Links between different classes of storm tracks and the flood trends in Spain

The identification of storm tracks that generate the annual maximum floods and the quantification of their air-moisture content is proposed to understand better the atmospheric generation processes of floods in Spain, as well as their decreasing trends identified previously. In this work, the role of the atmospheric component on hydrological changes on Spain by using storm track data generated by the hybrid single particle Lagrangian integrated trajectory model (HYSPLIT) is examined. Storm tracks associated with annual maximum flood series from 14 streamflow gauges located across Spain are obtained by using NCEP/NCAR Reanalysis data. They are classified into five clusters through use of the K-means algorithm. It was also shown that the use of five clusters was able to reproduce the five major types of storms identified in Spain reported in the literature. The posterior grouping of the five associated storm types into two bigger groups (Oceanic and Continental storms), led to a coherent seasonal and spatial behaviour of hydrological regimes. For some gauges, it was observed that distinct flood statistics, such as mean and variance, differ significantly as a result of atmospherically distinct generation processes, suggesting that local annual maximum flood series may be non-homogeneous as a result of contrasting atmospheric generation processes. By means of logistic regression, it is estimated that the probability of occurrence of Oceanic storms reduced significantly in the Spanish Atlantic region along the studied period. Furthermore, the existence of low-frequency cycles introduces significant variation in the occurrence of storm types in the country, with them being much more frequent during the period 1975-1979, while Oceanic storms were more frequent during the 1983-1987 and 1996-1999 periods. Those storms classified as Continental are observed in 59.3% of the studied cases, while those termed Oceanic in 40.7%. Continental storms also contribute with more moisture in the studied cases, that is to say, 63.1% of total moisture content, while Oceanic storms contributed with 36.9%.