Temporal and spatial representativeness of the information of extreme temperatures in a network of meteorological stations in Guanajuato, Mexico

Abstract

The meteorological information generated by a traditional or automated station network is important in order to perform various economic activities: agriculture uses it in operative (real time), planning (with historical data), and research activities. When work is being performed in any of the previously mentioned activities and there is no station available, but there are several surrounding stations, the question to answer is which of these data stations should be used to get a representative data for our worksite. This research determined for the meteorological station network of Guanajuato (1) the minimal number of years required for this kind of studies, and (2) the temporal and spatial variability of the distance that generates the influence area from where the information of each of the gage stations considered in this research is representative. To accomplish the first goal, the stabilization of the determination coefficient (r2) and the mean square error (e2) through time was used. The r2 and e2 statistics were computed by using regressions for similar daily weather variables (maximum or minimum temperature) between two gage stations. To achieve the second goal, the distances between a base station and each of the stations considered were determined and its respective square root of mean square error of the estimated values were also computed. It was found that r2 and e2 did not stabilize in one year. The distances in each station show variations for each month, among stations, and for the two variables considered. Finally, it was concluded that in order to characterize the temperatures in the winter season, a greater number of stations are required than in any other season.