Four principal steps were used to complete the analysis. First, the climate change projections from HadCM and CGCM were superimposed on the data from 72 weather stations across Mali.
Second, the biophysical models were run with the CO2 level set at 1.5 times the current level of 330 ppmv (parts per million by volume). The mean and standard deviation of yields across 85 agroecological zones in Mali were computed and weighted into sensitivity data for nine MASM production zones. (Information needed to set up EPIC simulations was not available for Tombouctou, the northernmost region of Mali. EPIC responses from the nearby Segou region were used as representative of Tombouctou.)
Besides estimating climate change impact, the EPIC was also used to estimate the consequence of degradation on yields. In particular, 40 years of simulated yield were used to estimate the regression
Yijst = atjst + btjst T + £ijst where Yijst is the yield of item i, in simulation zone j, on soil type s, for technology t; and T is a time trend variable. The intercept a in this regression shows the base year yield; the coefficient b shows how yields change over time; and j shows the error term in the regression. The coefficient b was generally found to be negative, showing degradation over time. Yield for 2030 was estimated by setting T equal to 34 in the estimated regression. Degradation rates estimated through regression were applied to the MASM base yields. (The information available restricted our calibration of EPIC to using observed crop yields only in Sikasso region. Hence, the trend regressions were estimated for 43 agroecological zones in Sikasso. The degradation estimates averaged across agroeco-logical zones in Sikasso-North, a relatively dry area, were applied to other areas in Mali.)
Third, the biophysical climate change and degredation scenarios were imposed on MASM. In particular, yields across the 12 states of nature were adjusted to reflect changes in both the mean and the dispersion of yields, using the procedure employed by Lambert et al. (1995). Therein, the new yields became where superscripts cc and b, respectively, represent with climate change/degradation and without (base) parameters; MeanYld is the mean of the yields in the MASM base model; Yieldf and Yieldf are yields for the i-th uncertain weather year under base conditions and climate change/degradation; ^ and a are, respectively, the mean and standard deviation of the simulated biophysical yields.
Fourth, the possible influences of climate change on international grain prices were imposed on MASM using the results from the international market component of the 2002 U.S. national assessment study of climate change impacts in the United States (Reilly et al., 2002).
MeanYld (+ (Yieldb - MeanYld
Was this article helpful?