Correlation Analysis and Residual Error between Re-Analysis Data of the CFSR Model and Meteorological Stations

Abstract

The article describes a work of analysis of correlations between Re-Analysis Data (RAD) of the climate-forecasting model known worldwide as CFSR (Climate Forecast System Reanalysis), and those provided by local Meteorological Stations (MS), installed in the geographical area study. We also seek to quantify and characterize the residual error in order to validate the RAD as a source of alternative information to the data of the MS (DMS). The study area that encompasses the work corresponds to the northern region of Chile Tarapacá region, adjacent to border areas with Peru, Bolivia and Argentina. There is evidence that MS Data in the Tarapacá region present quality problems to calculate the Climatic Extremes Indices (CEI). For this reason, this work performs the indicated analysis, through the use of Data Mining (DM) techniques, which include the correlation coefficients of Pearson, Spearman and Kendall. In addition, descriptive statistics are used, considering temperature and precipitation data, provided by 82 MS for the period 1983–2012. The study concludes that there are severe residual errors, and it is not recommended to use the RAD as an alternative information source to calculate the CEI, since these can be affected. However, the RAD manage to reproduce the behavior of the meteorological variables, which is why they can be used for qualitative studies of these).