Time Instability Of Hydrological Model Parameters Under Changing Environment

In the background of substantial environmental change, “stable” watershed hypothesis is being challenged, resulting in time instability of hydrological model parameters that reflecting the catchment characteristics, therefore,it is necessary to carry out the research on time varying parameters. Previous studies usually adopted conceptual lumped models to qualitatively analyze the influence of climate change on time instability of model parameters, while lack of quantitative research. In this thesis, a theoretical framework to estimate the time instability of model parameters was proposed, and the impacts of model structures onparameter instabilitywere analyzed. Taking Hanjiang River basin as an example, the relationships between hydrological parameters of the physically based model and vegetation coverage were established to improve the hydrological model, which was used to investigate the impact factors of streamflow change and the spatial gradient of relative contributions of these factors.It was suggested tousethechanges of observable proxy indicators as climate and vegetation conditions to express the changes of unobservable hydrological characteristics, which are usually represented by model parameters.Based on the environmental indicators, the theoretical framework to estimate parameter instability was established. The rationality source of this framework is the self-organizaiton of catchment characteristics due to co-evolution of the hydrological system, which implies environmental characteristics are correlated.Based on the theoretical framework, quadratic regression models of NDVI and hydrological parameters of the physically based model were built in growing and non-growing seasons in Hanjiang River basin, and the correlation coefficients were 0.42 to 0.75. The regression models were used to estimate time varying parameters toimprove the hydrological model, and were verified by the simulations in validation stations and validation periods. The results showed that the hydrological model with time varying parameters could significantly improve the simulations of both high flows and low flows.Four hydrological models with different model structure were applied to investigate the influences of precipitation, potential evapotranspiration and NDVI on the parameters related toevaporation and flow generation processes. The results indicated that the stronger physical mechanism of the hydrological model, the weaker influence of climate change on varying parameters, and the stronger influence of vegetation change on varying parameters. Moreover, the comparative study in Hanjiang River basin and Chaohe River basin which have significantly different hydrological characteristics, proved the universality of this conclusion to some extent.Attributing analysis of the streamflow change based on the improved model with time varying parameters was carried out in Hanjiang River basin, and compared with three common methods. The results showed that the climate change, especially decreasing precipitation, was the main reason for the decreasing streamflow, and the vegetation change also played an important role. Compared to the original hydrological model with fixed parameters, themodel with time varying parameters could take into consideration of the direct influence of vegetation change on hydrological processes, as well as the indirect influence of vegetation change through the time instability of hydrological parameters, thus improved the evaluation of the contribution of vegetation change to the streamflow change. In addition, the spatial analysis of impact factors of streamflow change indicated that the influence of climate change got weaker from upstream to downstream, while that of the land cover change got stronger.