水稻种植面积遥感提取中田间作业路的消除方法
|
摘要
利用多时相遥感数据提取水稻种植面积是一种快速、高效的方法,但在高空间分辨率的遥感数据源中很难在一个作物物候期内获取多时相影像,高分一号卫星(GF-1/WFV)在16 m分辨率上实现了较高的时间分辨率,但由于存在混合像元,阈值分割尺度的设定会影响水稻面积提取精度。针对此类问题,混合像元分解模型可以有效排除异质地物的干扰。本文以GF-1/WFV遥感影像为数据源,选取银川平原水稻集中分布区破碎化程度不同且仅含有田间作业路和水稻的3个水稻样地作为实验区。首先,利用水稻分蘖期水体特征与孕穗期植被特征较为突出的特点,通过阈值法分类初步获取水稻的空间分布范围;然后在地表反射率遥感影像上选取田间作业路和水稻端元波谱曲线。结合线性混合像元分解模型,根据水稻丰度比例提取最终的种植面积。最后利用高空间分辨率的高分二号遥感影像对提取结果进行精度验证。结果表明,耦合两种方法提取水稻面积的面积精度为96. 33%,比阈值法提取水稻面积的精度提高了14. 63%,有效地排除了田间作业路对水稻面积提取精度的影响,为农作物种植面积信息的精确提取提供参考。
Extracting rice planting area with the multi-temporal remote sensing images is a rapid and effective method,but it is difficult to obtain the multi-temporal images from the high spatial resolution remote sensing data sources during one crop growing season. The high-resolution satellite GF-1/WFV achieves high temporal resolution at 16 m resolution,but the setting of the scale of threshold segmentation is to affect the extraction accuracy of rice planting area. For such problem,the mixed-pixel decomposition model can effectively eliminate the interference from the relevant heterogeneous surface features. By taking GF-1/WFV remote sensing image as the data source,three sample rice plots consisting of field operation roads and rice with different fragmentation degrees are selected herein as the experimental plots. Firstly,the spatial distribution range of the rice is preliminarily obtained through the classification of threshold method with the more prominent characteristics of the feature of water body during rice tillering stage and the vegetation feature of rice booting stage,and then the endmember spectrum character curves of the field operation road and rice are selected from the surface reflectance. In combination with the linear mixed-pixel decomposition model,the final planting area is extracted with the abundance ratio of rice. Finally,the accuracy of the extracted result is verified with thehigh spatial resolution remote sensing image of GF-2/WFV. The result shows that the accuracy of the rice planting area extracted with the coupling of both the methods is 96. 33%,which is higher than that obtained from the threshold method by 14. 63%. In this way,not only the impact from the field operation road on the extraction accuracy of the rice planting area is effectively eliminated,but also a certain reference for the accurate extraction of the information of crop planting area is provided.
Extracting rice planting area with the multi-temporal remote sensing images is a rapid and effective method,but it is difficult to obtain the multi-temporal images from the high spatial resolution remote sensing data sources during one crop growing season. The high-resolution satellite GF-1/WFV achieves high temporal resolution at 16 m resolution,but the setting of the scale of threshold segmentation is to affect the extraction accuracy of rice planting area. For such problem,the mixed-pixel decomposition model can effectively eliminate the interference from the relevant heterogeneous surface features. By taking GF-1/WFV remote sensing image as the data source,three sample rice plots consisting of field operation roads and rice with different fragmentation degrees are selected herein as the experimental plots. Firstly,the spatial distribution range of the rice is preliminarily obtained through the classification of threshold method with the more prominent characteristics of the feature of water body during rice tillering stage and the vegetation feature of rice booting stage,and then the endmember spectrum character curves of the field operation road and rice are selected from the surface reflectance. In combination with the linear mixed-pixel decomposition model,the final planting area is extracted with the abundance ratio of rice. Finally,the accuracy of the extracted result is verified with thehigh spatial resolution remote sensing image of GF-2/WFV. The result shows that the accuracy of the rice planting area extracted with the coupling of both the methods is 96. 33%,which is higher than that obtained from the threshold method by 14. 63%. In this way,not only the impact from the field operation road on the extraction accuracy of the rice planting area is effectively eliminated,but also a certain reference for the accurate extraction of the information of crop planting area is provided.
引文
[1]邓长春.水稻种植面积遥感信息提取研究[D].乌鲁木齐:新疆农业大学,2008.
[2]孙九林.中国农作物遥感动态监测与估产总论[M].北京:中国科学技术出版社,1996.
[3]史舟,梁宗正,杨媛媛,等.农业遥感研究现状与展望[J].农业机械学报,2015,46(2):247-260.
[4]游炯,裴志远,王飞,等.基于改进多元纹理信息模型和GF-1影像的县域冬小麦面积提取[J].农业工程学报,2016,32(13):131-139.
[5] YANG Y,HUANGY,TIAN Q,et al. The Extraction model of paddyrice information based on GF-1 satellite WFV images[J]. Spectrosco-py and spectral analysis,2015,35(11):3255-3261.
[6]李耀辉,王金鑫,李颖.应用时间序列EVI的MERSI多光谱混合像元分解[J].遥感学报,2016,20(3):459-467.
[7] LV T T,LIU C. Study on extraction of crop information using time-series MODIS data in the Chao Phraya Basin of Thailand[J]. Ad-vances in space research,2009,45(6).
[8]王凯,赵军,朱国峰,等.基于GF-1遥感数据决策树与混合像元分解模型的冬小麦种植面积早期估算[J].遥感技术与应用,2018,33(1):158-167.
[9]马孟莉.基于HJ卫星混合像元分解的水稻生长监测技术研究[D].南京:南京农业大学,2011.
[10]张永宏,何静,阚希,等.遥感图像道路提取方法综述[J].计算机工程与应用,2018,54(13):1-10,51.
[11]王兴盛.宁夏水稻[J].北方水稻,2007(1):19-22,31.
[12]刘佳,王利民,姚保民,等.基于多时相OLI数据的宁夏大尺度水稻面积遥感估算[J].农业工程学报,2017,33(15):200-209.
[13]郑长春.水稻种植面积遥感信息提取研究[D].乌鲁木齐:新疆农业大学,2008.
[14]童李霞,燕琴,骆成凤,等.基于NDWI分割与面向对象的水体信息提取[J].地理空间信息,2017,15(5):57-60.
[15]郝卫平,梅旭荣,蔡学良,等.基于多时相遥感影像的东北三省作物分布信息提取[J].农业工程学报,2011,27(1):201-207+393.
[16]李志鹏,刘珍环,李正国,等.水稻空间分布遥感提取研究进展与展望[J].中国农业资源与区划,2014,35(6):9-18.
[17]王连喜,徐胜男,李琪,等.基于决策树和混合像元分解的江苏省冬小麦种植面积提取[J].农业工程学报,2016,32(5):182-187.
[18]蔡薇,郭洪海,隋学艳,等.基于混合像元分解提取小麦种植面积的技术与方法研究[J].山东农业科学,2009(5):12-16.
[19]邓书斌,陈秋锦,杜会建,等. ENVI遥感图像处理方法[M].北京:高等教育出版社,2014.
[20]潘腾.高分二号卫星的技术特点[J].中国航天,2015(1):3-9.
[2]孙九林.中国农作物遥感动态监测与估产总论[M].北京:中国科学技术出版社,1996.
[3]史舟,梁宗正,杨媛媛,等.农业遥感研究现状与展望[J].农业机械学报,2015,46(2):247-260.
[4]游炯,裴志远,王飞,等.基于改进多元纹理信息模型和GF-1影像的县域冬小麦面积提取[J].农业工程学报,2016,32(13):131-139.
[5] YANG Y,HUANGY,TIAN Q,et al. The Extraction model of paddyrice information based on GF-1 satellite WFV images[J]. Spectrosco-py and spectral analysis,2015,35(11):3255-3261.
[6]李耀辉,王金鑫,李颖.应用时间序列EVI的MERSI多光谱混合像元分解[J].遥感学报,2016,20(3):459-467.
[7] LV T T,LIU C. Study on extraction of crop information using time-series MODIS data in the Chao Phraya Basin of Thailand[J]. Ad-vances in space research,2009,45(6).
[8]王凯,赵军,朱国峰,等.基于GF-1遥感数据决策树与混合像元分解模型的冬小麦种植面积早期估算[J].遥感技术与应用,2018,33(1):158-167.
[9]马孟莉.基于HJ卫星混合像元分解的水稻生长监测技术研究[D].南京:南京农业大学,2011.
[10]张永宏,何静,阚希,等.遥感图像道路提取方法综述[J].计算机工程与应用,2018,54(13):1-10,51.
[11]王兴盛.宁夏水稻[J].北方水稻,2007(1):19-22,31.
[12]刘佳,王利民,姚保民,等.基于多时相OLI数据的宁夏大尺度水稻面积遥感估算[J].农业工程学报,2017,33(15):200-209.
[13]郑长春.水稻种植面积遥感信息提取研究[D].乌鲁木齐:新疆农业大学,2008.
[14]童李霞,燕琴,骆成凤,等.基于NDWI分割与面向对象的水体信息提取[J].地理空间信息,2017,15(5):57-60.
[15]郝卫平,梅旭荣,蔡学良,等.基于多时相遥感影像的东北三省作物分布信息提取[J].农业工程学报,2011,27(1):201-207+393.
[16]李志鹏,刘珍环,李正国,等.水稻空间分布遥感提取研究进展与展望[J].中国农业资源与区划,2014,35(6):9-18.
[17]王连喜,徐胜男,李琪,等.基于决策树和混合像元分解的江苏省冬小麦种植面积提取[J].农业工程学报,2016,32(5):182-187.
[18]蔡薇,郭洪海,隋学艳,等.基于混合像元分解提取小麦种植面积的技术与方法研究[J].山东农业科学,2009(5):12-16.
[19]邓书斌,陈秋锦,杜会建,等. ENVI遥感图像处理方法[M].北京:高等教育出版社,2014.
[20]潘腾.高分二号卫星的技术特点[J].中国航天,2015(1):3-9.