Multimodel uncertainty changes in simulated river flows induced by human impact parameterizations
Multimodel uncertainty changes in simulated river flows induced by human impact parameterizations
复制标题
DOI：
10.1088/1748-9326/aa5a3a
复制
发表时间：
2017-02
影响因子：
6.7
通讯作者：
Xingcai Liu;Q. Tang;H. Cui;M. Mu;D. Gerten;S. Gosling;Y. Masaki;Y. Satoh;Y. Wada
中科院分区：
文献类型：
--
作者：
Xingcai Liu;Q. Tang;H. Cui;M. Mu;D. Gerten;S. Gosling;Y. Masaki;Y. Satoh;Y. Wada
人类活动对全球水文循环的影响越来越大，甚至在某些地区主导着水文变化。水文学家试图通过在全球水文模型（GHMs）中参数化人为用水来确定人类影响引起的水文变化。由此增加的模型复杂性很可能在温室气体中引入额外的不确定性。在这里，使用四个ghm，根据1971-2000年两个实验模拟的平均河流流量的信噪比（SNR）对模式间不确定性进行了量化，其中有人类影响（VARSOC）和无人类影响（NOSOC）。这是对包含人类影响参数化的模式间不确定性的首次定量研究。结果表明，与NOSOC相比，VARSOC年流的模式间信噪比不确定性更大（全球约为2%，不同流域的信噪比大小不同），这在亚洲大部分地区和地中海北部地区尤为显著。盆地平均年流量的信噪比差异大多为负（- 20% ~ 5%，表明不确定性较高）。与NOSOC模拟相比，VARSOC高流量的不确定性略低，信噪比差异大多在- 20%到20%之间。两个试验的不确定度差异与流域灌溉面积的比例显著相关。由于人类影响的参数化，在VARSOC模拟中引入了大量额外的不确定性，因此迫切需要开发关于更好地理解人类影响的ghm。讨论了灌溉、水库调节和取水参数化的差异，以及未来GHM发展的潜在改进方向。我们还讨论了统计方法在减少模式间不确定性方面的优势，以及ghm校准的重要性，这不仅有助于提高历史模拟的性能，而且有助于在不断变化的环境下对水文变化进行更稳健和机密的未来预测。
Human impacts increasingly affect the global hydrological cycle and indeed dominate hydrological changes in some regions. Hydrologists have sought to identify the human-impact-induced hydrological variations via parameterizing anthropogenic water uses in global hydrological models (GHMs). The consequently increased model complexity is likely to introduce additional uncertainty among GHMs. Here, using four GHMs, between-model uncertainties are quantified in terms of the ratio of signal to noise (SNR) for average river flow during 1971–2000 simulated in two experiments, with representation of human impacts (VARSOC) and without (NOSOC). It is the first quantitative investigation of between-model uncertainty resulted from the inclusion of human impact parameterizations. Results show that the between-model uncertainties in terms of SNRs in the VARSOC annual flow are larger (about 2% for global and varied magnitude for different basins) than those in the NOSOC, which are particularly significant in most areas of Asia and northern areas to the Mediterranean Sea. The SNR differences are mostly negative (−20% to 5%, indicating higher uncertainty) for basin-averaged annual flow. The VARSOC high flow shows slightly lower uncertainties than NOSOC simulations, with SNR differences mostly ranging from −20% to 20%. The uncertainty differences between the two experiments are significantly related to the fraction of irrigation areas of basins. The large additional uncertainties in VARSOC simulations introduced by the inclusion of parameterizations of human impacts raise the urgent need of GHMs development regarding a better understanding of human impacts. Differences in the parameterizations of irrigation, reservoir regulation and water withdrawals are discussed towards potential directions of improvements for future GHM development. We also discuss the advantages of statistical approaches to reduce the between-model uncertainties, and the importance of calibration of GHMs for not only better performances of historical simulations but also more robust and confidential future projections of hydrological changes under a changing environment.