An effective projection pursuit classification model is suggested to evaluate co mpensatory effects of water recovery after drought in order to avoid the uncerta inty of the evaluation and the optimization, and to improve the accuracy of the evaluating model. It adopts an ant colony algorithm based on mutation and dyna mic pheromone updating to find the best projective direction, and then uses the best projective direction to calculate the contribution of each evaluation index to the compensatory effects. The test results showed that photosynthetic rate d irectly reflected the capacity of compensatory growth. Thus it is a critical f actor for compensatory effects, which conformed to the previous studies. The opt imal compensatory effects appeared in the fifth day after appropriate rewatering . 