Predictions of un-gauged sub-basins in Canadian prairies dominated by boreal forest

为预测流域径流而建立的水文模型大多是数据驱动的。这项研究将开发一套技术/工具来预测未测量流域(PUB)，这是由国际水文科学协会(IAHS)于2003年发起的一项倡议。可选择的研究地点包括区域尺度，加拿大草原上的阿萨巴斯卡河流域(ARB)，由北方森林(BF)主导，对陆地表面和大气之间的热量和水通量产生强烈影响。通过野外数据收集，修改土壤-植被-大气传输模型(SVAT)，如MISBA，我们可以科学地了解加拿大半干旱草原，特别是BF的气候、地形和对水文过程的控制。为了限制未测量盆地预报的不确定性，我们将尝试通过两个阶段的过程，例如GCM-区域尺度和区域-地方尺度，尽量减少从降尺度GCM尺度、欧洲中期天气预报中心(ECMWF)的ERA-40再分析资料和/或NASA的NCEP再分析资料获得的局地尺度气候输入的不确定性。我们将修改MISBA的模型结构以适应草原的水文和气候条件，并将另一个名为SWAP的SVAT中的BF冬季特征添加到MISBA中进行修改。最后，我们将在地理上将针对ARB内或附近的一些已测量的局部尺度子流域校准的模型参数转移到水文相似的未测量的子流域。我们希望通过一些随机方案调整模型参数，然后将它们转移到将对其性能进行定量评估的未测量的流域。这种技术的结合将提高我们对未测量的流域的预测能力，并为评估相对未知的河流流域的未来发展做好更好的准备。

Hydrologic models built for predicting streamflow of river basins are mostly data driven.  This study will develop a suite of techniques/tools to predict for un-gauged basins (PUB), an initiative started by the International Association of Hydrologic Sciences (IAHS) in 2003.  Study sites to be chosen are such as the regional scale, Athabasca River Basin (ARB) in the Canadian Prairies dominated by the Boreal Forest (BF) that exerts strong influence on the heat and water fluxes between land surface and the atmosphere.  Through field data collections, modifying soil-vegetation-atmosphere transfer models (SVAT) such as the MISBA, we can gain scientific understanding of climate, landscape and controls on hydrologic processes of the semi-arid Canadian Prairies, especially the BF.  To constrain uncertainties in un-gauged basin predictions, we will attempt to minimize uncertainty in local-scale climate input obtained from downscaling GCM scale, re-analysis data of ERA-40 of the European Centre for Mid-range Weather Forecasts (ECMWF) and/or NCEP re-analysis data of NASA, and regional scale observed data of say, ARB by a 2-stage process, e.g., GCM-regional scale and regional-local scale.  Next, we will modify MISBA's model structure to the Prairies' hydrologic and climatic conditions, and adding winter features of BF from another SVAT called SWAP to MISBA with modifications.  Lastly, we will geographically transfer model parameters calibrated for some gauged local-scale sub-basins located within or near ARB to the un-gauged sub-basin that are hydrologically similar.  We expect to adjust the model parameters through some stochastic schemes before transferring them to the un-gauged basin whose performance will be assessed quantitatively.  This combination of techniques will improve our ability to predict for un-gauged basins, and better prepared to assess future developments of relatively unknown river basins.