Assessment of the spatial distribution of lake evaporation with 3d hydrodynamic modeling

Abstract

ln arid and semiarid environments, the evaporation rates are a main water loss. This condition is specially conceming in the Brazilian Northeast region, as this is one of the most populous semiarid areas in the planet. ln a reality of poor water availability, fully understanding the complexities involved in surface water losses along with the vertical process in a reservoir is of great importance. This study aims to calibrate the thermal structure of Pentecoste reservoir, located in the Brazilian serniarid region, and assess the variability of evaporation rates on its surface by using the tridimensional model Delft3D-FLOW. Additionally, the spatial distribution results obtained by the 3D model are compared with those estimated by a remote sensing technique (RS). The calibration results showed an accurate reproduction of the water level variability (r2 of 0.997), along with a satisfactory calibration of the reservoir's thermal structure for the full water colurnn (MAE of 0.539 ºC, RMSE of 0.572 ºC, and NMAE of 0.008). Curves relating monthly evaporation rates with air temperature and wind speed showed strong correlation between those variables (r2 of O. 734 for air temperature and O. 853 for wind speed). Also, the averaged evaporation rates modeled by Delft3D were less than 5% different from those estimated by the RS. Regarding the spatial distribution results, it was found that the drying and flooding criteria of Delft3D impacted the heat distribution in the water body, with the adjacent cells substantially retaining heat. This condition was found more relevant on the days with notably low water level. For the day with higher volume, the spatial distribution of evaporation rates was similar to that of RS, due to the occurrence of heat transfer. Regarding he thermal structure, the potential energy anomaly (PEA) results indicated stratification condition in the wet season (r2 between PEA and water depth of 0.33), which was probably induced not by the heating of the water surface, but by the cooling of the bottom layer caused by low temperature inflow. ln fact, evaporation rates showed negative correlation with PEA with a relatively low coefficient of correlation (r2 of 0.24), which could indicate altemating negative/positive correlation depending on the season. To conclude, this study brings an innovative approach which may be used to optimize the location for techniques that lower evaporation rates, such as air diffusers, shading structures, and floating solar panels, and therefore be a tool in the process of enhancing water availability for the population.