Statistical downscaling rainfall using artificial neural network: significantly wetter Bangkok?

• Published in: Theoretical and applied climatology Vol. 126; no. 3-4; pp. 453 - 467
• Main Authors: Vu, Minh Tue, Aribarg, Thannob, Supratid, Siriporn, Raghavan, Srivatsan V, Liong, Shie-Yui
• Format: Journal Article
• Language: English
• Published: Vienna Springer Vienna 01.11.2016; Springer; Springer Nature B.V
• Subjects: Analysis; Aquatic Pollution; Atmospheric models; Atmospheric Protection/Air Quality Control/Air Pollution; Atmospheric Sciences; Climate change; Climate models; Climatology; Correlation coefficient; Earth and Environmental Science; Earth Sciences; Global climate; Neural networks; Original Paper; Principal components analysis; Rain and rainfall; Rainfall measurement; Rainy season; Statistical analysis; Waste Water Technology; Water Management; Water Pollution Control; Thailand; Bangkok Thailand; Thailand, Chacoengsao Prov., Bangkok; Artificial Neural Network Model; Statistical Downscaling; Artificial Neural Network; Southwest Monsoon; Reanalysis Data
• ISSN: 0177-798X; 1434-4483
• DOI: 10.1007/s00704-015-1580-1

Artificial neural network (ANN) is an established technique with a flexible mathematical structure that is capable of identifying complex nonlinear relationships between input and output data. The present study utilizes ANN as a method of statistically downscaling global climate models (GCMs) during the rainy season at meteorological site locations in Bangkok, Thailand. The study illustrates the applications of the feed forward back propagation using large-scale predictor variables derived from both the ERA-Interim reanalyses data and present day/future GCM data. The predictors are first selected over different grid boxes surrounding Bangkok region and then screened by using principal component analysis (PCA) to filter the best correlated predictors for ANN training. The reanalyses downscaled results of the present day climate show good agreement against station precipitation with a correlation coefficient of 0.8 and a Nash-Sutcliffe efficiency of 0.65. The final downscaled results for four GCMs show an increasing trend of precipitation for rainy season over Bangkok by the end of the twenty-first century. The extreme values of precipitation determined using statistical indices show strong increases of wetness. These findings will be useful for policy makers in pondering adaptation measures due to flooding such as whether the current drainage network system is sufficient to meet the changing climate and to plan for a range of related adaptation/mitigation measures.