Estimativa da radiação solar global para Viçosa (MG) com dados de entrada de fácil medição

Abstract

Global solar radiation is an important input variable in energy models and crop productivity. However, the recording of this variable is performed by instruments present in the automatic meteorological stations, which present a high maintenance and calibration cost, being necessary their estimation. The objective of this study was to adjust, evaluate and validate five models of estimation of global solar radiation for Viçosa, Minas Gerais. Three models use as input variable maximum and minimum temperature and one of them uses, in addition, the precipitation; while the other two models use the real insolation. For this, data were collected from the Meteorological Data Bank for Teaching and Research (BDMEP) and from the Integrated Environmental Data System (SINDA) platform, from 1999 to 2015. The adjustment and validation of the performance of each model were evaluated by means of the mean square error root (RQME), model trend (BIAS), correlation coefficient (r), Willmott concordance index (d), performance index (c) and weighted value (PV), significance of the coefficients and proximity between the observed and estimated values. In general, the models tended to overestimate radiation values below 10 MJ.m-2.dia-1 and underestimation to values higher than20 MJ.m-2.dia-1. The models that present the inclusion of the real insolation and the reason of the insolation were the ones that obtained better results, both in the adjustment and the validation.