13 1.28 0.84 References Jun 6.57 4.74 2.86 1.31 3.37 Jul 7.28 4.79 3.17 1.31 4.35 Aug 4.20 4.94 3.35 1.39 1.22 Allen, R. G., Morse, A., Tasumi, M., Trezza, R., Bastiaanssen, W. Sep 5.49 5.10 3.74 1.34 2.79 G. M., Wright, J. L., and Kramber, W.: Evapotranspiration from Oct 4.17 4.89 3.75 1.40 1.64 a Satellite-BASED Surface energy balance for the Snake Plain Nov 0.14 4.44 3.17 1.43 −2.56 Aquifer in Idaho, Proc. National USCID Conference, San Luis Dec 0.01 4.26 3.24 1.49 −2.50 Obispo, California, pp. 16, 8–10 July 2002. Bastiaanssen, W. G. M., Menenti, M., Feddes, R. A., and Holt- Total 35.9 56.5 37.4 16.6 0.1 slag, A. A. M.: The Surface Energy Balance Algorithm for Land (SEBAL): Part 1 formulation, J. Hydrol., (212–213) 198–212, 1998a. Table A2. Monthly water balance of the Sudd wetland for 1999, Bastiaanssen, W. G. M., Menenti, M., Feddes, R. A., and Holtslag, 3 units in Gm /month. A. A. M.: The Surface Energy Balance Algorithm for Land (SE- BAL): Part 2 validation, J. Hydrol., (212–213) 198–212, 1998b. Month PEa Qin Qout dS/dt Bastiaanssen, W. G. M., Noordman, E. J. M., Pelgrum, H., Davids, Jan 0.00 5.45 4.01 1.74 −3.18 G., Thoreson, B. P., and Allen, R. G.: SEBAL Model with Feb 0.11 6.01 3.48 1.61 −4.02 Remotely Sensed Data to Improve Water-Resources Manage- Mar 0.41 5.87 3.70 1.59 −3.34 ment under Actual Field Conditions, ASCE J. Irrig. Drain. Eng., Apr 3.46 6.99 3.77 1.54 −1.29 131(1), 85–93, 2005. May 4.09 6.97 4.06 1.45 −0.27 Bauer, P., Brunner, P., and Kinzelbach, W.: Quantifying the Net Jun 6.86 6.05 4.18 1.49 3.50 Exchange of Water Between Land and Atmosphere in the Oka- Jul 6.09 6.23 4.20 1.50 2.57 vango Delta, Botswana, Conf. Proceedings Model Care 2002, Aug 7.39 6.14 4.87 1.54 4.58 Prag, Czech Republic, 581–584, 17–20 June 2002. Sep 6.20 5.89 4.97 1.56 3.71 Betts, A. K., Ball, J. H., and Viterbo, P.: Basin-scale water and Oct 5.81 6.24 5.39 1.57 3.39 energy budgets for the Mississippi from the ECMWF reanalysis, − Nov 0.38 6.32 4.45 1.67 3.15 J. Geophys. Res., 104D, 19 293–19 306, 1999. Dec 0.00 5.74 4.35 1.68 −3.07 Brutsaert, W. H. and Sugita, M.: Regional surface fluxes from Total 40.8 73.9 51.4 18.9 −0.6 satellite-derived surface temperatures (AVHRR) and radiosonde profiles, Boundary-Layer Meteorology, 58, 355–366, 1992. Butcher, A. D.: The Sadd hydraulics. Technical report, Ministry of Table A3. Monthly water balance of the Sudd wetland for 2000, Public Works, Cairo, , 1938. units in Gm3/month. Choudhury, B. J.: Estimating evaporation and carbon assimilation using infrared temperature data: Vistas in modeling, in: Theory and applications in optical remote sensing, edited by: Asrar, G., Month PEa Qin Qout dS/dt John Wiley, New York, 628–690, 1989. Jan 0.00 5.45 3.23 1.62 −3.84 Coureault, D., Seguin, B., and Olioso, A.: Review about estimation − Feb 0.00 4.80 2.81 1.47 3.47 of evapotranspiration from remote sensing data: from empirical − Mar 0.45 5.21 2.98 1.46 3.23 to numerical modeling approach, and Drainage Sys- Apr 2.26 5.45 2.94 1.41 −1.65 tems, 19, 3–4, 223–249, 2005. May 4.60 5.64 3.23 1.33 0.87 Denny, P.: Africa, in Wetlands, edited by: Finlayson, M. and Jun 5.05 5.33 3.38 1.36 1.74 Jul 8.83 4.98 3.55 1.35 6.04 Moser, M., International Waterfowl and Wetlands Research Bu- Aug 6.03 5.53 3.64 1.41 2.73 reau, Facts on File, Oxford, UK, pp. 115–148, 1991. Sep 4.19 5.21 3.86 1.43 1.40 Eagleson, P. S.: The Emergence of Global-Scale Hydrology, Water Oct 4.79 5.77 4.52 1.46 2.09 Resour. Res., 22(9), 6s–14s, 1986. Nov 0.43 4.68 3.61 1.48 −2.11 Ek, M. B. and Holtslag, A. A. M.: Influence of Soil Moisture on Dec 0.00 5.05 3.53 1.51 −3.04 Boundary Layer Cloud Development, J. Hydrometeorol., 5, 86– 99, 2004. Total 36.6 63.1 41.3 17.3 −2.5 Eltahir, E. A. B.: A Feedback Mechanism in Annual Rainfall in Central Sudan, J. Hydrol., 110 , 323–334, 1989. www.hydrol-earth-syst-sci.net/10/507/2006/ Hydrol. Earth Syst. Sci., 10, 507–518, 2006 518 Y. A. Mohamed et al.: New lessons on the Sudd hydrology

Farah, H. O., Bastiaanssen, W. G. M., and Feddes, R. A.: Evaluation Lott, R. B. and Hunt, R. J.: Estimating evapotranspiration in natural of the temporal variability of the evaporative fraction in a tropi- and constructed wetlands, Wetlands, 21, 614–628, 2001. cal watershed, Int. J. Appl. Earth Obs. Geoinform., 5, 129–140. Migahid, A. M.: Report on a Botanical Excursion to the Sudd Re- 2004. gion, Fouad I University Press, Cairo, Egypt, 1948. Gilman, K.: Water balance of wetland areas, Conf. on “The bal- Menenti, M.: Evaporation, in: Remote Sensing in Hydrology and ance of water – present and future”, AGMET Gp. (Ireland) & Water Management, edited by: Schultz, G. A. and Engman, E. Agric. Gp. of Roy. Meteorol. Soc. (UK), Dublin, 123–142, 7–9 R., Springer Verlag, Heidelberg, Chapter 8, 157–188, 2000. Sep 1994. Mohamed, Y. A., Bastiaanssen, W. G. M., and Savenije, H. H. Howell, P. P., Lock, J. M., and Cobb, S. M. (Eds): The Jonglei G.: Spatial variability of evaporation and moisture storage in the canal: Impact and Opportunity. Cambridge University Press, swamps of the upper Nile studied by remote sensing techniques, UK, pp. 536, 1988. J. Hydrol., 289, 145–164, 2004. Hurst, H. E. and Philips, P.: The hydrology of the Lake Plateau and Mohamed, Y. A., van den Hurk, B. J. J. M., Savenije, H. H. G., and Bahr el Jebel, The Nile Basin Vol. V, Government Press, Cairo, Bastiaanssen, W. G. M.: Hydroclimatology of the Nile: Results Egypt, 1938. from a regional climate model, Hydrol. Earth Syst. Sci., 9, 263– Jacobs, J. M., Mergelsberg, S. L., Lopera, A. F., and Myers, D. 278, 2005a. A.: Evapotranspiration from a wet prairie wetland under drought Mohamed, Y. A., van den Hurk, B. J. J. M., Savenije, H. H. G., conditions: Paynes prairie preserve, Florida, USA, WELANDS, and Bastiaanssen, W. G. M.: The Impact of the Sudd wetland 22(2), 374–385, 2002. on the Nile Hydroclimatology, Water Resour. Res., 41, W08420, Jarvis, P. G.: The interpretation of the variations in leaf water po- doi:10.1029/2004WR003792, 2005b. tential and stomatal conductance found in canopies in the field. Penman, H. L.: Vegetation and hydrology, Tech. Comm. No. 53, Philosophical Transactions of the Royal Society of London, (B) Commonwealth Bureau of Soils, Harpenden, England, pp. 125, 273, 593–610, 1976. 1963. JIT: The Equatorial Nile Project and its effects in the Anglo- Rijks, D. A.: Evaporation from a papyrus swamp, Quart. J. Roy. Egyptian Sudan. Report of the Jonglei Investigation Team, Sudan Meteorol. Soc., 95, 643–649, 1969. Government, , 1954. Savenije, H. H. G.: New definitions for moisture recycling and the Jose,´ J. S., Meirelles, M. L., Bracho, R., and Nikonova, N.: A com- relation with land-use changes in the Sahel, J. Hydrol., 167, 57– parative analysis of the flooding and fire effects on the energy 78, 1995. exchange in a wetland community (Morichal) of the Orinoco Schar,¨ Ch., Luthi,¨ D., Beyerle, U., and Heise, E.: The Soil- Llanos, J. Hydrol., 242 (3–4), 228–254, 2001. Precipitation Feedback: A Process Study with a Regional Cli- Kustas, W. P. and Norman, J. M.: Use of remote sensing for evapo- mate Model, J. Climate, 12(3), 722–741, 1999. transpiration monitoring over land surfaces, Hydrol. Sci. J., 41(4) Stewart, J. B.: Modelling surface conductance of pine forest, Agri- 495–515, 1996. culture Forest Meteorology, 43, 19–35, 1988. Kustas, W. P., Diak, G. R., and Moran, M. S.: Remote sensing Sutcliffe, J. V. and Parks, Y. P.: The Hydrology of the Nile, IAHS of evaporation, Encyclopedia of Water Science, Marcel Dekker Special Publication no. 5, IAHS Press, Institute of Hydrology, Inc., doi:10.108/E-EW120010313, 267–274, 2003. Wallingford, Oxfordshire OX10 8BB, UK, 1999. Linacre, E. T., Hicks, B. B., Sainty, G. R., and Grauze, G.: The Travaglia, C., Kapetsky, J., and Righini, G.: Monitoring wetlands evaporation from a swamp, Agric. Meteorology, 7, 375–386, for fisheries by NOAA AVHRR LAC thermal data. FAO/SDRN, 1970. Rome, Italy, 1995.

Hydrol. Earth Syst. Sci., 10, 507–518, 2006 www.hydrol-earth-syst-sci.net/10/507/2006/