Regression‐based regionalization for bias correction of temperature and precipitation

Abstract

Statistical bias correction methods are inferred relationships between inputs and outputs. The constructed functions are based on available observations, which are limited in time and space. This study investigates the ability of regression models (linear and nonlinear) to regionalize a domain by defining a minimum number of training pixels necessary to achieve a good level of bias correction performance. Linear regression is used to divide northern South America into five regions. To correct the biases of temperature and precipitation, an artificial neural network (ANN) model was trained with selected pixels within each region and then used to reproduce bias‐corrected temperature and precipitation at all pixels within the delineated regions. The Community Climate System Model (CCSM) provided the climate model data. Results confirm that it is possible to identify regions in terms of physical features such as land cover, topography, and climatology over which models trained with a few pixels can correct the biases of climate variables with good accuracy over the entire domain. This approach saves computational time and reduces memory usage of using ANNs for correcting biases in climate model outputs.

A probabilistic climate change assessment for Europe

“…The spatial pattern of precipitation differences between the two distributions (baseline and projection) is not as smooth as for temperature. This is a reflection of the complex and highly nonlinear process of precipitation, which is consistent with the results obtained by Moghim and Bras (2017, 2019).…”

“…Similar to ECDF 1 , empirical CDFs for projection are constructed from 30 years (2020–2049) of monthly ensemble temperature and precipitation (ECDF Proj ). The KS measure has been used to show the performance skill of a model in a probability framework (Moghim and Bras, 2017, 2019). The second measure, Δ (see Figure 1b), can be interpreted as a change in intensities (value associated with a given percentile).…”

“…CRU product is one of the most common datasets that are widely used in hydro‐climatic studies (e.g., Moghim and Bras, 2017, 2019; Li et al ., 2010; Sheffield et al ., 2006, among many others). This gridded dataset is prepared by the School of Environmental Sciences and the Tyndall Centre at the University of the East Anglia, United Kingdom (data are publicly available at http://www.cru.uea.ac.uk/data).…”