Abstract: Based on remote sensing (RS) which has integrated and realistic characteristics, geographic information system (GIS) which has powerful spatial analysis ability, and artificial neutral network (ANN) which can optimize nonlinear complex systems, the forest volume in Huzhong forest area was estimated. The results showed that there was an obvious negative correlation between the forest volume and infrared band, indicating that infrared band had definite potential in estimating forest volume. The forest volume also negatively correlated with visible band and PC1. Among the topographic factors, altitude exerted more influence than aspect and slope on the estimation of forest volume. The correlation coefficient of predicted value and actual value reached to 0.973, when the optimal ANN parameter, suitable GIS information, and RS bands were adopted. After principal component transformation, the amount of observation data was effectively reduced, while the predicted precision only had a small decline (R²=0.934).

Key words: wheat, drought stress, source-sink regulation, water soluble carbohydrate, 1000-grain mass.