Comparison of weather measurements obtained on land surface and by Google Earth Engine (GEE) in Botucatu, São Paulo, Brazil

Abstract

The aim of this study was to compare weather data obtained from geospatial products in Google Earth Engine with measurements collected by an automatic weather station at the School of Agricultural Sciences of UNESP in Botucatu (SP) - Brazil. Scatter plots were created and the statistical indicators MBE, rMBE, RMSE, rRMSE and r were generated from data on air temperature, precipitation, evapotranspiration, wind speed, shortwave solar irradiation, and atmospheric pressure, obtained in 2018 at hourly, daily, and monthly temporal resolutions. The investigation pointed to a strong positive correlation in most of the weather data, however, those related to the amount of water present in the atmosphere, such as precipitation and evapotranspiration, showed a lower correlation, mainly in the hourly temporal resolution. The study demonstrated that geospatial products were an efficient alternative to obtain weather data for the city of Botucatu (SP) - Brazil, mainly because they were obtained in a simplified way from the Google Earth Engine cloud computing platform, which demonstrated be a possible alternative to traditional weather measurements, collected on the earth's surface, in areas where the necessary technological resources are not available to meet this demand.