Since 1980s of the 20th century, outbreak of oriental migratory locust has rampantly emerged again in some regions of China. It is particularly important to monitor timely and accurately the intensity of damage from oriental migratory locust (Locusta migratoria manilensis Meyen) for realizing efficient control and prevention of this kind of insect pest.In this study, Huanghua city in Hebei province was taken as the study area. From the point of view of the mechanism of developing the optical models, four algorithms used for LAI retrieval based on the gap fraction of vegetation canopy were quantitatively analyzed and compared. These methods are the Bonhomme & Chartier algorithm, the LAI-2000 algorithm, the improved LAI-2000 algorithm and the Campbell’s ellipsoid distribution algorithm, respectively. Three types of vegetation which are most popularly distributed in the study area, i.e. reeds, cottons and weeds, are selected as the targets and the original datum of vegetation get collected from 14 sampling sites and used in the study. A fish-eye digital camera manufactured by the Regent Instrument Company was utilized to obtain the spherical images of vegetation canopy, and WINSCANOPY2004a was made use of in processing of the datum of gap fraction of vegetation canopy. In order to compare the accuracy of LAI retrieved by four kinds of algorithms and find out one which is mostly applicable to LAI retrieval in the study area, the relations between LAI retrieved by the different kinds of algorithms, their logarithmic averages of the retrieved LAI and the corresponding vegetation index are statistically analyzed, respectively. The results show that, firstly, a positive correlation exists between the retrieved LAI and vegetation index for all three kinds of vegetation in the study area. Secondly, the accuracy of LAI retrieval by the LAI-2000 algorithm and the Campbell’s ellipsoid distribution algorithm is higher than that of the Bonhomme&Chartier algorithm and the improved LAI-2000 algorithm. Thirdly, with the Bonhomme&Chartier algorithm excluded, the fitting process between LAI retrieved by the other three kinds of algorithms and RDVI retrieved from TM indicates that these algorithms all perform much well to reveal the growing conditions of vegetation in the study area, and, among them the LAI-2000 algorithm is the best one in terms of the accuracy of LAI retrieval. Finally, the fitting between the logarithmic average of the LAI retrieved by three kinds of algorithms and RDVI retrieved from TM also shows that the LAI-2000 algorithm is the greatest for the LAI retrieval of the vegetation in the study area. In addition, it is found that a negative correlation exists in the relation between LAI and the area where the locust outbreak appeared. In other words, with the decrease of LAI, the area in which the locust outbreak appears increased. Following this step, a forecasting model based on the datum of outbreak areas and LAI was developed to predict the area where the locust outbreak(ALO)would emerge，which is ALO=－aln(LAI)+b．In this fitting, a and b are both adjustable coefficients.