Climate Model Biases, Downscaling and Implications on Crop Model Simulated Climate Change Impacts.

In making projections of responses of crops to future climates, it is necessary to understand
and differentiate between the sources of uncertainty in climate models, and biases in
projections, and how these affect crop model estimates. This study investigates the
complexities in using climate model projections representing different spatial scales within
climate change impacts and adaptation studies. This is illustrated by using original and bias
corrected downscaled weather data from a Regional Climate Model (RCM) and the biases
introduced to three crop models’ outputs. We offer a cautionary warning on the potential difficulties in interpreting climate change impacts based on input weather data containing biases. Original and bias corrected downscaled hindcast (1960-1990) weather data from the HadRM3 RCM were evaluated against observed data (13 sites in the UK). Subsequently, observed, modelled original and downscaled hindcast and future projection data were used within CropSyst, DSSAT and APSIM cropping systems models to investigate the effect of data source on a range of models outputs estimates. Though the bias correction improved the match between observed and
hindcast data, this did not always translate into better matching crop models' estimates. At four sites the original HadRM3 data produced near identical mean yield values as from the observed weather data, despite differences in the weather data. This was due to compensating errors in the input weather data and non-linearity in crop models processes, making interpretation of results problematical. Overall, downscaling the climate data improved the quality of models'
estimates. Understanding how introduced biases in climate data manifest themselves in crop models gives greater confidence in the utility of the estimates produced using 
downscaled future climate projections.