Results for linear regression models predicting the natural logarithm of annual area burned.

Global r2 or R2adj (for multiple regression models) reflects comparisons between climate metrics and area burned over the entire 107-year dataset. Median r2 or R2adj reflects the median value from the 86, 21-yr calibration periods, and median RE reflects the median value from 86, 86-yr cross-validations prediction periods. The product of median r2 (or R2adj) and median RE (“Score”) was used to rank each predictor variable by integrating accuracy (r2 or R2adj) and prediction skill (RE).Results for linear regression models predicting the natural logarithm of annual area burned.

全局决定系数r²或调整后决定系数R²_adj（适用于多元回归模型）用于表征整个107年数据集内气候指标与过火面积之间的拟合关联程度。中位数r²或调整后R²_adj代表86个21年校准时段的拟合系数中位数，而中位数RE则为86个86年交叉验证预测时段的RE中位数。将中位数r²（或调整后R²_adj）与中位数RE的乘积定义为评分（"Score"），以此结合拟合精度（r²或调整后R²_adj）与预测能力（RE）对各预测变量进行排序。本结果基于针对年度过火面积自然对数构建的线性回归模型所得。