Long-term trends in karst spring discharge and relation to climate factors and changes

Long karst spring discharge series were analyzed to determine trends, fluctuations and relationship to climate. The data came from two karst aquifers, Picentini and Matese massifs in southern Italy, and in some cases cover more than 100 years of records. The long-term hydrological series of these springs aid the understanding of water-resource availability variations under climate change, given the systematic discharge records and the absence of human-made alteration of the natural conditions in most of the spring catchments. The performances of five three-parameter probability distribution functions (generalized logistic, lognormal, Pearson type III, generalized extreme values and Weibull) for frequency analysis of the hydrological series were investigated, and evaluated also for the fit of extreme values of both tails of the distributions. The Weibull distribution was chosen in most cases. Then, each spring discharge value was transformed into the Z value of the standard normal distribution, which allows a statistical comparison between different spring discharge series and other climate variables. In particular, the relationship between the spring discharge, rainfall, North Atlantic Oscillation (NAO) and temperature series were investigated, also by linear correlations and trend detection tests. The analyses proved a general decrease of spring discharge over the last few decades. This decrease does not seem to depend only on precipitation change; an important role was played by the temperature increase. The discharge trends of these karst springs may provide a robust climate indicator of a wide area of southern Italy, as well as of a large sector of the Mediterranean area.