Planning agricultural activities requires that climatic risks in each location and time of year are estimated with maximum precision. In this estudy data from15 locations of Rio Grande do Sul, Brazil, were used with the objective of studing spatial and temporal variability of the precision estimates average of monthly rainfall, sunshine, solar radiation global and temperatures minimum, mean and maximum. It was estimated the estimation error (half-amplitude of the confidence interval) (EE) in percentage of estimate average monthly rainfall, sunshine, solar radiation global and temperatures minimum, mean and maximum for each month and location, which was used as a measure of precision. From the EE, for each meteorological elements, months and locations were clustered by hierarchical method of Ward. The Pearson's linear correlation coefficient between the arrays of the standardized average euclidian distance and graphics was calculated in order to check the clusters consistency. It was concluded that the precision estimates, average of months and locations, between the meteorological elements, increases in the following order: rainfall (EE=21.66%), sunshine (EE=6.74%), global solar radiation (EE=5.52%) and temperatures minimum (EE=4.40%), mean (EE=2.63%) and maximum (EE=2.27%). Results showed that there is variability in precision estimate of monthly average rainfall, sunshine, global solar radiation and minimum, mean and maximum temperatures in Rio Grande do Sul, both in temporal and spatial scale. Estimates based on 30 years data are less precision in April, May, June, July, August and September in relation to the others.
time series; sampling; error estimation; confidence interval; inferential statistics
