Skip to main content
ARS Home » Plains Area » El Reno, Oklahoma » Grazinglands Research Laboratory » Agroclimate and Natural Resources Research » Research » Publications at this Location » Publication #374641

Research Project: Uncertainty of Future Water Availability Due to Climate Change and Impacts on the Long Term Sustainability and Resilience of Agricultural Lands in the Southern Great Plains

Location: Agroclimate and Natural Resources Research

Title: Evaluation of statistical downscaling methods for simulating daily precipitation distribution, frequency, and temporal sequence

Author
item Zhang, Xunchang
item SHEN, MINGXI - Wuhan University
item CHEN, JIE - Wuhan University
item HOMAN, JOEL - Us Geological Survey (USGS)
item Busteed, Phillip

Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/26/2021
Publication Date: 3/30/2021
Citation: Zhang, X.J., Shen, M., Chen, J., Homan, J.W., Busteed, P.R. 2021. Evaluation of statistical downscaling methods for simulating daily precipitation distribution, frequency, and temporal sequence. Transactions of the ASABE. 64(3). https://doi.org/10.13031/trans.14097.
DOI: https://doi.org/10.13031/trans.14097

Interpretive Summary: Global Climate Models (GCM) project future climate change at a large spatial scale; however, plant growth computer models simulate crop production at field or farm scales, requiring climate data at that scale. This spatial mismatch is a major limitation to simulate the potential impact of climate change on crop production. Many spatial downscaling methods have been developed to bridge the spatial gap. It has been shown that there is not a universal downscaling method that performs the best for all climate regions. Thus, there is a need to select most suitable methods for particular climate regions. This study is evaluates nine different climate downscaling methods for selecting the most suitable methods for use in Oklahoma. Climate datasets were generally from 1940 to 2017, and were divided into calibration and validation periods. Performances of all methods were ranked based on the simulation error of each method, which was calculated by comparing the simulated climate with the observed climate of the validation period. The top four methods were Distribution-based Bias Correction (DBC), Generator for Point Climate Change (GPCC), SYNthetic weather generaTOR (SYNTOR), and LOCal Intensity scaling (LOCI). DBC and LOCI were two bias correction methods, and GPCC and SYNTOR were two weather generator-based methods. These results will be useful to meteorologist, climatologist, and modelers for selecting most suitable downscaling methods for use in Oklahoma.

Technical Abstract: Spatial mismatch between Global Climate Model (GCM) projection and input requirement of climate data by hydrological models is a major limitation for assessing the impact of climate change on soil erosion and crop production at local scales. Statistical downscaling techniques are widely used to bridge the gap while correcting biases of GCM projections. The objective of this study is to evaluate the ability of nine statistical downscaling methods in three downscaling categories to simulate daily precipitation distribution, frequency, and sequence at four Oklahoma stations representing arid to humid climate regions. The three downscaling categories include Perfect Prognosis (PP), Model Output Statistics (MOS), and Stochastic Weather Generator (SWG). To minimize the effect of GCM projection error on downscaling quality, the National Center for Environmental Prediction (NCEP) Reanalysis 1 data at a 2.5° grid resolution (treated as observed grid data) were downscaled to four stations (representing arid, semi-arid, semi-humid, and humid regions) using each downscaling method0. Datasets were generally from 1940 to 2017, and were divided into calibration and validation periods in a way that maximizes differences of annual precipitation means between the two periods for assessing the ability of each method in downscaling non-stationary climate changes. All methods were ranked with three metrics (Euclidean distance, sum of absolute relative error, and absolute error) for their ability in simulating precipitation distribution, frequency, and sequence at daily, monthly, yearly, and annual maximum scales. After eliminating the poorest two performers in simulating each aspect of precipitation mean, distribution, frequency, and sequence, the top four methods (in ascending order) were Distribution-based Bias Correction (DBC), Generator for Point Climate Change (GPCC), SYNthetic weather generaTOR (SYNTOR), and LOCal Intensity scaling (LOCI). DBC and LOCI were two bias correction methods, and GPCC and SYNTOR were two generator-based methods. The differences in performance among downscaling methods were smaller within each downscaling category than between the categories. Overall results indicated that weather generator-based methods had certain advantages in simulating daily precipitation distribution, frequency, and sequence for non-stationary climate changes.