Optimal automatic enhanced ERA5 daily precipitation data for environmental and agricultural monitoring tools in scarce data regions

Abstract

The development of monitoring and early warning tools for environmental and agricultural applications is highly restricted in scarce climate data regions. In particular, precipitation data is a key input for several environmental monitoring tools on which decision-makers rely. However, precipitation records are collected by rain gauge stations, but these are frequently inhomogeneous and scarce in some regions of the world, especially in South America and Africa. In such cases, the use of alternative precipitation data sources is necessary to correctly assess its spatial and temporal variations. Therefore, we evaluate the possibility of using the ERA5 data with diferent automatic enhancement methods. Three adjustment approaches were evaluated in Northern Patagonia, which is an example of a scarce data area: (1) modifying the ERA5 daily data with three diferent regression models, one depending on lag and lead days, a distributed lag model, and a simple linear regression model, (2) detecting the lower time window of precipitation accumulation that can represent the observed precipitation variations, and (3) determining a window size and cut-of frequency of a low-pass flter to have data that represent well the low-frequency variation. The lag-distributed models improved the ERA5 data precipitation. A combination of approaches 1 and 2 showed the best performance for enhancing the ERA5 precipitation data, with a minimum of 6-day time window accumulation. However, this enhanced performance is not spatially homogeneous and it is poor in the northeastern region. This tool allows the use of data from ERA5 in sites where daily precipitation input data is scarce or inaccurate for diferent environmental and agricultural applications aimed at ofering permanent and updated information, such as monitoring drought, food, wildfre risk, or pest outbreaks. These applications are key to reducing ecosystem, production, and infrastructure loss in regions where climate data is a strong restriction