Final technical report

Prudhomme, Christel; Crooks, Sue; Jackson, Christopher; Kelvin, Jon; Young, Andy. 2012 Final technical report. Wallingford, UK, NERC/Centre for Ecology & Hydrology, 118pp. (CEH Project Number: C04004, Science Report/Project Note: SC090016/PN9)

Before downloading, please read NORA policies.
[img] Text
Restricted to NORA staff only

Download (11MB)


This report is a technical summary of the Future Flows and Groundwater Levels (FFGWL) partnership project between NERC (Centre for Ecology and Hydrology (CEH) and British Geological Survey BGS), Wallingford HydroSolutions, the Environment Agency of England and Wales (EA), the Department for Environment, Food and Rural Affairs (Defra), and the UK Water Industry Research (UKWIR). This report evaluates the level of uncertainty associated with the methods used and the data generated. This is very important as uncertainty associated with all FFGWL data and products needs to be understood when interpreting findings based on them. It also provides a brief summary of the river flow changes for a single time horizon (2050s) projected for Great Britain GB by the project. FFGWL has delivered a nationally consistent ensemble of river flow and groundwater level time series (Future Flow Hydrology FFH) of 148 years simulated by surface and ground water models using Future Flows Climate FFC. FFC is a gridded daily precipitation and monthly potential evapotranspiration time series ensemble derived from HadRM3-PPE which was specifically bias-corrected and downscaled at 1- and 5-km in the project for hydrological applications. FFC contains an ensemble of 11 plausible time series from 1950 to 2098, capturing natural climate variability and some climate change uncertainty, as described by 11 different Global/Regional Climate Model variants. Five surface and ground water modelling codes were used in FFGWL, all with different model structure and calibration emphases but their performance was assessed using a common modelling protocol for the project study sites. For the majority of the sites the performance was found to be good for low, median and high flows but for some sites surface water modelling uncertainty could be large. Generally there is no seasonal pattern in the modelling errors but in the south and east of GB, uncertainty in summer flow is the largest, attributed to catchments with complex hydrological processes (in particular in groundwater-dominated areas) and with a fine equilibrium between precipitation and losses through evapotranspiration. For groundwater levels modelling errors show positive and negative modelling biases. There is no seasonal pattern in the errors linked to specific aquifer type. FFH uncertainty results from the combination of climate variability, climate modelling, bias-correction and downscaling uncertainty, and surface and ground water uncertainty. Comparison between FFH and gauged river discharge suggests that the greatest combined errors in FFH time series typically occur in dry conditions and drier regions. For groundwater levels biases show a tendency to underestimate levels. When surface and groundwater modelling uncertainty is removed (by comparing FFH with simulations driven with observed climate) modelling errors do not show any pattern linked with flow magnitude or the season for both surface and ground water. This suggests that while some hydro(geo)logical uncertainty is introduced in FFH it is smaller compared to climate variability and climaterelated uncertainty. The climate change signal is found to be independent of surface and ground water model structure hence suggesting that the national database of FFH can be compared even if the sites’ time series are simulated with different models. For all project surface water sites with available CERF model two methods were used to transfer the information contained in FFC to the hydrological models (also called downscaling method): (i) The transient method uses FFC directly to generate catchment average climate time series used to drive CERF to generate FFH; this produces an ensemble of 148-year daily river flow simulations from 1951 to 2098. Differences between statistics calculated from two independent periods of FFH (comparing pairs from the same ensemble member) provide an assessment of the climate change impact; (ii) The scaling method first calculates changes in precipitation and PE as differences between catchment monthly averages calculated for the periods 1961-1990 and 2040-2069 from FFC (comparing pairs from the same ensemble member), then applies the changes to observed precipitation and PE time series, which become input to CERF. The difference between statistics calculated on the paired CERF simulations (using observed and scaled climate input) provide an assessment of the climate change impact. This was used to produce the river flow maps of changes for the 2050s. The scaling method shows a larger range of river flow changes compared to the transient method and in particular larger winter flow increases and summer flow decreases. The systematic differences were not investigated in detail but could relate to change being applied to exceptional observed periods (e.g. 1975-1976 in some regions of GB) which might not appear in the FFC and resulting FFH for 1961-1990. Note that the variation in the signal of river flow change due to different downscaling methods is generally smaller than the variation between different ensemble members (expressing both climate variability and climate change uncertainty). For contrasting catchments FFH was compared with simulations based on the 10,000 UKCP09 probabilistic ensemble scenarios using the scaling downscaling method. While results are catchment specific, FFH generally captures most of the range associated with simulations using the UKCP09-probabilistic ensemble but does not describe the upper tail of the distribution for both surface and ground water sites well. The FFH distribution seems to be towards the lower half of the distribution of changes resulting from UCKP09 scenarios but this is difficult to generalise due to the small number of catchments used for the comparison. The overall pattern of changes for the 2050 within FFGWL is very complex spatially and no clear spatial pattern of changes is shared by all ensemble members. In winter (December, January, February) there is a mixed pattern in England and Wales with drier, similar or wetter signals, within - 20% to +40% change (one scenario with up to 60% in a small region). In contrast, river flows in Scotland show a small increase or decrease, although this is still mainly within ± 20% with changes in the west reaching up to 40%. In spring (March, April, May) more of the RCM scenarios are drier for most of the UK, with decreases of up to 40%. However, for 3 scenarios central England has increased flows (up to 60%). In summer (June, July, August) scenarios predominantly show decreases in runoff through the UK, but range from +20% to -80%. The largest percentage decreases are mainly in the north and west of the UK although the range in these areas between scenarios can be large (0 to - 80%). In autumn (September, October, November) there is a mixed pattern with a full range of percentage changes (+60 to -80%) across the UK. Most scenarios indicate decreases in flows, especially in the south and east (up to -80%) whilst in the west and north changes can be small. One scenario shows no change or an increase in runoff across the UK. For the first time a national ensemble of 148-year river flow and groundwater levels time series (FFH) and the associated gridded climate projections (FFC) were produced for Great Britain and made available to the research community and project partners. The 11-member ensemble spans both historical (starting in 1951) and future (ending in 2098) periods. These products enable water-related climate change impact studies to be be made across GB using a consistent method and data. The complex spatial pattern of changes suggest that all 11 members must be considered for a better understanding of the climate change uncertainty. The uncertainty associated with FFH was found to mainly result from climate variability and other uncertainty in FFC, with additional surface and ground water modelling uncertainty of smaller magnitude and modelling structure not overly impacting on the climate change signal. It is hence possible to compare results for different sites regardless of the impact model structure but it is not recommended to consider directly FFH time series as this will contain all combined modelling uncertainty. Instead relative changes between different time periods of the same ensemble members can be calculated. It is highly recommended to understand the level of modelling uncertainty associated with FFH fully when interpreting the results. A summary ‘fact sheet’ containing site specific information regarding FFC and FFH is associated with every FFH data package.

Item Type: Publication - Report (UNSPECIFIED)
Programmes: BGS Programmes 2010 > Groundwater Science
CEH Topics & Objectives 2009 - 2012 > Water > WA Topic 1 - Variability and Change in Water Systems
CEH Topics & Objectives 2009 - 2012 > Water > WA Topic 3 - Science for Water Management
CEH Sections: Reynard
Funders/Sponsors: Environment Agency, UK Water Industry Research, Defra, Wallingford HydroSolutions Ltd
Additional Information. Not used in RCUK Gateway to Research.: Available online - click on Official URL link for full text
Additional Keywords: GroundwaterBGS, Groundwater, Climate change, Groundwater modelling
NORA Subject Terms: Hydrology
Data and Information
Related URLs:
Date made live: 04 Dec 2012 12:29

Actions (login required)

View Item View Item

Document Downloads

More statistics for this item...