Comparison of statistical and dynamical downscaling results from the WRF model. (February 2018)

• Record Type: Journal Article
• Authors: Le Roux, Renan; Katurji, Marwan; Zawar-Reza, Peyman; Quénol, Hervé; Sturman, Andrew
• Is Part Of: Environmental modelling & software. Volume 100(2018)
• Page Start: 67
• Page End: 73
• Publication Date: 2018-02
• Subjects: Environmental monitoring -- Computer programs -- Periodicals; Ecology -- Computer simulation -- Periodicals; Digital computer simulation -- Periodicals; Computer software -- Periodicals; Environmental Monitoring -- Periodicals; Computer Simulation -- Periodicals; Environnement -- Surveillance -- Logiciels -- Périodiques; Écologie -- Simulation, Méthodes de -- Périodiques; Simulation par ordinateur -- Périodiques; Logiciels -- Périodiques; Computer software; Digital computer simulation; Ecology -- Computer simulation; Environmental monitoring -- Computer programs; Periodicals; Electronic journals; 363.70015118
• DOI: 10.1016/j.envsoft.2017.11.002
• Languages: English
• ISSNs: 1364-8152

Abstract

Abstract: This study compares two methods of downscaling the Weather Research and Forecasting model output temperatures to 1 km resolution over the largest vineyard area in New Zealand. The WRF dynamical downscaling is obtained via a four-level nested grid configuration to create a 1-km grid. The statistical downscaling is achieved using a Support Vector Regression (SVR) between WRF 3-km output temperatures and terrain at 1 km resolution. The bias of the two approaches is evaluated using automatic weather stations, and the averages of both 1-km and 3-km model output are associated with a cold bias. The sensitivity of the methods to the input sample size is assessed using statistical indicators. The results demonstrate that for an equivalent sample size, there is no need to dynamically downscale the model temperatures from 3 to 1 km, as statistical downscaling seems to provide results very close to those of dynamical downscaling, while requiring less computer resources. Highlights: Statistical, dynamic downscaling of 2 m air temperatures. Statistical downscaling took 1% of the execution time. Tests with independent measures suggest robust performance.