基于NDVI时间序列的黄河中下游典型史前都邑性遗址的作物物候观察

发布时间：2019-01-01 19:15

【摘要】：一些浅埋藏在地下的考古遗迹在高空间分辨率遥感影像上可以表现为可见的作物标记。但是由于高空间分辨率数据具有价格昂贵,其可获取的数据的空间和时间具有非常高的限制。而另一方面,Landsat影像作为可以自由获取的中分辨率图像,其生成的时间序列可以用来检测作物生长变化,进一步可以在大面积中显示微小的表面作物标记。同时,MODIS数据由于其粗糙的空间分辨率使其无法用于检测微妙细小的地表特征,因此很少被用于考古勘探,但是MODIS的高时间分辨率的优势使其可以提供大量的作物物候生长信息。为弥补这个缺陷,选择具有大面积特征的典型的史前都邑性遗址陶寺和二里头作为研究对象,采用MODIS NDVI时间序列数据观察遗址区作物的物候生长规律并检测作物在各个物候期的长势差异。计算230期覆盖10年范围的MODIS NDVI变化率并通过皮尔逊相关系数进行分析,并采用最大最小距离来测量考古遗址内作物生长的相似性。另一方面,在遗址内部选择埋藏遗迹集中区(如宫殿区和祭祀区)作为研究区,而以其周围的遗迹分布密度较低的空白区域作为参考区。对小麦和玉米两年生长期的Landsat影像进行皮尔逊相关系数和欧几里德距离计算并分析。使用单向Hausdorff距离计算对照遗迹集中区及其对应空白区之间的作物生长相似性或差异性以获取监测和检测作物生长异常的最佳时间窗口。根据作物生长的物候学,生成冬小麦和夏玉米的时间序列NDVI图像,分析考古遗址中的作物异常现象。结果表明,两种作物对水分需求的关键阶段,最大最小值和单向Hausdorff距离都比较高,而这些阶段的Landsat NDVI图像显示出了由埋藏的考古遗迹引起的作物异常。这表明对NDVI时间序列图像进行Hausdorff距离计算可用于检测作物异常。同时也表明了作物关键需水期的Landsat NDVI图像可以提供一定的地下埋藏信息,而MODIS也可以提供有用的作物生长信息在进行田野发掘之前。
[Abstract]:Some shallow buried archaeological sites can be seen as crop markers on high spatial resolution remote sensing images. However, because of the high spatial resolution data is expensive, its available data space and time have very high limit. On the other hand, as a freely available medium resolution image, Landsat images can be used to detect crop growth and display small surface crop markers in large areas. At the same time, MODIS data are rarely used in archaeological exploration because their rough spatial resolution prevents them from being used to detect subtle and small surface features. But MODIS's advantage of high temporal resolution allows it to provide a large amount of crop phenological growth information. In order to make up for this defect, the typical prehistoric capital Yinji Temple and Erlitou site with large area characteristics were selected as the objects of study. MODIS NDVI time series data were used to observe the phenological growth of crops in the site area and to detect the growth difference of crops in different phenological periods. The variation rate of MODIS NDVI covering 10 years was calculated and analyzed by Pearson correlation coefficient, and the similarity of crop growth in archaeological sites was measured by the maximum and minimum distance. On the other hand, the concentrated area of buried relics (such as palace area and sacrificial area) is chosen as the research area within the site, while the blank area with low distribution density around it is used as the reference area. Pearson correlation coefficient and Euclidean distance were calculated and analyzed in Landsat images of wheat and maize. Using one-way Hausdorff distance to calculate the similarity or difference of crop growth between the control relic concentration area and its corresponding blank area to obtain the best time window for monitoring and detecting crop growth anomaly. According to the phenology of crop growth, the time series NDVI images of winter wheat and summer corn were generated, and the abnormal crop phenomena in archaeological sites were analyzed. The results show that the maximum minimum and one-way Hausdorff distance of the two crops are relatively high in the critical stages of water demand. The Landsat NDVI images of these stages show crop anomalies caused by buried archaeological relics. This indicates that the Hausdorff distance calculation of NDVI time series images can be used to detect crop anomalies. At the same time, it is shown that the Landsat NDVI images of the critical water demand period of crops can provide some underground burial information, and MODIS can also provide useful crop growth information before the field excavation.
【学位授予单位】：中国科学院大学(中国科学院遥感与数字地球研究所)
【学位级别】：博士
【学位授予年份】：2017
【分类号】：S-09

【相似文献】