The skill of General Circulation Models (GCMs) in mimicking the observed climate is assessed through various procedures and are ranked. The performance of a GCM is site‐specific and the ranking pattern varies spatially. In general, a set of best performing GCMs is extracted to study the impact of climate change. As, there is no universally accepted ranking procedure, the ranking of GCMs is prone to subjectivity. In this study, it is aimed to address the effect of this subjectivity on the GCM rankings. The past performance of GCMs from Coupled Model Intercomparison Project phase 6 (CMIP6) in simulating the maximum and minimum temperature across India are evaluated and ranked by different ranking procedures. These ranking procedures involve combinations of various components present in the ranking procedure such as model evaluation criteria, criteria weightage allocation methods, Multicriteria Decision Making methods (MCDM) and reference gridded datasets. Different criteria and methods involved in the ranking procedures are carefully selected to address the subjectivity involved in ranking of GCMs. The effect of each individual component on the ranking pattern is systematically analysed and the spatial distribution of grids with same ranking patterns across all the combinations are considered as grids with same ranking. The performance of best performing GCM is attributed to homogenous climatic zones and its corresponding topological features. An ensemble of frequently performing top five GCMs among 16 different ranking procedures are extracted for each climate zone as the most suitable set of GCMs.
In the present study a total of 201 research articles from around the globe over the review period from 2001 to 2021 are identified and investigated. The study has identified some of the grand challenges associated with ensemble flood forecasting (EFF). In addition, future opportunities are identified including the development of quality-controlled datasets for longer duration, new techniques to weigh the ensembles, inclusion of more physically based datasets in data assimilation techniques, simulation of more observable states by hydrological models to facilitate data assimilation, application of artificial intelligence and machine learning, and optimization of computational efficiency to issue timely flood warnings. Moreover, it is worth mentioning that in addition to the technical aspects, the effectiveness of the operational flood forecasting system depends on the non-technical aspects.
10 April 2023 in Journal of Water and Climate Change
Climate change significantly impacts the natural systems, accelerating the global water cycle, and impacting various ecosystem services. However, the expected effects of climate change on the frequency and severity of extreme events on hydrological systems vary significantly with location. The present study investigates the uncertainties engulfed in hydrological predictions for the Tungabhadra River Basin. The ensemble streamflow projections were generated using four hydrological models, five climate models, and four climate scenarios to illustrate the associated uncertainties. The uncertainty in hydrological components such as streamflow (QQ), water availability (WA), and potential evapotranspiration (PET) was analysed in the future period (2015–2100). The results suggest that, in the monsoon period, precipitation projections increase by about 10.43–222.5%, whereas QQ projections show an increment between 34.50 and 377.7%. The analysis of variance (ANOVA) technique is used to further quantify the contribution of different sources to the total uncertainty. Furthermore, the ensemble spread is optimized using quantile regression forests (QRF), and the post-processed flows are likely to decrease up to 7% in June and increase up to 70% in September. This study is envisaged to give insights into the quantification of uncertainties in the prediction of future streamflow for rational and sustainable policymaking.