Disturbance Detection and Diagnostics (D3) Algorithm

This study examines the effectiveness of using the Normalized Difference Vegetation Index (NDVI) derived from 1326 different Landsat Thematic Mapper and Enhanced Thematic Mapper images in finding low density development within the Commonwealth of Virginia’s forests. Individual NDVI images were stacked by year for the years 1995–2011 and the yearly maximum for each pixel was extracted, resulting in a 17-year image stack of all yearly maxima (a 98.7% data reduction). Using location data from housing starts and well permits, known previously forested housing starts were isolated from all other forest disturbance types. Samples from development disturbances and other forest disturbances, as well as from undisturbed forest, were used to derive vegetation index thresholds enabling separation of disturbed forest from undisturbed forest. Disturbances, once identified, could be separated into Development Disturbances and Non-Development Disturbances using a classification tree and only two variables from the Disturbance Detection and Diagnostics (D3) algorithm: the maximum NDVI in the available recovery period and the slope between the NDVI value at the time of the disturbance and the maximum NDVI in the available recovery period. Low density development disturbances of previous forest land cover had an F-measure, combining precision and recall into a single class-specific accuracy (β = 1), of 0.663. We compared our results to the NLCD 2001–2011 land cover changes from any forest (classes 41, 42, 43, and 90) to any developed (classes 21, 22, 23, and 24), resulting in an F-measure of 0.00 for the same validation points. Landsat time series stacks thus show promise for identifying even the small changes associated with low density development that have been historically overlooked/underestimated by prior mapping efforts. However, further research is needed to ensure that (1) the approach will work in other forest biomes and (2) enabling detection of these important, but spatially and spectrally subtle, disturbances still ensures accurate detection of other forest disturbances.

本研究旨在探讨利用从1326幅不同Landsat主题制图仪和增强型主题制图仪图像中提取的归一化植被指数（NDVI）在弗吉尼亚州联邦森林内寻找低密度开发情况的有效性。针对1995至2011年间的数据，将单个NDVI图像按年度堆叠，并提取每个像素的年度最大值，从而形成一个17年的年度最大值图像堆叠（数据量减少了98.7%）。通过利用住房开工和井许可的位置数据，将已知的先前森林地区的住房开工与其他森林干扰类型区分开来。从开发干扰、其他森林干扰以及未受干扰的森林中选取样本，以推导出植被指数阈值，从而实现受干扰森林与未受干扰森林的分离。一旦识别出干扰，可以使用分类树将其分为开发干扰和非开发干扰，仅使用来自干扰检测与诊断（D3）算法的两个变量：可用恢复期内的最大NDVI以及干扰发生时的NDVI值与可用恢复期内的最大NDVI之间的斜率。先前森林土地覆盖的低密度开发干扰的F度量，结合精确率和召回率成为一个单一类别的特定准确率（β = 1），达到了0.663。我们将我们的结果与2001至2011年NLCD森林土地覆盖变化（任何森林类别41、42、43和90）到任何开发类别（21、22、23和24）进行了比较，在相同的验证点上，F度量达到了0.00。因此，Landsat时间序列堆叠在识别与低密度开发相关的小幅变化方面显示出巨大潜力，这些变化在以往的制图工作中已被忽视或低估。然而，为了确保（1）该方法适用于其他森林生态系统，以及（2）检测这些重要但空间和光谱上微妙的变化仍能保证准确检测其他森林干扰，仍需进行进一步的研究。