基于HJ-CCD时序数据提取镇江市水稻种植分布

Mapping Rice Paddy Distribution in Zhenjiang City Based on Multi-temporal HJ-CCD Imagery

基于传统经验公式法和辐射传输法,分别对江苏省镇江市水稻分布状况进行提取。经验公式法可基于植被指数与叶面积指数的统计关系求得,以归一化植被指数和增强植被指数分别建立统计模型,而辐射传输模型采用PROSAIL模型,建立查找表,通过对比模拟和卫星遥感反射率数据对水稻的LAI(叶面积指数)进行确定。试验依据水稻生育期间LAI的时相分布,确定镇江水稻的种植分布状况。结果表明:(1)由经验公式法得到的镇江水稻面积为7.95×104hm2,相对误差为14.9%,由辐射传输法得到的镇江水稻面积为8.65×104hm2,相对误差为7.5%。辐射传输法的结果更接近实际值。(2)在统计模型中,增强植被指数EVI与LAI相关性比归一化植被指数NDVI与LAI的相关性更高,用EVI-LAI方程对LAI作进一步反演效果更理想;(3)实测值验证分析表明,统计模型反演的LAI标准误差为1.32,PROSAIL模型反演的LAI标准误差为0.97,说明在遥感反演植被生物物理参数时,辐射传输法比传统的经验公式法稳定性更强。研究结果对探讨两种方法在提取作物分布范围中的应用,准确了解作物长势和种植面积以指导生产具有重要意义。

The distribution of rice in Zhenjiang, Jiangsu province was separately extracted based on traditional empirical formula method and radiate transfer method. Empirical tormula method can be calculated by the statistical relationship between VI and LAI. NDVI and EVI were applied to establish the statistical model separately. As for radiative transfer method, the PROSAIL model was adopted and LUT was established firstly. Then the LAI of rice was determined through the comparison of the simulated reflectance data and the satellite remote sensing reflectance data. The distribution of rice planting in Zhenjiang was determined by the phase distribution of LAI during the period of rice growth. The results showed as followings： （ 1 ） by empirical formula method, Zhenjiang＇s rice area was 7.95 × 10^4 ha and relative error was 14. 9%. While by radiate transfer method,the rice area was 8.65 × 10^4 ha and relative error was 7.5%. As a result,the result simulated by radiate transfer was closer to actual value than by traditional empirical formula method. （2） In statistic model, the correlation between EVI and LAI was higher than the correlation between NDVI and LAI. Thus ,the result of further inversion of LAI would be more ideal when EVI - LAI method was used. （3） The validation analysis of measured values showed that LAI RMSE inverted by statistical model was 1.32 and LAI RMSE inverted by PROSAIL model was 0. 97. So radiate transfer method was more stable than traditional empirical formula method when remote sensing inversion of vegetation biophysical parameters. Research results of this paper are significant in the discussion of the application of traditional empirical formula method and radiate transfer method when crop distribution scopes are extracted. Besides, results play an important role in accurate understanding of crop＇s growth and planting area and have a good guidance for production.