New algebraic formulations of evapotranspiration extracted from gene-expression programming in the tropical seasonally dry regions of West Africa

Abstract

Within hydrological nonlinear complex functions, taking only few parameters into the modeling process is still a challenging task. The present paper has for objective to investigate for the first time the predictive ability of the Gene-expression Programming (GEP) for modeling reference evapotranspiration (ETo) using routing weather data from the tropical seasonally dry regions of West Africa in Burkina Faso. The regions under study are located in three agro-climatic zones, Bobo Dioulasso in the Guinea Savanna zone, and Dedougou and Fada N’Gourma in the Sudan zone, and Ouagadougou in the Sudano-Sahelian Savanna zone. Several meteorological data combinations are used as inputs to the GEP to estimate ETo, and their performances are evaluated using R2 and RMSE. Statistically, it was found that GEP can be an alternative to the conventional methods, and its accuracy improves significantly up to R2 (0.979) and RMSE(0.108) when critical variables are taking into account in the model. The results revealed that GEP model is fairly a promising approach with the advantage to provide successfully simple algebraic formulas ease to use without recourse to the full set of meteorological data requirement for accurately estimate ETo in Sub-Saharan Africa regions.