Spatial and time correlation of thermometers and pluviometers in a weather network database

A basic issue that arises when analysing data bases from weather networks is the correlation system that characterizes the set of weather stations. Some statistical models being used for simulating temperature and precipitation or estimating missing data often exploit the Pearson’s correlation coefficient, whereby a selection of predictors is carried out. In this paper, a specific analysis was made to understand the relationship between the distances (between the stations) and the correlation structure (of the network) and to assess the evolution of the stations ranking over the time from the network establishment, given that they were ranked on the basis of their correlation coefficient values with a target station. This study was first carried out over the whole of the Veneto region in Northeast Italy, and subsequently, it was repeated, subdividing the area into three main climatic zones: mountain, plain and coast. The variables that are involved in this study are the following: daily precipitation and daily maximum, mean and minimum temperature. Generally, the correlation coefficients of the database of precipitation are, on average, inversely proportional to the mean distances from the target station. Considering that the same behaviour was not detected on analysing the temperature database, the main results of this work can be summarized as follows: (1) the most correlated stations of precipitation are generally closer to a target station than the most correlated stations of temperature (entire area); (2) starting from 5.5 years after the network was established, the temperature variable is characterized by a high stability (over time) of the correlation rankings, up to a wide radius from the target station; (3) this trend is not so clear in precipitation data. However, when taking into account the first result, (4) generally, the most correlated stations are placed within the radius of stability, more frequently so for precipitation than for temperature.