Search site

Estimating climate impacts on river flow duration curves

Objectives: The Environment Agency and the University of Reading used the UKCP09 projections to estimate changes in the flow duration curve at a location. The study aimed to answer questions such as: how might Q95 change in a particular river by the 2050s? And, how might in-stream habitat suitability change by the 2030s?

UKCP09 products used: Observed data; User Interface

How were UKCP09 products used?

  1. A catchment hydrological model was calibrated for the Harper's Brook catchment. UKCP09 observed data was used to run the model was with catchment average daily rainfall data for the period 1961-1990 and daily potential evaporation constructed from MORECS monthly potential evaporation.
  1. Changes in mean monthly precipitation, mean monthly temperature, monthly relative humidity and cloud cover were downloaded for grid cell 1548. The 2050s were chosen for all three emissions scenarios and then each decade was examined for the Medium scenario.
  1. Data was also downloaded for an adjacent cell and for the Anglian region, which again looked at the 2050s under a Medium emissions scenario. The data sets were downloaded containing 10,000 scenarios for each climate variable for the site, year and emissions scenario of interest.
  1. Changes in mean monthly precipitation were applied to the observed baseline data (daily rainfall 1961-1990) for the catchment to produce a perturbed 30 year time series with a different mean. Changes in potential evaporation were calculated by applying changes in temperature, relative humidity and cloud cover to the CRU monthly data.
  1. Each of the scenarios had to be run separately through the hydrological model to produce an ensemble of 10,000 possible hydrological series.
  1. Finally, probability distributions were constructed for individual flow duration quantities.

Difficulties & limitations

One difficulty that was encountered was that the full set of 10,000 scenarios has to be used in order to produce an estimate of the distribution of possible outcomes. It is not possible to use the UKCP09 CDFs for this. Using a "small" number of scenarios can give misleading results.

Lessons learned

The study found that there is relatively little difference between the emissions scenarios. It also showed that the range in climate projections is large compared to natural, multi decadal variability alone.

Running a large number of climate scenarios through a further impact model requires careful and thorough data management.

How will the results be communicated to the target audience?

The main means of presenting the results are by using flow duration curves. These can be used for a number of flow quantiles and for annual or seasonal runoff characteristics. This approach also enables comparison of different emission scenarios, for example, or to examine change through time as a consequence of climate change.

Find out more

  • Contact details: Nigel Arnell, Walker Institute, University of Reading, and Glenn Watts, Environment Agency.