Rice yield variability estimates at different time scales of SPI index

Abstract

The objective of this work was to evaluate the accuracy of the drought index SPI (standardized precipitation index), at different time scales, for determining the drought occurrences and for estimate their effects on upland rice adjusted yield in six different microregions in Goiás state, Brazil. Precipitation daily data were used from rain gauge stations located at Uruaçu, Aruanã, Alexânia, Mineiros, Catalão and Itumbiara, in Goiás state, as well as upland rice yield data set for the microregions Porangatu, Rio Vermelho, Entorno de Brasília, Southwest Goiás, Catalão and Meia-Ponte, also located in Goiás state. The temporal time scales used in this study were monthly, quarterly, semiannual and annual. The differences among the SPI temporal scales were evaluated for the number and frequency of drought occurrences, the percentage accuracy of the variation estimate related to the upland rice adjusted yield data, and the deviations between these estimates and the observed yield. Increasing SPI temporal scale makes it possible the identification of persistent drought, and the annual scale (SPI-12) showed the best performance for estimating the variability of upland rice adjusted yield.