Towards numerical modelling of surface water temperature in a tropical lake in central-western México: Atmospheric forcing, validation and verification through satellite-based observations.

The atmospheric forcing from both the North American Regional Analysis (NARR) and National Solar Radiation Database (NSRDB) was compared statistically based on the Effect Size (ES) concept and was then used as an input to estimate the weekly Lake Surface Water Temperature (LSWT) of Lake Chapala, with two models: (i) a simple heat balance equation and (ii) the numerical model Delft3D. To validate and verify the numerical outputs, twenty-eight years of LSWT data from a composite of several satellites' sensors were used for comparison. In case (i) the application of NARR data resulted in an adequate estimation of LSWT compared to the composite, with a R2 of 0.81 and an ES of 0.06 (meaning a very small significant difference), whereas with NSRDB data resulted in an overestimation of the LSWT, despite a R2 of 0.87 and an ES of 0.67 (a moderate, nearly large, significant difference). However, a simple offset correction applied to the NSRDB data improved the temperature to an ES of 0.09 (a very small significant difference). In case (ii), an optimization procedure was required. Similarly, an air2water model was applied to verify the LSWT based only on the air temperature, and an acceptable R2 was obtained with the satellite composite of 0.73 with NARR and 0.78 with NSRDB data. Coupling satellite data and models is an excellent tool for analyzing physical processes in water bodies at different temporal and spatial scales and can improve operational observations of poorly monitored lakes and enhance knowledge of surface heat fluxes.