ABSTRACT

The high computational requirements of physically based fully distributed hydrological models (PDHM) constrain the use of all available general circulation models (GCMs) to assess climate change impacts. Here, an approach of ensembling GCMs using clustering based on future climatological variables was compared with model democracy while using a PDHM, SHETRAN, forced with six GCMs. The methodology is applied to hydrological projections in Netravathi River basin from the present to the near (2021–2050) and far (2071–2100) future. The results demonstrate that some GCMs project increase (50%, 30%) while others show decrease (10%, 11%) in the far future relative to the historical period (1980–2005) for streamflow and sediment load, respectively. The spread in the projection of climatological and hydrological variables from ensembled GCMs was retained as in model democracy whereas actual evapotranspiration showed overestimation relative to individual GCMs in the far future due to limitations of the clustering approach. Hence, we suggest employing individual GCMs for hydrological impact studies.

摘要

基于物理的完全分布式水文模型 (PDHM) 的高计算要求限制了使用所有可用的大气环流模型 (GCM) 来评估气候变化影响。在这里，将使用基于未来气候变量的聚类的 GCM 集成方法与模型民主进行比较，同时使用 PDHM、SHETRAN，强制使用六个 GCM。该方法适用于 Netravathi 河流域从现在到近期（2021-2050 年）和远期（2071-2100 年）的水文预测。结果表明，相对于历史时期（1980-2005 年），一些 GCM 项目的流量和泥沙负荷分别增加（50%、30%），而其他 GCM 项目在遥远的未来分别显示减少（10%、11%）。来自集合 GCM 的气候和水文变量预测的扩散与模型民主一样被保留，而由于聚类方法的限制，在遥远的将来，实际蒸散量相对于单个 GCM 显示出高估。因此，我们建议采用单独的 GCM 进行水文影响研究。