The objective of this work was to quantify the contribution of the auxiliary variable elevation in the estimate of the spatial distribution of annual average rainfall in the state of São Paulo, Brazil. The statistic mean square error (MSQR) was used in two observation sets for the mean annual rainfall (1957 to 1997): a full set, with 1,027 observations, and a reduced one, with 445. Areas of rainfall concentration were clearly defined in spatial variability maps that used the full data set, indicating possible microclimates. Ordinary kriging geostatistical interpolation had a performance 82 times more accurate than that of interpolation by the inverse square of the distance when the MSQR was used as a benchmark for the full set. For the reduced set, this magnitude was of two times. The estimation errors obtained by ordinary kriging were lower in the full set, whereas those obtained by ordinary cokriging were lower in the reduced one. This indicates that these interpolators should be used for determining the spatial distribution of annual average rainfall. The use of altitude as an auxiliary variable benefits the ordinary cokriging interpolator and defines more uniform micro regions as to the spatial distribution of annual average rainfall.
ordinary cokriging; geostatistics; ordinary kriging; interpolation methods; microclimate; climatic risk
