User Tools

Site Tools


clipc:start

CLIPC Bias-Correction

Available bias-corrected data

Bias-corrected EUR-44 CORDEX and EUR-11 CORDEX RCM data made by SMHI, IPSL and Met Norway for CLIPC are now available (with others) through the ESGF project “CORDEX-Adjust”. More information can be found here.

PLEASE REPORT to Andreas Dobler, if you find something strange or errors in the Met Norway data.

Please have a look at some Evaluation plots considering the bias-corrected data from Met Norway.

An overview on the bias-corrected data from Met Norway can be found here BiasCorrectedDataList.pdf

Project and deliverable summary

The CLIPC project provides access to climate information of direct relevance to a wide variety of users, from scientists to policy makers and private sector decision makers. Information will include data from satellite and in-situ observations, climate models and re-analyses, transformed data products to enable impacts assessments and climate change impact indicators.

This page documents the datasets produced by the Meteorologisk institutt within the CLIPC project under the deliverable ”D6.1 : Climate model data for Europe, bias-corrected when necessary, for CCII-T1 calculation. Documented dataset”. The datasets is one of the first ensemble of multi-method bias-corrected datasets derived from the EURO-CORDEX RCM ensemble at two spatial resolutions and complemented by bias-corrected GCM data:

  1. Bias-corrected regional climate model (RCM) data at the EURO-CORDEX 0.11° (~ 12 km × 12 km) grid
  2. Bias-corrected RCM data at the EURO-CORDEX 0.44° (~ 50 km × 50 km) grid
  3. Bias-corrected global climate model (GCM) data interpolated to a common 2° × 2° grid,

The Meteorologisk institutt provides bias-corrected data at the EURO-CORDEX 0.11° and the EURO-CORDEX 0.44° grid. These will also contribute to the Bias Correction Intercomparison Project BCIP.

Terms of use

All EURO-CORDEX simulations are published under the CORDEX terms of use. Until further notice, the terms of use for the bias-corrected data from CLIPC are the same as those from the uncorrected EURO-CORDEX simulations obtained from ESGF servers. These are:

  • All simulations have non-restricted use

We acknowledge the following grants and work. Please acknowledge/cite these when you use or redistribute the bias-corrected CLIPC data from Met Norway.

  • The CLIPC EU project (grant agreement #607418)
  • The EURO4M FP7 project (grant agreement #242093) and Hāggmark, L., Ivarsson, K.-I., Gollvik, S. and Olofsson, P.-O., 2000: MESAN, an operational mesoscale analysis system. Tellus A, 52: 2–20. doi: 10.1034/j.1600-0870.2000.520102.x for the EURO-CORDEX 0.11° reference/training data MESAN
  • Gudmundsson, L., Bremnes, J. B., Haugen, J. E., and Engen-Skaugen, T., 2012: Technical Note: Downscaling RCM precipitation to the station scale using statistical transformations – a comparison of methods. Hydrol. Earth Syst. Sci., 16, 3383-3390, doi:10.5194/hess-16-3383-2012 for the bias-correction method used by Met Norway (quantile mapping).

Description of the Met Norway method

The method from the Meteorologisk institutt uses the R package qmap: Statistical transformations for post-processing climate model output. Basically, it performs a quantile mapping of the simulated time series to observed ones for each grid point. It does so by estimating values for regularly spaced quantiles of the empirical cumulative distribution function and using these estimates to perform a quantile mapping. Details on the method can be found in Gudmundsson et al. (2012, doi:10.5194/hess-16-3383-2012). The set-up includes a linear interpolation between the fitted transformed values, and simulated values lying outside the range of the training period (e.g., from climate projections) are extrapolated using the correction found for the highest quantile of the training period, as suggested by Boé et al. (2007, doi: 10.1002/joc.1602). Furthermore, the method includes an adjustment of wet-day frequencies for precipitation. For temperature, the quantile mapping takes into account the main trend by subtracting the (5-year) mean before the quantile mapping and adding it again afterwards. This is expected to reduce the impact on the mean climate change signal.

  • For the bias-corrected EUR-11 CORDEX data, the daily 0.11° MESAN data has been used as reference/training data covering 1989-2010. The complete historical RCM data has been used for calibration.
  • For the bias-corrected EUR-44 CORDEX data, the daily 0.44° EOBS12 data has been used as reference/training data covering 1981-2010. For calibration the RCM data has been restricted to the same period (1981-2010) following the BCIP experiment design.

(Scripts etc.: Procedure)

Known issues

Please note the following:

  • There is a small remaining overall bias due to the use of “only” 101 quantiles to estimate the empirical cumulative distribution functions within qmap. See pdfs on Evaluation.
  • There are additional remaining seasonal biases due to the calibration process which is carried out without splitting into seasons or months. See pdfs on Evaluation.
  • The bias-correction method does not guarantee the conservation of climate change signals. The impact on the projected changes in mean conditions is relatively small, as can be seen here:

ClimChangeSignal_tas.pdf
ClimChangeSignal_pr.pdf
ClimChangeSignal_pr_rel.pdf

  • There are some days in the MESAN data when tasmax < tas or tas < tasmin. In the current set-up this is handed down to the bias-corrected data.

While the first three points are a result of the idea to keep the method very simple (at least for the beginning), the last point is under investigation and may be fixed in a later version.


Andreas Dobler 2015/07/15 11:00

clipc/start.txt · Last modified: 2016-11-03 13:19:38 by andreasd