The Influence of Meteorological Variables on Energy Demand in the Federal District of Brazil

The effects of climate change are present in all segments of society in general, and especially in the energy segment. Brazil plays a leading role in the use of renewable energy, with the majority of its matrix coming from renewable sources. In this sense, evaluating the impact of meteorological variables on the injected energy load is fundamental for the efficient use of energy. The present study aims to analyze the influences of meteorological variables on the energy load demand in Brasília - Federal District, Brazil. We analyzed observed data from the National Institute of Meteorology (INMET) weather station, reanalysis data from the National Center for Environmental Prediction (NCEP), reanalysis from the European Center for Medium-Range Weather Forecast (ECMWF), Modern-Era Retrospective Analysis for Research and Application Aerosol Reanalysis (MERRA-2) and South American Mapping of Temperature (SAMeT). Pearson's correlation coefficient was used to quantify the linear relationship between the observed data from the meteorological station (INMET) and the monthly load injected in Brasília in the period 2016-2022. Subsequently, statistical metrics, commonly used for model checking, were applied to regularly spaced global numerical model datasets with assimilation: CFSR (NCEP), ERA5 (ECMWF), MERRA2 (NASA), and SAMET (INPE). A high direct correlation of the injected monthly load with the monthly averages of maximum and average temperatures (0.65 and 0.51), respectively, and an inverse correlation with the observed average relative humidity (-0.50) was noted. Furthermore, the representativeness of temperatures from the data sets was investigated, aiming to expand the analysis to other regions that do not have meteorological station data. In validating the maximum and average temperature, it was possible to identify a high representative potential of the sets covered. Highlights include SAMET and ERA5, which presented the highest correlation coefficients (higher than 0.90) and standard deviation proportional to observational data.