Comparative Analysis Between Meteorological Measurements of the Conventional and Automatic Station of Fazenda Lageado in the Municipality of Botucatu, São Paulo, Brazil

Abstract

The desable process of a conventional meteorological station (EMC) and replacing it with an automatic one (EMA) must be done after comparing concomitant measurements between both. Therefore, the objective of this study was to compare 2015 and 2016 measurements of the EMC and EMA of the School of Agricultural Sciences (UNESP) of Botucatu (SP) - Brazil to evaluate the if there is a difference in measurements and create a linear regression models for the construction of a single measurement serie between stations. Measures of maximum, minimum and average air temperature, average relative humidity, wind speed and precipitation were compared using the resampling (Bootstrap) and cross-validation (K-Fold) method, as well as the statistical indicators of index of concordance, mean bias error, relative mean bias error, root mean square error, relative root mean square error, correlation and determination coeficiente ($R^2$). The biggest difference was found in wind speed ($R^2$ = 0.746) and the smallest in precipitation ($R^2$ = 0.984). The results demonstrate that despite the difference in the composition of the instruments, there is agreement between the elements, making it safe to disable the EMC and allowing the estimation of the EMA measurements for the reconstruction of a single historical series.

Keywords
comparison; validation; homogenization