Hybrid approach in statistical bias correction of projected precipitation for the frequency analysis of extreme events

Myoung Jin Um, Hanbeen Kim, Jun-Haeng Heo

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

A general circulation model (GCM) can be applied to project future climate factors, such as precipitation and atmospheric temperature, to study hydrological and environmental climate change. Although many improvements in GCMs have been proposed recently, projected climate data are still required to be corrected for the biases in generating data before applying the model to practical applications. In this study, a new hybrid process was proposed, and its ability to perform bias correction for the prediction of annual precipitation and annual daily maxima, was tested. The hybrid process in this study was based on quantile mapping with the gamma and generalized extreme value (GEV) distributions and a spline technique to correct the bias of projected daily precipitation. The observed and projected daily precipitation values from the selected stations were analyzed using three bias correction methods, namely, linear scaling, quantile mapping, and hybrid methods. The performances of these methods were analyzed to find the optimal method for prediction of annual precipitation and annual daily maxima. The linear scaling method yielded the best results for estimating the annual average precipitation, while the hybrid method was optimal for predicting the variation in annual precipitation. The hybrid method described the statistical characteristics of the annual maximum series (AMS) similarly to the observed data. In addition, this method demonstrated the lowest root mean squared error (RMSE) and the highest coefficient of determination (R2) for predicting the quantiles of the AMS for the extreme value analysis of precipitation.

Original languageEnglish
Pages (from-to)278-290
Number of pages13
JournalAdvances in Water Resources
Volume94
DOIs
Publication statusPublished - 2016 Aug 1

Fingerprint

frequency analysis
extreme event
general circulation model
method
climate
prediction
air temperature
climate change

All Science Journal Classification (ASJC) codes

  • Water Science and Technology

Cite this

@article{d9f6a6b710b4475f8495976dc09de74f,
title = "Hybrid approach in statistical bias correction of projected precipitation for the frequency analysis of extreme events",
abstract = "A general circulation model (GCM) can be applied to project future climate factors, such as precipitation and atmospheric temperature, to study hydrological and environmental climate change. Although many improvements in GCMs have been proposed recently, projected climate data are still required to be corrected for the biases in generating data before applying the model to practical applications. In this study, a new hybrid process was proposed, and its ability to perform bias correction for the prediction of annual precipitation and annual daily maxima, was tested. The hybrid process in this study was based on quantile mapping with the gamma and generalized extreme value (GEV) distributions and a spline technique to correct the bias of projected daily precipitation. The observed and projected daily precipitation values from the selected stations were analyzed using three bias correction methods, namely, linear scaling, quantile mapping, and hybrid methods. The performances of these methods were analyzed to find the optimal method for prediction of annual precipitation and annual daily maxima. The linear scaling method yielded the best results for estimating the annual average precipitation, while the hybrid method was optimal for predicting the variation in annual precipitation. The hybrid method described the statistical characteristics of the annual maximum series (AMS) similarly to the observed data. In addition, this method demonstrated the lowest root mean squared error (RMSE) and the highest coefficient of determination (R2) for predicting the quantiles of the AMS for the extreme value analysis of precipitation.",
author = "Um, {Myoung Jin} and Hanbeen Kim and Jun-Haeng Heo",
year = "2016",
month = "8",
day = "1",
doi = "10.1016/j.advwatres.2016.05.021",
language = "English",
volume = "94",
pages = "278--290",
journal = "Advances in Water Resources",
issn = "0309-1708",
publisher = "Elsevier Limited",

}

Hybrid approach in statistical bias correction of projected precipitation for the frequency analysis of extreme events. / Um, Myoung Jin; Kim, Hanbeen; Heo, Jun-Haeng.

In: Advances in Water Resources, Vol. 94, 01.08.2016, p. 278-290.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Hybrid approach in statistical bias correction of projected precipitation for the frequency analysis of extreme events

AU - Um, Myoung Jin

AU - Kim, Hanbeen

AU - Heo, Jun-Haeng

PY - 2016/8/1

Y1 - 2016/8/1

N2 - A general circulation model (GCM) can be applied to project future climate factors, such as precipitation and atmospheric temperature, to study hydrological and environmental climate change. Although many improvements in GCMs have been proposed recently, projected climate data are still required to be corrected for the biases in generating data before applying the model to practical applications. In this study, a new hybrid process was proposed, and its ability to perform bias correction for the prediction of annual precipitation and annual daily maxima, was tested. The hybrid process in this study was based on quantile mapping with the gamma and generalized extreme value (GEV) distributions and a spline technique to correct the bias of projected daily precipitation. The observed and projected daily precipitation values from the selected stations were analyzed using three bias correction methods, namely, linear scaling, quantile mapping, and hybrid methods. The performances of these methods were analyzed to find the optimal method for prediction of annual precipitation and annual daily maxima. The linear scaling method yielded the best results for estimating the annual average precipitation, while the hybrid method was optimal for predicting the variation in annual precipitation. The hybrid method described the statistical characteristics of the annual maximum series (AMS) similarly to the observed data. In addition, this method demonstrated the lowest root mean squared error (RMSE) and the highest coefficient of determination (R2) for predicting the quantiles of the AMS for the extreme value analysis of precipitation.

AB - A general circulation model (GCM) can be applied to project future climate factors, such as precipitation and atmospheric temperature, to study hydrological and environmental climate change. Although many improvements in GCMs have been proposed recently, projected climate data are still required to be corrected for the biases in generating data before applying the model to practical applications. In this study, a new hybrid process was proposed, and its ability to perform bias correction for the prediction of annual precipitation and annual daily maxima, was tested. The hybrid process in this study was based on quantile mapping with the gamma and generalized extreme value (GEV) distributions and a spline technique to correct the bias of projected daily precipitation. The observed and projected daily precipitation values from the selected stations were analyzed using three bias correction methods, namely, linear scaling, quantile mapping, and hybrid methods. The performances of these methods were analyzed to find the optimal method for prediction of annual precipitation and annual daily maxima. The linear scaling method yielded the best results for estimating the annual average precipitation, while the hybrid method was optimal for predicting the variation in annual precipitation. The hybrid method described the statistical characteristics of the annual maximum series (AMS) similarly to the observed data. In addition, this method demonstrated the lowest root mean squared error (RMSE) and the highest coefficient of determination (R2) for predicting the quantiles of the AMS for the extreme value analysis of precipitation.

UR - http://www.scopus.com/inward/record.url?scp=84976600577&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84976600577&partnerID=8YFLogxK

U2 - 10.1016/j.advwatres.2016.05.021

DO - 10.1016/j.advwatres.2016.05.021

M3 - Article

VL - 94

SP - 278

EP - 290

JO - Advances in Water Resources

JF - Advances in Water Resources

SN - 0309-1708

ER -