Present and future mean hydrologic trends in Serbia as a function of climate trends

Abstract

Past multiyear temperature (T), precipitation (P) and river discharge (Q) trends across Serbia are presented in the paper, both annual and monthly basis. The first objective of the research is to find observed multiyear T, P and Q trends in Serbia, which could be similar to the long-term trends and to assess the correlations between them. The results indicate that the long-term average yearly trends are approximately: temperature increase of 0.6°C/100 years, a slight decrease in precipitation, but with significant differences between western and eastern part of the country, and a decrease in river discharge of 30%/100 years. The second objective, and most important finding is the result of average correlations between air temperature increase and changes in river discharges and precipitation. The conclusion is that all the selected monitoring stations report an inversely proportional correlation between average annual temperatures and annual river discharges. On average, a 1°C increase in annual temperatures roughly corresponds to a 20% reduction in average annual river discharge and a 7% reduction in average annual precipitation. It is shown that an average annual temperature increase of 2°C will likely result in half the river discharge in Serbia, on average. The methodology described in the paper may be very useful for estimating near future (approx. next 30 years) average river discharges in many parts of the world, particularly in regions where a decreasing precipitation trend has been recorded. The third objective and important conclusion is related to low-discharge months (July through October). A considerably lower negative river discharge trend (close to zero) is noted, as a result of an upward precipitation trend during these months, but also in places due to human impact. The fourth objective is to generally compare the results of this research based only on observed changes, in which regional climate and hydrologic models (RCMs) were not used, with the results obtained by RCMs for the near future in different projects and studies by other authors.