An Index for Predicting Precipitation in the North Coast of Peru Using Logistic Regression

Abstract

The north coast of Peru has a desert-like climate. Since precipitation is so scarce, strong convective events have a big impact on this region, especially when they are associated with an increase in sea surface temperature. More tools are needed to accurately forecast regions of storm development. This is why this study aims at developing a new index for predicting precipitation using Logistic Regression (LR). Satellite-based radar data will be used as a predictand while ERA5 reanalysis parameters will be used as predictors. The sensitivity of the LR results are assessed via factors such as, meteorological variables and precipitation and validation thresholds. After examining the values of the probability of detection, false alarm ratio and critical success index, and other validation indices. the best combination of predictors from the ERA5 parameters are mixing ratio at 700 hPa, divergence at 950 and 250 hPa, and the Galvez-Davison Index. This combination yields a logistic regression equation that ultimately takes the form of a new proposed index for the prediction of rainfall in the north coast of Peru called RAMI.