基于FY-3 MERSI遥感数据的水稻种植分布提取

Extraction of Pddy Rice Planting Area Based on Multi-Temporal FY-3 MERSI Remote Sensing Images

快速准确获取水稻种植分布对于制定区域农业生产政策、保护区域农业稳定具有科学意义。以风云系列为代表的卫星资料在农业遥感方面已得到了广泛使用,但目前鲜有基于国产风云遥感数据反演水稻种植空间分布的研究。为快速准确获取水稻种植分布,挖掘风云遥感资料在反演水稻种植信息领域的数据价值,以盘锦市作为实验区域,开展基于FY-3 MERSI卫星资料的水稻种植空间分布反演。利用5景2019年研究区域水稻生育期风云三号中分辨率MERSI(Medium Resolution Spectral Imager)光谱成像仪数据计算归一化植被指数(NDVI)、归一化水体指数(NDWI)、两者之间的差值(NDWI-NDVI)及比值植被指数(RVI),对盘锦市水稻以及其他地物类型(建筑用地、水体、自然植被、天然湿地和旱地)感兴趣区域进行植被指数时序分析,利用NDVI、 NDWI、 RVI和NDWI-NDVI时间序列曲线确定最佳识别模式及其阈值,发展水稻种植空间分布遥感反演算法。首先根据水稻移栽期NDWI-NDVI>-0.14和抽穗期NDWI-NDVI<-0.4对水稻种植分布进行粗提取,在此基础上依据水稻与其他地物类型NDVI、 NDWI和RVI曲线特征差异对其他地物类型进行掩膜,得到2019年研究区域水稻种植空间分布。基于实地调查数据对研究区域水稻种植空间分布反演结果开展验证评价,总体精度为75%。基于目视解译的水稻空间分布数据开展验证评价,总体精度、制图精度以及用户精度均达到了80%以上,Kappa值为0.61。研究区域2019年水稻面积为116 618.75 hm^(2),与2019年盘锦市统计年鉴公布数据基本一致。研究表明,基于风云三号卫星资料反演水稻种植空间分布能够满足区域农作物种植分布遥感监测的要求,FY-3 MERSI遥感数据在农作物种植空间分布提取中具有应用价值。该研究丰富了农作物种植分布监测的遥感数据源,对于深入风云卫星资料的实际应用具有重要科学意义。

Rapid and accurate monitoring of paddy rice planting areas distribution plays an important role in formulating regional agricultural production policies and protecting regional food security.With the successful launch of FY series satellites,domestic satellite data have been increasingly used in crop information monitoring,but there are few studies on the extraction of paddy rice planting distribution information based on FY data.In order to quickly and accurately obtain paddy rice planting distribution information and explore the application potential of FY remote sensing data in monitoring paddy rice planting distribution,the study was conducted to extract paddy rice planting distribution based on FY-3 MERSI data in Panjin county,Liaoning Province.Five images of FY MERSI data during the growth period of paddy rice in 2019 were used to calculate the Normalized Difference Vegetation Index(NDVI),Normalized Difference Water Index(NDWI),Ratio Vegetation Index(RVI)and NDWI-NDVI.The temporal sequence analysis of these vegetation indices was carried out on the interest areas of six land cover types in Panjin county,including paddy rice,building land,water body,natural vegetation,natural wetland and dry land.The optimal recognition mode and threshold were determined using NDVI,NDWI,RVI and NDWI-NDVI time series curves,and the remote sensing extraction model of paddy rice planting distribution was established.First,the paddy rice planting distribution was roughly extracted according to NDWI-NDVI>-0.14 at the transplanting stage and NDWI-NDVI<-0.4 at the heading stage.Then,other land cover types were masked based on the difference of NDVI,NDWI and RVI curve characteristics between paddy rice and other land cover types,and the spatial distribution of paddy rice planting in the study area in 2019 was obtained.Based on field survey data,the accuracy of paddy rice planting distribution in the study area was verified,and the overall accuracy was 75%.Accuracy verification was also conducted based on remote sensing visual interpretation data,the overall accuracy,Kappa coefficient,paddy rice mapping accuracy and user accuracy were 80.80%,0.61,80.00%and 86.96%,respectively.The paddy rice planting area in the study area in 2019 was 116618.75 hm^(2),consistent with the data published in the 2019 Panjin Statistical Yearbook.The study shows that extracting paddy rice planting distribution based on FY-3 MERSI remote sensing image can satisfy the requirements of remote sensing monitoring of regional crop planting distribution.FY-3 MERSI has great application potential in extracting crop planting distribution information.The study enriches the remote sensing data sources for crop planting distribution monitoring and provides a theoretical basis for promoting the practical application of FY data.