【作者】 冯秀丽；

【导师】 王珂；

【作者基本信息】 浙江大学 ， 农业遥感与信息技术， 2006， 博士

【摘要】 土地利用更新调查是政府和各级土地管理部门制定政策和落实各项管理措施的科学依据,对于浙江这样经济快速发展的省份,建设用地需求量大,用地类型转换频繁,及时掌握准确的土地利用现状尤为重要。传统土地利用更新调查通常是在已有土地利用现状数据的基础上,实行人工野外实地调查与测量,更新图件,再数据汇总上报,工作效率低,费时,费力,时效性差;或者采用航测方式,于室内进行土地利用信息矢量数据的提取,再野外调绘和数据汇总,精度高,是经典的普遍认可的土地利用调查方法,但费用昂贵,而且同样费时费力,不可能在较短的周期内重复采用。 利用卫星遥感影像进行土地利用更新调查有着非常明显的优势。首先,卫星遥感影像具有周期性、现势性、客观性和系统性的特点;其次,卫星遥感影像具有反映地面信息丰富、覆盖面积大、实时性强、费用较低(为航空遥感的1/10)等优点:再次,卫星遥感影像的分辨率越来越高,逐步弥补了精度上的不足,而费用却有下降的趋势。土地利用更新调查一般为1:1万尺度,从空间分辨率、经济性、时效性以及商业化程度等多方面综合考虑,SPOT5卫星影像具有较高性价比,其全色及多光谱波段影像空间分辨率分别为2.5m、10m,理论上可满足1:1万更新调查的要求。国土资源部也曾推荐利用SPOT5开展新一轮土地利用基础图件更新,但更新调查的关键技术、技术流程和调查精度缺乏系统的研究。 基于SPOT5数据的1:1万土地利用更新调查的关键技术以及更新调查成果所能达到的更多还原

【Abstract】 Land use survey by remote sensing was able to obtain the land use status timely and accurately, and provide the scientific basis for land management especially in the developed region, such as Zhejiang province, where needs large amount of construction land and the land use changing is frequent. The traditional method of land use updating is performed through field investigation and survey based on land use statue data, whicn is inefficient, time-consuming and untimely. Another classical method is aviation photography, which are too expensive and complicated,and can’t totally meet the demand to be fast and periodical at present.Remote sensing technology appears to be a powerful tool for land use survey according to the basic need of land use database updating and land use dynamic monitoring in China because of its high-resolution, easy-obtaining and economy. The land use survey is generally at 1: 10,000 scale. Compared to other satellite data, SPOT5 image has obvious advantage in spatial resolution of 2.5 m for panchromatic image satisfied with the demand of land use updating survey at 1: 10,000 scale theoretically. The ministry of land and resources of China once recommend to update the land use map using SPOT5image. But unfortunately, few studies have been done on the key technology and the overall presicion. Furthermore, the feasibility of land use survey at 1:10,000 scale based on the SPOT5 image even has still been in debate. In this thesis, the SPOT5 image processing, procedure as well as the evaluating index for the land use survey based on SPOT5 image at 1:10,000 scale was undertaken. A system accuracy analysis for the result of land use survey of 1: 10,000 scale is performed. To guarantee the result of the application to satisfy with the national standards, the methods as well as the technology flow of the application is definitized.The main results were concluded as follow:(1) Effect of piexl resample of SPOT5 image to geometric correction precision The fusion between panchromatic image resampled from 2.5m to 1m andmultispectral image resampled from 10m to 2m using bilinear interpolation approach was able to produce satisfying merged image, in which features and borderlines are propitious to interpretation and position while the resultant area presicion was also improved. The geometric correction presicion of SPOT5 image using different resample methods and piexl size was analysed, and the geometric correction presicion can reach X 1.9m, Y 2.0m, RMS ±2.4m.(2) Enhancing the interpretation for SPOT5 imageThe main method of enhancement after image fusion is contrast enhancement including histogram equalization, linear stretch and piecewise linear stretch. Our results indicated that the effect of piecewise linear stretch based on the analysis of the object spectrum was obvious, but not for histogram equalization. The corresponding breakpoints could be found through analysising of the object spectrum, and then be used for the piecewise linear stretch to the SPOT5 image. As a result, the precision for area can be increased about 4%.(3) Scheme of geometric correction combination for SPOT5 imageThe geometric correction precision of SPOT5 image is obviously influenced by the different geometric correction models and the numbers of GCPs. Under the preconditionof accurate GCPs, satisfied precision can be obtained using quadratic or cubic polynomial model with about 30 well-distributed GCPs in plain area. The geometric correction precisions were X RMS-±2.0 m, Y RMS-±1.8 m, total RMS-±2.8 m. High geometric correction precision of orthorectification can be obtained adopting the polynomial model with the metadata, the accurate DEM and about 20 well-distributed GCPs in hill area. The geometric correction precisions were X RMS-±1.6 m, Y RMS-±1.0 m, total RMS-±1.9 m. (4) Relations among land use types, patch size and patch area precisionThe patch area presicion was influenced by many factors such as land use type, patch size, image temporal, image quantity, land use situation, patch fragmentation, land use degree etc. The results indicated that, on the 1:10,000 scale, the patch area precision of each land use type reached about 90% overall, except some upland patchs with small area. In fact, for some land use types with frequent conversion, such as residence and construction, the patch average area precision could reach about 95%, the maximum of area precision was about 99%. The study also showed that the bigger the patch area was, the higher the precision could be obtained.(5) Relation between different combinations of land use type and patch area precisionThe borderline discerniblity of different land use type combinations was evaluated using the e-band method. The results showed that higher value usually correspond to construction-related land use type combinations, which means the borderline could be more discernible. Therefore, it was reliable to extract the patch information of construction with a high area precision when the^changing was between construction and other land use types. However, the borderline was unsharp when upland borders upon orchard plot, it was difficult to extract the patch information accurately and the area precision was correspondingly low.The key technology for land use survey based on the SPOT5 image on 1:10,000 scale were studied, the innovation or new development of this research were summarized as follows:Piecewise linear stretch was applied according to the spectral analysis of land use types, based on the linear stretch theory. As a result, the image with dramatic spectral difference was produced for interpretation. The spectral difference enhancement for similar objects leaded discernability improvement for different land use type combinations, which was helpful to extract Jand use information and improve the precision of area.The e-band method was adopted to determine the borderline discernability between different land use type combinations. Satisfied correlation could be found between the discernability degree and precision of area. Approviding a method for analyzing precision of land use updating.Based on the data derived from contemporary aerial survey and IKONOS image as true references, the area precision was analyzed. The area precisions for different land use types and different patch size were achieved in the land use survey using SPOT5 image on 1:10000 scale. At the same time, the minimum patch area was determined for information extraction according to certain precision standard.更多还原