Comparison of climate change scenario construction methodologies for impact assessment studies T. Mavromatis a,* , P.D. Jones b a National Center for Atmospheric Research, ESIG, P.O. Box 3000, Boulder, CO 80307, USA b Climatic Research Unit, University of East Anglia, Norwich NR4 7TJ, UK Received 6 December 1997; accepted 12 February 1998 Abstract This paper explores three important aspects of studies that assess the possible effects of climatic change on agricultural productivity at regional spatial scales. First, long-term historic and stochastically generated (WGEN) weather records are compared in terms of their statistical attributes using the climate conditions found in central France. Second, our results show that the use of CERES-wheat coupled with WGEN produced weather data provides an ef®cient method for assessing the impacts of changing climate on average agricultural production. Less con®dence can be placed, however on the estimation of future agricultural risk and variability assessment. Finally, time series of climate variables with changed mean and variability are either constructed according to the methodology proposed by Mearns et al. (1992) or simulated with WGEN using the approach suggested by Riha et al. (1996). The climate change scenarios are compared in terms of their effects on wheat development and predicted yield with CERES-wheat using daily data from the sulphate integration of the HadCM2 General Circulation Model to drive the crop model. The comparison of the different approaches for the construction of climate change scenario demonstrates the relative importance of changes in the mean climate and short/long-term variability in the prediction of crop yield on a regional basis. The results also indicate that the strength of the yield response to such combined scenarios and sometimes even its sign, depends on the qualitative nature of the change. Therefore, assessments of future agricultural productivity based on this methodological approach must be regarded as speculative. # 1998 Elsevier Science B.V. All rights reserved. Keywords: Wheat (Triticum aestivum L.)-modelling; Yield; Climate change; Climate impacts; Weather generators 1. Introduction There is considerable quantitative uncertainty concerning how agricultural crops respond to changes in climate variability, although it is known qualita- tively that changes in variability can have serious effects (Parry and Carter, 1985). Moreover, in the face of possible climatic change, there is mounting evidence that changes not only in the mean state of the climate but also in the higher order moments will occur (e.g., Mearns et al., 1995a, b; Gordon et al., 1992 and IPCC, 1992). Crop simulation models incorporate a mixture of non-linear responses of the crop to its environment and the simulations critically depend on the sequence of weather events. It is thus equally important to include Agricultural and Forest Meteorology 91 (1998) 51±67 *Corresponding author. 0168-1923/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved. PII S0168-1923(98)00063-X