Fig. 3.

(a) Importance of four climate variables in (a) CEMs and (b) BRTs for modelling rainfed rice cultivation. In (a) the y-axis is the mean correlation coefficient (Pearson's r) (and SE) from model projections made with a single climate variable against predictions made by using all four variables. In (b) the y-axis plots the relative influence of each variable (higher numbers indicate stronger influence). Refer to Section 2.3 for a brief description and Friedman and Meulman (2003) for full details.