Yield estimation of semilate rice using remote sensing data in Honghu city, China

China has the hugest population in the world. Accurate crop yield data is especially needed by Chinese agriculture and food policymakers. Traditional measuring techniques couldn´t meet the requirement. Nowadays remote sensing data has been widely applied to crop yield estimation. However, the existing studies on crop yield estimation were mostly done at the large scale such as global, country, province, and city. Few cases were done at town scale, which cause a requirement for suitable method of crop yield estimation at the small scale. In this paper, one scene of IRS-P6 multi-spectral image and 33 field data of semilate rice´s yield were collected in Honghu city, China during the semilate rice growing. Based on it, relationship analysis was firstly researched between IRS-P6 data (i.e. NDVI, RVI and Near Infrared data) and yield per unit of semilate rice The results show there was strong correlativity (Pearson = 0.808) between RVI index and yield per unit of semilate rice. Then, several regression models, such as linear, twice, three times and compound power, were respectively established for yield estimation of semilate rice. According to the results from F test, the three-times regression model composed of RVI index and per unit area yield was selected, which is Y=-13.774X²+5.111X³+341.503, where X is RVI index and Y is yield per unit. Using this model, yield of semilate rice was estimated at the town scale. Based on the yield estimation results, a spatial distribution pattern could be found, i.e. whether as for the yield per unit or the total area and the total yield, the western towns are all higher than the eastern towns. Moreover, there is still plenty of median-yield and low-yield paddy fields, which shows the semilate rice will have great developing potentiality in Honghu City, which could be realized by scientific cultivation, arranging and reconstructing these fields. The above evidence-based research achievements will help local g- vernment make science agricultural economics policies.