Comparison of baseline period choices for separating climate and land use/land cover change impacts on watershed hydrology using distributed hydrological models.

Separating impacts of land use/land cover change (LUCC) and climate change on hydrology is essential for watershed planning and management. This is typically done via hydrological modelling in combination with the one-factor-at-a-time analysis. However, it remains unclear how large the differences in isolated hydrological impacts would be when selecting different baseline periods. In this study, we compared baseline period choices for separating climate change and LUCC impacts on watershed hydrology in a typical inland river basin in northwest China, i.e. the Upper Heihe River Basin, with two hydrological models, i.e., Soil and Water Assessment Tool and Distributed Hydrology Soil Vegetation Model. In the real LUCC case which considers the actual land use changes between 2000 and 2011, the absolute contributions of LUCC to the variations in water yield and ET are slight and almost have the same magnitude for different baseline period choices, whereas those of climate change are substantial and with varying magnitudes. Compared with the absolute contributions, the relative contributions of climate change and LUCC seem less sensitive to the choices of baseline periods. In the hypothetical LUCC case which assumes an extreme land use conversion (i.e., grassland converts to farmland completely), both climate change and LUCC contribute to the changes in water yield and ET significantly. Moreover, both the absolute and relative contributions diverge noticeably between various baseline period choices. The influences of baseline period choices on the partitioning of hydrological impacts diverge significantly between different hydrological models. This study highlights that baseline period choice is an important source of uncertainty when disentangling the impacts of LUCC and climate change on hydrology. Some useful recommendations regarding baseline period selection have been proposed, which may help to reduce the uncertainties associated with baseline period choices.