Deterministic and stochastic methods to analyse the spatial distribution of precipitation: The case of Mauritius, East Africa

Precipitation is one of the most important climatic parameters displaying significant changes across space and time. The accurate modeling of precipitation has become an important part of climate research for hydrological studies, the forecast of events such as droughts and floods and the estimation of ground and surface water resources. For this reason, several interpolation methods have been applied and compared for the accurate generation of models. In this study, the spatial distribution of annual mean total precipitation of Mauritius, located east of Africa, was investigated by applying deterministic methods, namely Thiessen Polygon (TP) and Inverse Distance Method (IDW), and stochastic methods, namely Ordinary Kriging (OK), using precipitation data from 53 meteorological stations for the period 1981–2010. The accuracy of the models was tested using the Cross Validation method and the models were compared using the Mean Error (ME), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the Coefficient of Determination (R²). The stochastic method, OK, provided the highest performance results, generating ME, MAE, RMSE and R² values of -17,66, 527,21, 329,53 mm and 0,88 respectively. In contrast, the deterministic method, Thiessen Polygon (TP),generated the lowest performance results, generating ME, MAE, RMSE, R²values of -78,83, 453,92, 621,58 mm and 0,60 respectively. Therefore, according to the results obtained, it can be concluded that stochastic methods provide more accurate models as compared to deterministic methods.