Evaluation of the TRMM 3B43 gridded precipitation estimates over Greece

Abstract

The goal of this study is the assessment of the reliability of the V7 TRMM (Tropical Rainfall Measuring Mission) 3B43 gridded precipitation estimates over Greece. The validation was carried out using gridded monthly precipitation data derived from 96 meteorological stations over the wider region of Greece. The TRMM 3B43 gridded estimates are on a calendar month temporal resolution and 0.25°×0.25° spatial resolution concerning the 11-year period 1998–2008. In order to compare directly the TRMM 3B43 products with the rain gauge data, the latter were interpolated at the same spatial resolution grid with that of TRMM 3B43 (0.25°×0.25°) and a new Greek gridded precipitation database (GGPDB) was constructed, using Kriging spatial interpolation method, taking into account the altitude of the meteorological stations. Further, to ensure reliable ground truth, we performed the comparison using data only over the 0.25° grid box when/where at least one reporting gauge is available. The annual and seasonal spatial distribution of the differences between the two datasets along with the spatial distribution of the correlation coefficients are presented and analyzed.The findings of the analysis revealed that the correlations between the two examined gridded databases (satellite and ground based) in winter were very high (r>0.90) for the entire Greek domain with small exceptions over sparse data continental mountainous regions and restricted areas over north and south Aegean Sea. Regarding the other seasons, small correlations occur in the northern high altitude areas (Olympus Mt.) and the central/south mountainous Peloponnese. Besides, it is worth noting the remarkable small correlations over the eastern Aegean Sea and the Cyclades complex, in summer and the Cyclades complex in autumn. The TRMM 3B43 gridded data overestimate the precipitation over the Aegean and Ionian Sea, the coastal areas of Asia Minor and western Greece, in winter (>50mm), spring (>25mm) and autumn (>60mm). On the contrary, underestimation is demonstrated on central continental highlands (Pindus mountain range) and central mountainous Peloponnese in winter (>150mm), spring (25–100mm) and autumn (60–90mm). Summer is generally characterized by small irregularities. Generally, the findings of the performed validation indicate an underestimation at higher precipitation appeared over high altitude areas against some overestimation at lower precipitation over plains and coastal regions. The TRMM 3B43 products performance was assessed by using three contingency statistics for probability of detection (POD), false alarm ratio (FAR) and critical success index (CSI). TRMM 3B43 has POD>80% and CSI>60% when the monthly precipitation is lower than 84.0mm. Moreover, the smallest POD of 44% and CSI of 34% appear when the monthly precipitation is greater than 191.5mm. This result implies that satellite products have poor performance in estimating high monthly precipitation totals, over the Greek region.