Determinação de regiões homogêneas de precipitação e temperatura no Estado do Rio Grande do Sul.

Abstract

Homogeneous subregions were determined from the climatological point of view of temperature and precipitation in the state of Rio Grande do Sul. Objective techniques of Multivariate Analysis: Cluster Analysis (TAA) and Principal Component Analysis (ACP) were used. The available data were monthly and decendial climatological averages for 41 stations, provided by the Rio Grande do Sul Agroclimatic Atlas - IPAGRO. The variables were analyzed separately. The TAA for temperature presented five groups (monthly basis) and four groups (decendial basis), delimiting regions according to isotherms (and, indirectly, the relief of the State). The ACP highlights the presence of a single predominant regime throughout the region. It is concluded that the temperature is homogeneous in the State and that the groups delimited by the TAA are coherent. The difference in behavior between two sites is basically determined by their means and standard deviations, making the use of PCA irrelevant for this variable. For precipitation five regionalization criteria were considered using both TAA and ACP and combinations of both. The criterion that divides the state into eight groups based on the components considered most significant was considered the most appropriate, mainly for approximation and simulation of data series. This grouping delimits regions characterized by a combination of relief and maritime - continentality effects. The ACP and the TAA cannot clearly separate at the climate level the influence of the various weather systems acting on time and space in the region. For this, it is recommended to analyze shorter and more detailed time series. New precipitation auto-vectors specific for the eight groups were evaluated. The simulation of decendial precipitation series based on these new auto vectors (method developed by CEBALLOS & BRAGA, 1991) showed good results, requiring, in general, no more than two main components to obtain average deviations of less than 30% of the annual standard deviation of. each group.