基于OLI影像的四川丘陵地区水稻种植面积监测

摘要

[目的]为使用Landsat8 OLI遥感影像准确监测四川丘陵地区水稻种植面积.[方法]根据丘谷相对高差分别选定浅丘、深丘水稻样方各4个,通过地面样方调查和同生长季Google Earth影像解译获取的样方水稻种植面积作为验证数据,评价基于OLI融合影像的川中丘陵的水稻种植面积监测精度.OLI影像经正射校正后,使用15 m全色波段影像和30 m多光谱影像融合,得到15 m分辨率的融合影像,使用最大似然法进行监督分类,获取监测结果.[结果]与样方验证数据比较,在样方面积相同的情况下基于OLI影像的水稻种植面积监测结果显示,浅丘区的平均精度为93.7％,误差范围为1.0％～8.7％;深丘区的为92.5％,误差范围为1.5 ％～15.8％.[结论]根据浅丘区、深丘区监测结果的误差范围差异,随着丘谷高差增大,地形趋于复杂、地块趋于狭长与破碎,OLI影像监测结果的不确定性增加,精度有下降的趋势.该研究为改进OLI影像监测四川丘陵地区水稻种植面积精度提供参考.%[Objective] The present paper was conducted to study the accuracy of using Landsat8OLI remote sensing image to monitor the rice planting area in hilly areas of Sichuan province.[Method] Four kinds of rice cultivars from shallow hill and deep hill rice were selected respectively according to the valley height.And the area of rice planting was obtained by ground sampling survey and Google Earth image interpretation in the same growing season as validation data.This result was used to evaluate the monitoring accuracy of rice planting area in the hilly region of Sichuan Province based on OLI fusion images.After orthophoto correction of OLI image,15 m panchromatic band image and 30 m multispectral images were used to fusion to get 15 m resolution fusion image.And supervised the classification with maximum likelihood method to get the monitoring result.[Result] Compared with the quadrat verification data,the results of rice planting area monitoring based on OLI images with the same sample area showed that the average accuracy of the shallow hill area was 93.7 ％,the error range was 1％ to 8.7 ％.And the depth area was 92.5 ％,the error range was 1.5 ％ to 15.8 ％.[Conclusion] According to the error range of the monitoring results in the shallow hilly area and the deep hilly area,with the increase of the valley height,the terrain tended to be complicated,the plot tended to be narrow and broken,and the uncertainty of the OLI image monitoring result increases.And in the same time,the accuracy of OLI image monitoring has a downward trend.This study provides a reference for improving the accuracy of OLI image monitoring of rice planting area in hilly areas of Sichuan province.