Estimation of monthly rainfall missing data in Southwestern Colombia: comparing different methods

ABSTRACT Historical rainfall records are relevant in hydrometeorological studies because they provide information on the spatial features, frequency, and amount of precipitated water a specific place, therefore, it is essential to make an adequate estimation missing data. This study evaluated four methods for estimating monthly data at 46-gauge stations southwestern Colombia covering 1983-2019. The performance Normal Ratio (NR), Principal Components Regression (PCR), Least Square (PLSR), Artificial Neural Networks (ANN) were compared using three standardized error metrics: Root Mean Error (RMSE), Percent BIAS (PBIAS), Absolute (MAE). results generally showed better nonlinear ANN method. Regarding linear methods, best was registered by PLSR, followed PCR. suggest applicability method regions with low density high percentage data, such as Colombia.