Fig. 1.

Geographic distributions of observed (a) rainfall (mm/d), (b) daily maximum surface air temperature (°C), (c) incident solar radiation (W/m2), (d) evapotranspiration (mm/d), and (e) wind speed at 10-m height (m/s) averaged across the growing season during 1979 to 2005.

 


Fig. 2.

Geographic distribution of cotton yield relative mean biases [] for the cross-validation, where denotes the mean of cotton yields during 1979 to 2005 while the subscripts p and o represent predictions and observations, respectively.

 


Fig. 3.

Geographic distributions of (a) observed and (b) modeled 1979 to 2005 mean cotton yields.

 


Fig. 4.

Geographic distributions of GOSSYM simulated mean (a) water stress factor, (b) carbohydrate stress factor, (c) N stress factor on the cotton boll, (d) maximum leaf area index (LAI), and (e) maximum plant height as averaged across the growing season during 1979 to 2005.

 


Fig. 5.

Geographic distributions of cotton yield interannual standard deviations for (a) observations and (b) GOSSYM simulations, and (c) their interannual correlation during 1979 to 2005.

 


Fig. 6.

Evolution during the growing season (1 April–27 October) of the percentage of area with significant interannual correlations between cotton yield and 30-d running windows of rainfall (Rain), incident solar radiation (Rswi), daily maximum temperature at 2-m height (Tmax), and root-zone soil temperature at 1-m depth (ST1m) as observed (solid lines) and simulated by GOSSYM (dashed lines) during 1979 to 2005. Assuming yearly independence, statistical significance occurs at the 95% level when the correlations are >0.32. The result is shown for rainfed (left) and irrigated (right) lands.

 


Fig. 7.

Evolution from 1 May to 30 October of interannual relationships between annual cotton yields and preceding 30-d running means of leaf area index as measured by the gross correlation (solid thick line) and mean correlation (dashed thick line) as well as the frequency of positive (solid thin line) and negative (dashed thin line) correlations significant at the 95% confidence level across the harvest areas of the entire U.S. Cotton Belt.