Uncertainty in hydrological change modelling

Lauren Paige Seaby

Uncertainty in hydrological change modelling

Lauren Paige Seaby

Abstract

Hydrological change modelling methodologies generally use climate models outputs to force hydrological simulations under changed conditions. There are nested sources of uncertainty throughout this methodology, including choice of climate model and subsequent bias correction methods. This Ph.D. study evaluates the uncertainty of the impact of climate change in hydrological simulations given multiple climate models and bias correction methods of varying complexity. Three distribution based scaling methods (DBS) were developed and benchmarked against a more simplistic and commonly used delta change (DC) approach. These climate model projections were then used to force hydrological simulations under climate change for the island Sjælland in Denmark to analyse the contribution of different climate models and bias correction methods to overall uncertainty in the hydrological change modelling methodology for basin discharge and groundwater heads. The ensemble of 11 climate models varied in strength, significance, and sometimes in direction of the climate change signal. The more complex daily DBS correction methods were more accurate at transferring precipitation changes in mean as well as the variance, and improving the characterisation of day to day variation as well as heavy events. However, the most highly parameterised of the DBS methods were less robust under climate change conditions. The spatial characteristics of groundwater head and stream discharge were best represented by DBS methods applied at the grid scale. Flux and state hydrological outputs which integrate responses over time and space showed more sensitivity to precipitation mean spatial biases and less so on extremes. In the investigated catchments, the projected change of groundwater levels and basin discharge between current and future conditions was found to be modest across multiple climate models and bias correction methods. The choice of RCM contributes almost all of the uncertainty across precipitation inputs, explaining 99% of uncertainty in total annual precipitation, while choice of bias correction method contributes far less to overall uncertainty, contributing to 7% of variance in total annual precipitation values. Similarly choice of climate model contributes to almost all of the variance in stream discharge, 97% in total annual, while choice of bias correction method contributes to 10% of uncertainty in total annual discharge.

Originalsprog	Engelsk

Forlag	Department of Geosciences and Natural Resource Management, Faculty of Science, University of Copenhagen
Antal sider	184
Status	Udgivet - 22 okt. 2013

Navn	Danmarks og Grønlands geologiske undersøgelse
Nummer	58
Vol/bind	2013

Adgang til dokumentet

Thesis.LaurenAccepteret manuskript, 6,5 MB

https://rex.kb.dk/primo-explore/fulldisplay?docid=KGL01008787534&context=L&vid=NUI&search_scope=KGL&tab=default_tab&lang=da_DK

Andre filer og links

Sign in to request a library copy

Citationsformater

@phdthesis{8290017239654ab3b7d4ba9f13cebee1,

title = "Uncertainty in hydrological change modelling",

abstract = "Hydrological change modelling methodologies generally use climate models outputs to force hydrological simulations under changed conditions. There are nested sources of uncertainty throughout this methodology, including choice of climate model and subsequent bias correction methods. This Ph.D. study evaluates the uncertainty of the impact of climate change in hydrological simulations given multiple climate models and bias correction methods of varying complexity. Three distribution based scaling methods (DBS) were developed and benchmarked against a more simplistic and commonly used delta change (DC) approach. These climate model projections were then used to force hydrological simulations under climate change for the island Sj{\ae}lland in Denmark to analyse the contribution of different climate models and bias correction methods to overall uncertainty in the hydrological change modelling methodology for basin discharge and groundwater heads. The ensemble of 11 climate models varied in strength, significance, and sometimes in direction of the climate change signal. The more complex daily DBS correction methods were more accurate at transferring precipitation changes in mean as well as the variance, and improving the characterisation of day to day variation as well as heavy events. However, the most highly parameterised of the DBS methods were less robust under climate change conditions. The spatial characteristics of groundwater head and stream discharge were best represented by DBS methods applied at the grid scale. Flux and state hydrological outputs which integrate responses over time and space showed more sensitivity to precipitation mean spatial biases and less so on extremes. In the investigated catchments, the projected change of groundwater levels and basin discharge between current and future conditions was found to be modest across multiple climate models and bias correction methods. The choice of RCM contributes almost all of the uncertainty across precipitation inputs, explaining 99% of uncertainty in total annual precipitation, while choice of bias correction method contributes far less to overall uncertainty, contributing to 7% of variance in total annual precipitation values. Similarly choice of climate model contributes to almost all of the variance in stream discharge, 97% in total annual, while choice of bias correction method contributes to 10% of uncertainty in total annual discharge. ",

author = "Seaby, {Lauren Paige}",

note = "Mundtligt ph.d.-forsvar blev afholdt 22. november 2013 p{\aa} IGN",

year = "2013",

month = oct,

day = "22",

language = "English",

series = "Danmarks og Gr{\o}nlands geologiske unders{\o}gelse",

number = "58",

publisher = "Department of Geosciences and Natural Resource Management, Faculty of Science, University of Copenhagen",

}

TY - BOOK

T1 - Uncertainty in hydrological change modelling

AU - Seaby, Lauren Paige

N1 - Mundtligt ph.d.-forsvar blev afholdt 22. november 2013 på IGN

PY - 2013/10/22

Y1 - 2013/10/22

N2 - Hydrological change modelling methodologies generally use climate models outputs to force hydrological simulations under changed conditions. There are nested sources of uncertainty throughout this methodology, including choice of climate model and subsequent bias correction methods. This Ph.D. study evaluates the uncertainty of the impact of climate change in hydrological simulations given multiple climate models and bias correction methods of varying complexity. Three distribution based scaling methods (DBS) were developed and benchmarked against a more simplistic and commonly used delta change (DC) approach. These climate model projections were then used to force hydrological simulations under climate change for the island Sjælland in Denmark to analyse the contribution of different climate models and bias correction methods to overall uncertainty in the hydrological change modelling methodology for basin discharge and groundwater heads. The ensemble of 11 climate models varied in strength, significance, and sometimes in direction of the climate change signal. The more complex daily DBS correction methods were more accurate at transferring precipitation changes in mean as well as the variance, and improving the characterisation of day to day variation as well as heavy events. However, the most highly parameterised of the DBS methods were less robust under climate change conditions. The spatial characteristics of groundwater head and stream discharge were best represented by DBS methods applied at the grid scale. Flux and state hydrological outputs which integrate responses over time and space showed more sensitivity to precipitation mean spatial biases and less so on extremes. In the investigated catchments, the projected change of groundwater levels and basin discharge between current and future conditions was found to be modest across multiple climate models and bias correction methods. The choice of RCM contributes almost all of the uncertainty across precipitation inputs, explaining 99% of uncertainty in total annual precipitation, while choice of bias correction method contributes far less to overall uncertainty, contributing to 7% of variance in total annual precipitation values. Similarly choice of climate model contributes to almost all of the variance in stream discharge, 97% in total annual, while choice of bias correction method contributes to 10% of uncertainty in total annual discharge.

AB - Hydrological change modelling methodologies generally use climate models outputs to force hydrological simulations under changed conditions. There are nested sources of uncertainty throughout this methodology, including choice of climate model and subsequent bias correction methods. This Ph.D. study evaluates the uncertainty of the impact of climate change in hydrological simulations given multiple climate models and bias correction methods of varying complexity. Three distribution based scaling methods (DBS) were developed and benchmarked against a more simplistic and commonly used delta change (DC) approach. These climate model projections were then used to force hydrological simulations under climate change for the island Sjælland in Denmark to analyse the contribution of different climate models and bias correction methods to overall uncertainty in the hydrological change modelling methodology for basin discharge and groundwater heads. The ensemble of 11 climate models varied in strength, significance, and sometimes in direction of the climate change signal. The more complex daily DBS correction methods were more accurate at transferring precipitation changes in mean as well as the variance, and improving the characterisation of day to day variation as well as heavy events. However, the most highly parameterised of the DBS methods were less robust under climate change conditions. The spatial characteristics of groundwater head and stream discharge were best represented by DBS methods applied at the grid scale. Flux and state hydrological outputs which integrate responses over time and space showed more sensitivity to precipitation mean spatial biases and less so on extremes. In the investigated catchments, the projected change of groundwater levels and basin discharge between current and future conditions was found to be modest across multiple climate models and bias correction methods. The choice of RCM contributes almost all of the uncertainty across precipitation inputs, explaining 99% of uncertainty in total annual precipitation, while choice of bias correction method contributes far less to overall uncertainty, contributing to 7% of variance in total annual precipitation values. Similarly choice of climate model contributes to almost all of the variance in stream discharge, 97% in total annual, while choice of bias correction method contributes to 10% of uncertainty in total annual discharge.

UR - https://rex.kb.dk/primo-explore/fulldisplay?docid=KGL01008787534&context=L&vid=NUI&search_scope=KGL&tab=default_tab&lang=da_DK

M3 - Ph.D. thesis

T3 - Danmarks og Grønlands geologiske undersøgelse

BT - Uncertainty in hydrological change modelling

PB - Department of Geosciences and Natural Resource Management, Faculty of Science, University of Copenhagen

ER -