《Table 3 Parameter estimation results from the multiple linear regression model》

  提示：宽带有限、当前游客访问压缩模式

本系列图表出处文件名：随高清版一同展现

《"Environmental factors influencing snowfall and snowfall prediction in the Tianshan Mountains, Northwest China"》

1. 获取 高清版本忘记账户？点击这里登录

1. 下载图表忘记账户？点击这里登录

Note:RH，relative humidity.

The ntree，mtry and nodesize parameters in the calibration period of the RF model were 500，3and 5，respectively，according to the selection standard for the specific parameter（see Section 2.3）.A larger proportion of the total variance in snowfall was explained in this model，with R2 values of0.70 and 0.74，d values of 0.90 and 0.93，and RSR values of 0.55 and 0.51 during the calibration and validation periods，respectively（Table 4）.These results clearly showed the desirable performance and prediction capability of the RF model.The observed and predicted values of snowfall for the validation samples from the RF and MLR models are presented in Figure 3.Generally speaking，the RF model performed much better than the MLR model in predicting snowfall.The predicted results are better when the observed snowfall values are less than 30 mm in both models.However，the overall predicted snowfall amounts were overestimated for low values and underestimated for high values.