The largest khor in HADR is Khor Toshka. This khor cannot be eliminated due to its function as a spillway for the HAD. Furthermore, the intake of the Shiekh Zayed canal is located on this khor. This canal will irrigate the new reclamation lands in the South of Egypt [21]. Therefore, Khor Kalabsha and Khor El-Alaky were selected as large khors to be studied. In addition, Khor Sara and Khor Korosko with their small surface areas were selected due to their high evaporation rates. Lake maximum level was 181.6 m AMSL Thus, the khors’ surface areas were measured with respect to it. The evaporation loss rates for each khor was estimated using the annual distribution of the evaporation losses over the lake presented in Figure 3.

Khor Kalabsha is the second largest khor in the lake. It is about 30 km upstream of the HAD, in the Western Desert. It has a surface area of 600 Km², about 10% of the lake’s entire area. It loses about 2700 mm of water annually, due to evaporation. Thus, it is expected that eliminating it will lead to a substantial reduction in evaporation losses. Khor El-Alaky is about 100 km upstream of the HAD and extends into the Eastern Desert. It has a large surface area of 500 km² and experiences 2500 mm evaporation losses annually. Thus, it is expected that eliminating this khor will help also in substantially decreasing evaporation losses. The other two khors investigated in this study, Koroso and Sara, are not as large as khors El-Alaky or Kalabasha; however, both experience high evaporation rates. Khor Korosko is found in the Eastern Desert at about 180 km upstream of the HAD and covers about 103 km² at water level 182 m. It is steep sided and relatively narrow with a rocky bottom. It loses about 3000 mm annually through evaporation. Khor Sara is located about 325 km upstream of the HAD reaching into the Western Desert. It is one of the miniature khors in HADR and has a surface area of only 50 km². It suffers from high evaporation losses of about 3100 mm per year.

The lake is meteorologically monitored to estimate the daily evaporation losses by six meteorological stations covering the entire lake. Data on daily evaporation losses at each station since its operation until 2010 were stored and assembled in a new meteorological database for HADR (HADRMTDB). Thus the daily, monthly and annually averages have been estimated. The mean daily evaporation losses per month for the entire lake, and the mean annual evaporation losses per station are presented in Figure 3. The annual evaporation losses estimated from these stations were converted into a GIS layer, and interpolated to produce the distribution of mean annual evaporation losses along the lake for remote areas as shown in Figure 3. Using the HADRMTDB, the mean evaporation losses are estimated to be 7.4 mm/day, equal to 2700 mm/year.

The lake surface water level and the lake flow measured daily at Aswan station since 1964 were stored and assembled in a new hydrological database (HADRHYDB). There are more than 16000 water level records covering the period from 1964 to July 2010 as shown in Figure 4. The lowest water level of 150.62 m AMSL was recorded on 21^st July 1988. The highest level of 181.6 m AMSL was measured during the period from 10th to14th November 1999. According to the HADRHYDB, the mean water level is about 178 m AMSL.

Lake Nasser covers large inaccessible areas, thus satellite imagery was used to define the lake actual morphology. New imagery was needed due to the lack of new up-todate topographic maps and the fact that the lake edges at different elevations have changed due to erosion events since HAD construction. Landsat ETM7 and Landsat TM5 were utilized in constructing new contour lines for HADR at different elevations. The lake is generally captured by a set of four images. Thus, over 300 images were downloaded to select the appropriate sets for each water level. The surface water level for every set was identified by using the developed HADRHYDB database. Hence, proper sets with similar water levels were selected and used in identifying the surface water area as listed in Table 1. The lake sides have a smooth incline, thus a step of 1.5 m is adequate to embrace various surface areas. Twenty imagery sets were used, consisting of 80 Landsat images for the years 1984 to 2005. The least surface area was obtained from the imagery set of August 1987 at water level of a about 154.5 m AMSL, while the largest area was attained from the imagery set of November 1999 when the lake reached its peak water level of 181.6 m AMSL.

All the images were transformed together. Each set of four images was mosaiced into one imagery using ERDAS IMAGINE. Band 4 is the near infrared band with wave length ranging between 0.77 and 0.90. Consequently, it emphasizes biomass content and shorelines as stated on USGS website http://landsat.usgs.gov/best_spectral_bands_to_use.php.

Thus the mosaiced imagery sets were imported into ENVI EX to delineate lake shorelines using band 4. The outputs of lake delineation were exported to ERADS IMAGINE to manually edit the edges to correct the classification errors occurring around the lake shore lines. The adjusted water layers were transformed then to shape files. These shape files were subsequently edited to eradicate all small polygons. Hence, the main polygon for each altitude was converted to a polyline representing contour lines of individual water levels. These polylines were attached to a new shape file layer for the contour lines at 154.5 m to 181.6 m AMSL as shown in Figure 5. The interpolation of these individual contour lines was

needed for water levels over 154 m and below 182 m AMSL. Using old aerial photo acquired in the 1960s which has the original Nile morphology before HAD construction, the contour line of 92 m was extracted.

Each year, the MWRI conducts a bathymetric survey for Lake Nasser, to identify the lakebed along many sections as shown in Figure 5. For altitudes below 154 m, the lakebed elevations computed using the bathymetric survey, were used as spot heights. Moreover, the MWRI monitors lake water quality and performs a geological survey along the lake shores. Past surveys showed that the sedimentation rate is particularly high in Lake Nubia which required surveying as presented in Figure 5. Elba visited HADR in January 2007, spending 30 days on the survey ship collecting the required data. Figure 5 shows 200,000 points along the lake where water depths were measured in 2007. The depths are subtracted from the lake surface water level measured during this phase to detect the lake bed altitudes above mean sea level. These levels are generated as point shape file. The generated layer was utilized as spot heights for the lake levels below 154 m within the lake Egyptian division, and below 160 m within the Sudan division due to the high sedimentation rates there [9].

Since the area around HADR was not subject to any erosion or sedimentation events, old topographic maps of scale 1:100,000 produced in the 1960s before HAD construction was used to delineate the contour lines for the region outside the lake margin at its maximum water level of 181.6 m AMSL The contour lines within the lake margin were erased using ARCGIS because they had changed.

The DEM was created using the contour lines extracted from the satellite images, the spot heights measured from the bathymetric survey of 2007, and the old topographic maps. It was generated using the Topo to raster interpolation method in ARCGIS with 25 m spatial resolution for the area extending between 21˚N and 24˚N and 31˚30'E and 33˚30'E. To study the lake hydrological features, HADR DEM (HADRDEM) was isolated using the lake boundary at 182 m as demonstrated in Figure 6.

HADRDEM was used initially to build mathematical models for HADR hydrological characteristics. Three mathematical models were built to indicate the relationship between the different hydrological characteristics: water level versus surface area; water level versus lake

volume; and surface area versus lake volume. The number of pixels at each elevation were imported from HADRDEM to excel files, to estimate the lake surface area and water volume per meter of lake altitude. Elevation, calculated area and volume were analyzed to estimate the proper curve using regression functions in SPSS, and develop the proper type for each equation. It was found that power equations fits the obtained mathematical models as illustrated in Equations (1)-(3).

WL = 37.724 A^0.184           (1)

WL = 96.332 V^0.126            (2)

A = 74.833 V^0.894            (3)

where WL = Water Level in m AMSL, A = Surface Area in (km²), and V= Water Volume in (BCM). Using equation 2, the lake volume was estimated to be 150 BCM at the water level of 181 m AMSL. This implies that the lake water capacity is decreased by about 6 BCM. This reduction is due to the 5 BCM sediment deposits in the Sudan portion and 1 BCM sediment deposits within the Egyptian part. Equation (1) was used to estimate the current evaporation losses, with respect to each water level, using the rates stored in the HADRMTDB.

The HADRDEM was evaluated to investigate its quality and significance in estimating the surface area and the water volume of the HADR. Figure 7 shows a scatter diagram of the surface area estimated by the HADRDEM and the NWS. The points are highly correlated and the relationship between them can be presented with a linear equation. The differences between both areas are ±100 km² at maximum due to the morphological changes of the HADR for the last four decades. Figure 8 shows the HADR’s altitudes and the corresponding water volume estimated by the HADRDEM and the NWS. Both lines are parallel and have a difference of about 7 BCM between both volumes due to the sediment events over the last five decades.

The developed mathematical models were also evaluated to investigate their significance in estimating the surface area and the water volume of the HADR with respect to the water levels. Figure 9 shows a scatter diagram depicting the surface area estimated by the HADRDEM and from the developed mathematical model. The points are highly correlated and the relationship between them can be presented as a linear equation with a slope of one. Figure 10 shows a scatter diagram between the water volume estimated by the HADRDEM and the developed mathematical model. The points are highly correlated and the relationship between them can be presented as a linear equation with a slope of one.

To study the effect of eliminating the chosen khors, the four khors were isolated from the HADRDEM, for analysis of their hydrological characteristics individually. Using the isolated DEMs, average depth, surface area, and water volume were computed for each khor. Some alternatives were suggested to select the best location for eliminating each khor. The alternatives were analyzed and evaluated according to the cross section of the dam; the length of the dam; the saved surface area downstream the dam; and the water volume downstream of the dam.

According to these criteria, the most cost-effective and practical solution for each khor was selected. Each khor was removed from HADRDEM at the selected sections in order to study the impact of eliminating them on hydrological features of HADR individually. The new DEMs were again investigated to determine the new surface area and the new lake water volume for each elevation, and to build new mathematical models between them.

For the four case studies, the newly constructed mathematical models were used to estimate the saved eva-

poration losses at each water level under current climatic conditions. According to the MWRI prepared climatic scenarios, the ECHAM5 model predicts that due to climate change impact evaporation losses will increase by about 2.7% in 2030, 5% in 2050, and 10% in 2100, while HadCM3 expects higher evaporation by about 3% in 2030, 4.5% in 2050, and 8.5% in 2100 [14]. These rates are used in forecasting the expected reduced evaporation losses after eliminating each khor at each water level.