Development of a partial copula-based algorithm for disclosing variability of dependence structures between hydro-meteorological factors under consideration of covariate-effect

Investigating variability of dependence structures between hydro-meteorological factors (DSHF) is of critical importance for throwing light on mechanisms of hydrological cycle under changing environment. Previous studies mostly focused on examining variability of single hydro-meteorological variable in forms of change point while neglecting complicated interactions among hydro-meteorological variables and thus failing to fully reveal variability of hydrological cycle. In this study, a partial copula-based likelihood-ratio (PaCoLR) test algorithm is developed for disclosing change point of DSHF whilst controlling for the effect of covariate. Eight major catchments across the Loess Plateau (LP), China, are selected as study regions. Results indicate that change points of dependence structures between streamflow and precipitation/potential evapotranspiration varies across the LP due to the spatial heterogeneity of climate and anthropogenic activities. Moreover, change points of dependence structure between precipitation and potential evapotranspiration on the LP match well with the time of global warming hypothesis, i.e. the early 1980s. Additionally, we find that controlling for the effect of precipitation is extremely essential when examining change point of dependence structure between streamflow and potential evapotranspiration on rain-dominated region, such as the LP. Further, the proposed PaCoLR algorithm is also applicable for detecting change points of DSHF in transient snow and snowmelt dominant regions.