Functional data analysis of models for predicting temperature and precipitation under climate change scenarios

Abstract

Abstract Evaluating the impact of climatic change on hydrologic variables is highly important for sustainability of water resources. Precipitation and temperature are the two basic parameters which need to be included in climate change impact studies. Thirty years (1985–2015) climatic data of Astore, a sub-catchment of the Upper Indus River Basin (UIRB), were analyzed for predicting the temperature and precipitation under different climate change scenarios. The station data were compared with the results of two global climate models (GCMs) each with two emission scenarios, including Representative Concentration Pathway (RCP) 2.6 and 8.5. The Mann–Kendall test and Sen's slope were applied to explore various properties of precipitation and temperature data series for a trend analysis. The commonalities and dissimilarities between the results of various GCMs and the trend of the station data were investigated using the functional data analysis. Two cross distances were estimated on the basis of Euclidean distances between the predicted time series; subsequently, the differences in their first derivatives were used to evaluate their mutual dissimilarities. The long-term predictions by GCMs show a decreasing trend in precipitation and a slight increase in temperature in some seasons. The result of GCMs under both the emission scenarios showed almost the same pattern of changes in the two hydrologic variables throughout the century with their values reporting slightly higher for the RCP8.5 scenario as compared to those for RCP2.6. Validation of the GCM results using GCM-CSIRO-Mk3.6 revealed an overall agreement between the different models. The dissimilarity analysis manifested the difference between the results of temperature predicted by various GCMs.