Abstract
Large ensembles of climate models are increasingly available either as ensembles of opportunity or perturbed physics ensembles, providing a wealth of additional data that is potentially useful for improving adaptation strategies to climate change. In this work, we propose a framework to evaluate the predictive capacity of 11 multi-model ensemble methods (MMEs), including random forest (RF), to estimate reference evapotranspiration (ET0) using 10 AR5 models for the scenarios RCP4.5 and RCP8.5. The study was carried out in the Segura Hydrographic Demarcation (SE of Spain), a typical Mediterranean semiarid area. ET0 was estimated in the historical scenario (1970–2000) using a spatially calibrated Hargreaves model. MMEs obtained better results than any individual model for reproducing daily ET0. In validation, RF resulted more accurate than other MMEs (Kling–Gupta efficiency (KGE) M=0.903, SD=0.034 for KGE and M=3.17, SD=2.97 for absolute percent bias). A statistically significant positive trend was observed along the 21st century for RCP8.5, but this trend stabilizes in the middle of the century for RCP4.5. The observed spatial pattern shows a larger ET0 increase in headwaters and a smaller increase in the coast.
Highlights
Concerns about anthropogenic Climate Change (CC) have increased in the last decades
We propose a framework centered on the use of Random Forest (RF) as a machine learning algorithm, both as a daily series ensemble technique, and as part of a Random Forest Regression Kriging (RK) interpolation model (RFRK)
The ANOVA analyses performed on the goodness-of-fit statistics of model ensemble methods (MMEs) and regionalized models show significant differences in the three analyzed statistics, both when each series is considered separately (F(26, 1269), p < 0.0001 in all three cases) and when the series are aggregated by type (F(4, 1291), p < 0.0001 in all three cases)
Summary
Concerns about anthropogenic Climate Change (CC) have increased in the last decades. It has been studied by many researchers and institutions, such as the Intergovernmental Panel on Climate Change (IPCC). The coupled ocean–atmosphere general circulation models (AOGCM) are the basic tool for making these projections. Their low spatial resolution (>100 km) prevents them from being used to study CC impact and adaptation at a regional scale. To solve this problem, regionalization or downscaling techniques, which can be statistical or dynamic, have been used. Regionalization or downscaling techniques, which can be statistical or dynamic, have been used These techniques adapt global projections to regional or local characteristics, which are highly influenced by orography, land–water contrast and land use, among other variables. More information and links to access these data are described in AEMET [7]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
