Triple Collocation of Summer Precipitation Retrievals from SEVIRI over Europe with Gridded Rain Gauge and Weather Radar Data

Abstract

uantitative information on the spatial and temporal error structures in large-scale (regional or global) precipitation datasets is essential for hydrologic and climatic studies. A powerful tool to quantify error structures in large-scale datasets is triple collocation. In this paper, triple collocation is used to determine the spatial and temporal error characteristics of three precipitation datasets over Europe?that is, the precipitation-properties visible/near infrared (PP-VNIR) retrievals from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) instrument on board Meteosat Second Generation (MSG), weather radar observations from the European integrated weather radar system, and gridded rain gauge observations from the datasets of the Global Precipitation Climatology Centre (GPCC) and the European Climate Assessment and Dataset (ECA&D) project. For these datasets the spatial and temporal error characteristics are evaluated and their performance is discussed. Finally, weather radar and PP-VNIR retrievals are used to evaluate the diurnal cycles of precipitation occurrence and intensity during daylight hours for different European climate regions. The results suggest that the triple collocation method provides realistic error estimates. The spatial and temporal error structures agree with the findings of earlier studies and reveal the strengths and weaknesses of the datasets, such as inhomogeneity of weather radar practices across Europe, the effect of sampling density in the gridded rain gauge dataset, and the sensitivity to retrieval assumptions in the PP-VNIR dataset. This study can help us in developing satisfactory strategies for combining various precipitation datasets?for example, for improved monitoring of diurnal variations or for detecting temporal trends in precipitation.