Using multi‐model ensembles of CMIP5 global climate models to reproduce observed monthly rainfall and temperature with machine learning methods in Australia

Abstract

Global climate models (GCMs) are useful tools for assessing climate change impacts on temperature and rainfall. Although climate data from various GCMs have been increasingly used in climate change impact studies, GCMs configurations and module characteristics vary from one to another. Therefore, it is crucial to assess different GCMs to confirm the extent to which they can reproduce the observed temperature and rainfall. Rather than assessing the interdependence of each GCM, the purpose of this study is to compare the capacity of four different multi-model ensemble (MME) methods (random forest [RF], support vector machine [SVM], Bayesian model averaging [BMA] and the arithmetic ensemble mean [EM]) in reproducing observed monthly rainfall and temperature. Of these four methods, the RF and SVM demonstrated a significant improvement over EM and BMA in terms of performance criteria. The relative importance of each GCM based on the RF ensemble in reproducing rainfall and temperature could also be ranked. We compared the GCMs importance and Taylor skill score and found that their correlation was 0.95 for temperature and 0.54 for rainfall. Our results also demonstrated that the number of GCMs ensemble simulations could be reduced from 33 to 25 in RF model while maintaining predictive error less than 2%. Having such a representative subset of simulations could reduce computational costs for climate impact modelling and maintain the quality of ensemble at the same time. We conclude that machine learning MME could be efficient and useful with improved accuracy in reproducing historical climate variables.

Selection of multi-model ensemble of general circulation models for the simulation of precipitation and maximum and minimum temperature based on spatial assessment metrics

“…Ensemble simulation is estimated by averaging the predictions of all trees, which is considered as the final simulation. Wang et al (2018) and reported that the performance of RF varies with the number of trees (n tree ) and the number of variables randomly sampled (m try ) at each split in developing the trees. In those studies, it was observed that RF performance increases with the increase in the value of ntree.…”

“…Salman et al (2018a) generated the MME mean for maximum and minimum temperature over Iraq using four CMIP5 GCMs and reported that RF-based MME performed better compared to individual GCMs. Likewise, Wang et al (2018) conducted a comprehensive study to evaluate the performance of different machine learning techniques including RF, a support vector machine, Bayesian model averaging, and the arithmetic ensemble mean in generating MMEs. They considered 33 CMIP5 GCMs for precipitation and temperature over 108 stations located in Australia and concluded that RF and SVM can generate better-performing MMEs compared to other techniques.…”

“…The methods used for the generation of MMEs are broadly divided into two groups, (1) simple composite method (SCM) and (2) weighted ensemble method (WEM) (Wang et al, 2018). In SCM all ensemble members are equally weighted, while in the WEM ensemble members are weighted according to their performance in simulating the past climate (Wang et al, 2018;Oh and Suh, 2017;Giorgi and Mearns, 2002). The SCM is relatively simple to apply and found to perform better than individual GCMs (Weigel et al, 2010;Acharya et al, 2013;Wang et al, 2018).…”

“…In SCM all ensemble members are equally weighted, while in the WEM ensemble members are weighted according to their performance in simulating the past climate (Wang et al, 2018;Oh and Suh, 2017;Giorgi and Mearns, 2002). The SCM is relatively simple to apply and found to perform better than individual GCMs (Weigel et al, 2010;Acharya et al, 2013;Wang et al, 2018). However, WEM is preferred as it can remove the systematic biases and improve the prediction capability since higher weights are assigned to better GCMs (Krishnamurti et al, 1999(Krishnamurti et al, , 2000.…”

Selection of multi-model ensemble of GCMs for the simulation of precipitation based on spatial assessment metrics

“…generated MME mean for maximum and minimum temperature over Iraq using four CMIP5 GCMs and reported RF performed better compared to individual GCMs. Likewise,Wang et al (2017) conducted a…”