Forecasting austral summer precipitation along the western coast of South America (WCSA)

Abstract

Forecasting precipitation could help prevent flooding and drought disasters along the western coast of South America (WCSA), stretching from northern Peru to Ecuador. This study constructed a multiple linear regression (MLR) model to forecast precipitation anomalies with high spatial resolution across WCSA during the austral summer (December-January-February, DJF) for the period 1982–2023. The predictors of the MLR model are the central and eastern Pacific El Niño (C and E) and the central and east Pacific Intertropical Convergence Zone (CPITCZ and EPITCZ) indices. Furthermore, we readjusted the MLR model using forecasts from the Geophysical Fluid Dynamic Model (GFDL) model from the Seamless System for prediction and Earth System Research (SPEAR) called the GFDL-SPEAR (MLRGFDL-SPEAR) as predictors. The MLR model predicts DJF precipitation anomalies across WCSA because the E, CPITCZ and EPITCZ indices strongly correlate with DJF WCSA precipitation due to their influence on atmospheric circulation to trigger deep convection over far-eastern Pacific Ocean. The MLROBS model exhibits the highest performance over most WCSA (r > 0.6, p < 0.05), except along the coast of Ecuador and the Peru-Ecuador border by present high root mean square error values (above 20 mm month−1). The GFDL-SPEAR model provides more accurate forecasts of the DJF time series for the CPITCZ index than for the E and EPITCZ indices, due to Central Pacific ITCZ responses linearly to warm SST anomalies over western Pacific Ocean and it realistically simulates DJF precipitation patterns over Southern Pacific Ocean and Peru.