Ensuring continuous, good quality data from weather station networks requires a knowledge of spatial and temporal variability. This knowledge is essential to identify suspect data and provide estimates for data gaps. This study was conducted aiming to quantify and contrast the spatial and temporal variability for daily weather variables for a tropical climatic condition of the Instituto Agronômico (IAC) weather station network, located in the State of São Paulo, Brazil. For a period of 20 years (1981-2000) data were available from 19 weather stations. The daily meteorological variables studied were maximum and minimum air temperature and precipitation. In the spatial analysis, a central station was paired with each of the other stations in the area. The coefficient of variation (R²) and standard error of estimate (SEE) were calculated by regression of daily measurements between pairs of the same weather variables for various station within the area. The SEE and R² were plotted against linear distance from the central station. Best fit lines were determined for the variograms (R²) and errograms (SEE). Analysis were performed for each month. Generally, the R² decreased while the SEE increased with distance of separation between sites. A significant seasonal cycle was found in the SEE data for maximum and minimum air temperature and precipitation. Results indicated that the accuracy of estimated data and associated confidence limits varied with time of the year. As linear distance between sites increased, the SEE for maximum and minimum air temperature were greater, especially during the spring and winter seasons, respectively. The SEE data for precipitation were greater during the summer season. For 150 km of distance, the maximum air temperature presented SEE data values up to 3.0°C and 2.3°C during the spring and summer seasons, respectivelly, while precipitation presented SEE data values up to 4 and 15mm during winter and summer seasons, respectivelly. Based on the analysis of climate data from the tropical conditions of the State of São Paulo, spacing requirements varied with time of the year. An 80 km spacing is required to explain 90% of the variation between sites, for maximum daily air temperature during the spring and summer months, and 90 km for autumn and winter; while for minimum air temperature that spacing reduces to 55 km during the summer, 75 km for winter and spring, and increases to more than 90 km during the autumn. A 27 km spacing is required to explain the variation for precipitation during the winter season, up to 20 km for spring and autumn, but during the summer the spacing reduces to 12 km.
maximum and minimum air temperature; rainfall; spatial and temporal variability; quality control; weather station networks; estimation error; climate characterization
