基于多源卫星遥感的湖北省作物耕作面积提取及其动态监测

发布时间：2018-04-24 05:29

本文选题：湖北省 + 油菜　；参考：《长江大学》2015年硕士论文

【摘要】：农作物的种植面积作为国家农情基础数据是农作物估产的必要参数,同时也是国家的粮食安全重要指标。农作物面积获得的主要方法是大面积实地测量调查或者是统计部门逐级上报,但是这种方法耗时耗力而且缺乏空间分布信息。“3S”技术的快速发展为监测农作物种植面积信息提供了一个实用高效的科技手段。实时准确的获取农作物的种植范围是粮食生产和粮食安全的重要保障。由于遥感技术的客观性、空间性、时效性、内容丰富和成本相对较低的优势；与其他方法相比利用遥感技术可以获得不同时间分辨率和空间分辨率下的大规模的地表信息,它在作物面积监测方面具有无可比拟的优势。本文的研究区湖北省是国家的农业大省,其耕地面积占到我国国土面积的20%左右,而全省70%的耕地主要集中分布在江汉平原、鄂东沿江汉平原及鄂中丘陵地区；兼有水田旱地,且水田稍多于旱地,在鄂南水田占耕地的70%左右,普遍栽培双季稻；而在鄂北和鄂西北,旱地占耕地的68.3%,主要作物是油菜、小麦、玉米、水稻等；及时的掌握农作物的种植面积和产量信息,对强化农业生产的管理、调整农业结构、辅助政府有关部门制定科学合理的农业政策具有重要意义。本论文对研究的背景及意义、国内外近年来的研究进展以及拟采用的研究方法进行了详细的介绍。通过总结前人在相关的研究中已经取得的成果和存在的相关问题,提出了选题研究的必要性和可行性。农作物种植面积提取的关键是准确地识别作物类型,但因高分辨率多光谱遥感影像重访周期长,且数据获取容易受到天气影响,基本上不可能得到农作物的实际种植面积,高额的成本,导致了高分辨率数据的使用是困难的；所以本文选用中分辨率成像仪(MODIS)、环境减灾小卫星(HJ-1-A/B)、高分一号等遥感卫星数据同时结合所研究农作物的物候以及其他的辅助数据采用一种方法来大规模的进行农作物(主要是油菜)种植面积提取研究；并分析农作物在2005年-2014年这10年期间油菜种植面积的动态变化。详细介绍了具体的研究内容、方法和结果。(1) MODIS的植被指数(MODIS-NDVI)时间序列数据可以持续反应植被的覆盖状况,是农作物遥感监测的重要数据源。为研究基于MODIS数据的油菜种植分布信息提取技术,选取湖北省为研究区,利用2009-2010年15个时相的MODIS-NDVI时序数据,结合农作物的物候和地面调查样本等辅助数据,通过建立油菜种植面积的提取模型,采用多次阈值比较方法提取了2010年湖北省油菜种植分布信息,总体提取精度为85%左右。最后利用环境小卫星HJ-1A CCD数据进行精度验证,证明了MODIS-NDVI时序数据及本文方法在油菜种植面积提取中的可靠性。及时掌握油菜种植面积对加强农业的生产管理、调整农业的结构以及辅助政府相关部门制定科学合理的农业政策具有重要意义。(2)研究提取了2005年～2014年十年期间湖北省油菜种植面积,并分析其变化率和动态度。研究结果表明：2005年～2007年这三年湖北省油菜种植面积逐年递减,从2008年至今油菜种植面积有稳步增加的趋势,从整体来看这十年的油菜种植面积是增加的。(3)利用MODIS-EVI产品数据提取了湖北省2010年的冬小麦种植分布信息。通过从MODIS数据产品MOD13Q1提取的MODIS-EVI数据,建立了整个冬小麦生育期内的湖北省的MODIS-EVI时序数据；利用16m分辨率的高分一号卫星影像通过监督分类得到天门市冬小麦的种植分布信息影像,用得到的参考区冬小麦影像对湖北省MODIS-EVI时序数据进行掩膜处理得到湖北省冬小麦参考MODIS-EVI数据；最后利用MODIS-EVI数据采用混合像元分解法和波谱分析法提取整个湖北省的冬小麦种植分布信息。将湖北省10个县市的冬小麦遥感监测面积与统计冬小麦面积进行相关性分析,R2=0.85。(4)利用MODIS反射率数据进一步计算与植被、水体、土壤相关的各种遥感指数,例如归一化植被指数(NDVI),归一化水指数(NDWI),归一化土壤指数(NDSI).将生长季(第113天～273天)的各个指数的不同时序数据叠加后进行主成分分析,获得每一种植被指数的第一主成分,然后将三种指数的第一主成分叠加,将叠加的图利用SVM分类获得旱地和水田的信息,最后在旱田的基础上利用玉米和大豆特殊时期的差异区分玉米和大豆,从而得到湖北省2014年精细作物分类状况。
[Abstract]:As the basic data of national agricultural situation, the planting area of crops is a necessary parameter for crop yield estimation, and it is also an important index of national food security. The main method of crop area acquisition is large area survey survey or statistical department report by level, but this method is time-consuming and lack of spatial distribution information. "3 The rapid development of S technology provides a practical and efficient technology for monitoring crop planting area information. Real time and accurate harvest of crops is an important guarantee for grain production and food security. The advantages of remote sensing technology are objective, spatiality, timeliness, rich inside capacity and relatively low cost; Compared with remote sensing technology, the method can obtain large scale surface information with different time resolution and spatial resolution. It has unparalleled advantage in crop area monitoring. In this paper, Hubei province is a major agricultural province of the country. Its cultivated area accounts for about 20% of the land area in China, and 70% of the cultivated land in the province. Mainly concentrated in Jianghan Plain, east of Hubei Province along Jianghan Plain and hilly area of central Hubei Province, there are dry land in the paddy field, and the paddy field is slightly more than dry land. In the southern Hubei Province, about 70% of the cultivated land is cultivated. In Northern Hubei and northwest of Hubei, 68.3% of the cultivated land, the main crops are rape, wheat, corn, rice and so on. The planting area and yield information of crops is of great significance to strengthening the management of agricultural production, adjusting the agricultural structure and assisting the relevant departments of the government to formulate scientific and rational agricultural policies. This paper gives a detailed introduction to the background and significance of the research, the research progress in recent years at home and abroad and the research methods used to be used. The key to the extraction of crop planting area is to identify the type of crop accurately, but because the high resolution multi spectral remote sensing image revisits a long period, and the data acquisition is easily affected by the weather, it is basically impossible. It is difficult to use the actual planting area of crops and high cost, which leads to the use of high resolution data. Therefore, this paper uses the medium resolution imager (MODIS), the environmental disaster reduction satellite (HJ-1-A/B), the high score number one and other remote sensing satellite data, which combines the phenology of the research crops and other auxiliary data. Methods to study crop area extraction on a large scale (mainly rapeseed), and analyze the dynamic changes of crop planting area during the 10 years of -2014 in 2005. Detailed research contents, methods and results were introduced in detail. (1) the vegetation index (MODIS-NDVI) time sequence data of MODIS can continue to respond to vegetation. The coverage situation is an important data source for crop remote sensing monitoring. In order to study the extraction technology of rape planting distribution based on MODIS data, Hubei province is selected as the research area, and the MODIS-NDVI time series data of 2009-2010 years and 15 phases are used, and the auxiliary data of Crop Phenology and ground investigation sample are combined to establish the planting area of rape. Extraction model and multiple threshold comparison method were used to extract the distribution information of rape planting in Hubei Province in 2010. The overall extraction precision was about 85%. Finally, the accuracy of the environmental small satellite HJ-1A CCD data was used to verify the reliability of the MODIS-NDVI time series data and the method of this method in the extraction of rape planting surface. The area is of great significance to strengthen the production management of agriculture, adjust the structure of agriculture and assist the relevant government departments to formulate scientific and rational agricultural policies. (2) the area of rape planting in Hubei Province during the period of ten years from 2005 to 2014 was extracted, and its change rate and dynamic degree were analyzed. The results showed that the three year lake from 2005 to 2007 was the result of the study. The planting area of rape in northern province has decreased year by year. From 2008 to now, the planting area of rape has been steadily increasing. In the whole ten years, the planting area of rape is increasing. (3) the distribution information of winter wheat planting in Hubei province in 2010 was extracted by using MODIS-EVI product data. The MODIS-EVI data extracted from MOD13Q1 data products, The MODIS-EVI time series data of Hubei Province during the whole winter wheat growth period were established, and the image of the planting distribution information of Winter Wheat in Tianmen was obtained through supervised classification by the high score satellite image of 16m resolution, and the winter wheat image of the reference area was used to mask the MODIS-EVI time sequence data of Hubei province to get the winter small of Hubei province. Wheat reference MODIS-EVI data; finally, using MODIS-EVI data to extract the distribution information of winter wheat planting in Hubei province by using the mixed pixel decomposition method and spectral analysis method. The correlation analysis between the winter wheat monitoring area of the 10 counties and the statistical winter wheat area in the 10 counties of Hubei province is analyzed, and the R2= 0.85. (4) uses the MODIS reflectance data to further calculate the data. Various remote sensing indices related to vegetation, water and soil, such as the normalized vegetation index (NDVI), normalized water index (NDWI), and normalized soil index (NDSI), are added to the principal component analysis of each index of the growing season (113rd days to 273 days) to obtain the first principal component of each planting index, and then three The first principal component of the species index is superimposed, and the superimposed map is used to obtain the information of the dry land and the paddy field by SVM classification. Finally, corn and soybean are distinguished from the special period of maize and soybean on the basis of the dryland, thus the classification of fine crops in Hubei Province in 2014 is obtained.

【学位授予单位】：长江大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：S127;S31

【参考文献】