Attribution Analysis Of Evapotranspiration And Runoff Changes In China Based On The Budyko Hypothesis

Impacts of changes in climate and catchment properties on land surface hydrological processes is always a hot topic in the hydro-meteorological researches,especially over the regions with high population density and uneven distribution of water resources.In this study,we firstly used the observational hydro-meteorological dataset over 426 hydrological divisions(HDs)to analyze the spatiotemporal variations of evapotranspiration(ET)and streamflow(Q)during the 1980s(1981–1990)and 1990s(1991–2000)relative to the baseline(1961–1980).Then,contributions of precipitation(P),potential evapotranspiration(ET₀)and Budyko parameter n(referes to the land surface properties)to changes in ET and Q were identified.At last,the spatiotemporal differences of Budyko parameter n was quantitatively explained.The major results can be summarized as follows:(1)Compared with the baseline period,the 1980s ET in the mid-eastern Yangtze River Basin,and some HDs of the Pearl River Basin,and the Southeast Rivers obviously decreased,while the Q in this period showed evident decreases mainly in some HDs of the Pearl River Basin,particully for HDs in the southeast coastal basins,the northern Yangtze River Basin,and the Songhua River Basin with larger reductions.During 1990s,the HDs in the central Yellow River Bains,the southern Yangtze River Basin and the Pearl River Basin corresponded to obvious changes(increases or decreases)in ET.Regarding the 1990s Q,the evident reductions were found in the Huai River Bains and the northern Yangtze River Basin,but the larger increases mainly existed in the southeast coastal basins.(2)In 1980s and 1990s,more than 55%of the HDs showed the dominant factor of the Budyko parameter n,while there existed more than 35%and less than 10%of the HDs with the domiant factors of P and ET₀,respectively.For the dominant factors cotrolling ET changes,the Budyko parameter n(P)mainly happened in the southern(northern)basins of China,while the HDs with the dominant factor of ET₀ exhibited a sporadic distribution across China.Regarding for the domiant factors of the changed Q,these factors always showed a complex spatial distribution.Notably,some HDs experienced shifts in the domiant factors,and the shifits mainly occurred between Budyko parameter n and P.(3)In the Yellow River,the western Yangtze River and the Southwest River Bains,the Budyko parameter n was smaller.In 1980s and 1990s,the parameter n both increased in nearly70%of the HDs compared to the baseline period.At last,we utilized the multivariate adaptive regression spline(MARS)model to identify the major catchment property(e.g.,human activities,topography,soil property,vegetation,and climatological characterisitics)driving the spatiotemporal dynamics in Budyko parameter n(1990s relative to 1980s).It is found that the interaction terms between time-dynamic and time-statistic variables(could be refered to time-dynamic variables multiplied by time-statistic variables)should be responsible for spatial differences in the parameter n,while the time-dynamic variables and their interactions with time-statistic variables could explain the temporal changes in the Budyko parameter n over>50%of the HDs.