• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Remote Sensing and Photogrammetry Society annual conference 2011 >A multispectral multiplatform based change detection tool for vegetation disturbance on Irish peatlands
【24h】

A multispectral multiplatform based change detection tool for vegetation disturbance on Irish peatlands

机译：基于多光谱多平台的爱尔兰泥炭地植被扰动变化检测工具

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

In Ireland, maintenance of peatland Carbon (C) stock has taken on a renewed importance with articles 3.3 and 3.4 of the Kyoto Protocol / Marrakech Accords recognising soil organic carbon as a biosphere sink. Peatlands in Ireland cover approximately 20% of the land surface and accounts for between 53% (1071.13 Mt C) to 62% (1503 Mt C) of national soil carbon stock. Extensive anthropogenic disturbance to peatland habitats can have negative impacts on sequestration rates, and in extreme cases, can convert such habitats from net sinks to a net source's of C emissions. Irish peatlands are spatially extensive and relatively inaccessible, therefore satellite based multispectral imagery is ideally suited to the monitoring peatland vegetation due to its high spatial and temporal resolution. However Ireland's extensive cloud cover means that over 78% of all optical based satellite imagery taken throughout the year is completely obscured. An alternative approach in a change detection study that requires a high temporal resolution is to use multiplatform data. In this study satellite data from six different multispectral sensors (TM, SPOT 2, 4 and 5, IRS P6, Aster VNIR) were used in a change detection study of vegetation disturbance on Irish peatlands. All data was corrected for atmospheric scattering using dark object subtraction and converted from digital numbers to Enhanced Vegetation Index 2 (EVI2). Radiometric normalisation was performed using Temporally Invariant Clusters (TIC) and cross calibration applied using linear regression of radiometrically stable ground-based targets. Erdas Imagine's Spatial Modeller was used to create a change detection model using pixel-to-pixel based subtraction with a Standard Deviation (SD) threshold. The model can be adjusted by altering the SD threshold as well as a pixel based clump function which was used to mask isolated pixels that may have occur due to local anomalies or mis-registration. Initial results after TIC normalisation show EVI2 cross platform correlation values (R2), regressed against a Landsat TM master image, of 0.9346 for Aster VNIR, 0.9487 for IRS P6, 0.9246 for SPOT 4 and 0.9641 for SPOT 5. Change detection analysis of prior and post cross calibrated data revealed an average decrease of 15.4% in the area of detected change which was attributed to spectral and radiometric variability across platforms as opposed to actually vegetation change on the ground. Auxiliary data detailing type, date and extent of know disturbance events (e.g. burning, bog burst, afforestation, peat harvesting) is currently being used to train and validate the change detection model to ensure that SD and clump adjustment parameters are set to accurately detect vegetation disturbance on the ground. It is hoped that this change detection model, combined with other models which have been created as part of this research for image preprocessing, will all be combined to form a tool/protocol for the quantification of vegetation disturbance on Irish peatlands.
机译：在爱尔兰，随着《京都议定书》 /《马拉喀什协议》第3.3和3.4条的认可，泥炭地碳（C）储量已变得具有新的重要性，承认土壤有机碳是生物圈汇。爱尔兰的泥炭地覆盖了大约20％的土地表面，占全国土壤碳储量的53％（1071.13 Mt C）至62％（1503 Mt C）。对泥炭地生境的广泛人为干扰可能会对固存率产生负面影响，在极端情况下，可能会将此类生境从净汇转换为净碳排放源。爱尔兰泥炭地空间广泛，相对难以到达，因此基于卫星的多光谱图像由于其高时空分辨率而非常适合监测泥炭地植被。但是，爱尔兰广泛的云层覆盖意味着全年所拍摄的所有基于光学的卫星图像中有78％以上被完全遮盖。需要高时间分辨率的变更检测研究中的另一种方法是使用多平台数据。在这项研究中，来自六个不同的多光谱传感器（TM，SPOT 2、4和5，IRS P6，Aster VNIR）的卫星数据被用于爱尔兰泥炭地植被扰动的变化检测研究中。使用暗物减法校正了所有数据的大气散射，并将其从数字转换为增强植被指数2（EVI2）。使用临时不变聚类（TIC）进行辐射归一化，并使用辐射稳定的地面目标的线性回归进行交叉校准。 Erdas Imagine的Spatial Modeller用于创建变化检测模型，该模型使用具有标准偏差（SD）阈值的逐像素减法来创建。可以通过更改SD阈值以及基于像素的聚集函数（用于掩盖可能由于局部异常或配准错误而发生的孤立像素）来调整模型。 TIC归一化后的初步结果表明，与Landsat TM主图像相对的EVI2跨平台相关值（R2）对于Aster VNIR为0.9346，对于IRS P6为0.9487，对于SPOT 4为0.9246，对于SPOT 5为0.9641。交叉校准后的数据显示，检测到的变化区域平均下降了15.4％，这归因于平台之间的光谱和辐射变化，而不是地面上实际的植被变化。当前正在使用辅助数据详细信息类型，日期和已知干扰事件的程度（例如，燃烧，沼泽爆发，绿化，泥炭收割）来训练和验证变化检测模型，以确保将SD和丛集调整参数设置为可准确检测植被地面上的干扰。希望将这种变化检测模型与本研究中为图像预处理而创建的其他模型相结合，以形成量化量化爱尔兰泥炭地植被扰动的工具/协议。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号