• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>ESA living planet symposium >EDGE-PRESERVING DATA ASSIMILATION FOR FIRE MONITORING USING OPTICAL DATA
【24h】

EDGE-PRESERVING DATA ASSIMILATION FOR FIRE MONITORING USING OPTICAL DATA

机译：使用光学数据进行火灾监测的保留边缘数据的同化

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Detecting burned area using data from moderate resolutionsensors is fundamentally a change detection problem:as the fire affects the vegetation, it changes its opticalproperties. This change is detected and assessed todistinguish fire from other changes in the land surface(e.g. flooding, crop harvesting...). Two important challengesneed to be overcome: the variation in reflectanceought to be related just the fire, and thus acquisitiongeometry-derived effects ought to be minimised, and arobust way of deciding whether a fire took place by consideringthe spectral pre- and post-fire reflectance needsto be in place. In this contribution, we propose a firedetection approach based on (i) a more advanced signaltracking method using edge-preserving data assimilationtechniques to fit linear kernel models to the observationsand (ii) an interpretation of the change signal using aspectral linear model.For signal tracking, we use linear kernel models to interpretthe surface reflectance observations. The kernelweights are inferred using a 4DVAR DA system with aweak constraint. We implement regularisation in a waythat is ”edge-preserving”, i.e., abrupt changes in the signalare not smoothed over, resulting in adequate regularisationin the pre- and post-fire periods, but no regularisationover the fire itself. The proposed method is efficientlyimplemented as an iterative linear problem, andultimately produces estimates of pre- and post-fire reflectancefor a given geometry.To interpret the change in reflectance caused by the fire,we assume that fire causes defoliation (hence exposedsoils) and a combination of char and ash. We use a simplelinear mixture model where the post-fire signal is assumedto be a mixture of the pre-fire reflectance plus atypical burn signal, typical a mixture of dry soils, charand ash. This burned signal can be spectrally modelledby a constrained quadratic function. The proposed modelis also linear.Using MODIS data, we illustrate the proposed method.The method is not dependent on sensor type, and wereview its usefulness, opportunities for inter-sensor datablending, and practical complications and limitations.
机译：使用来自中等分辨率传感器的数据检测烧毁区域从根本上来说是一个变化检测问题：由于火势影响植被，因此会改变其光学特性。可以检测到这种变化并进行评估，以将火与陆地表面的其他变化区分开来（例如洪水，农作物收成...）。需要克服两个重要的挑战：反射率的变化应仅与火相关，因此应将采集几何衍生的影响降至最低，以及通过考虑火前和后的光谱反射率来确定是否发生火的可靠方法。就位。在这项贡献中，我们提出了一种基于（i）使用边缘保留数据同化技术的更先进的信号跟踪方法以将线性核模型拟合到观测结果的火探测方法，以及（ii）使用纵横线性模型对变化信号的解释。 ，我们使用线性核模型来解释表面反射率观测值。使用具有弱约束的4DVAR DA系统推断内核权重。我们以“保留边缘”的方式实施正则化，即信号的突变不会被平滑，从而导致火灾前和后时期有足够的正则化，但没有针对火灾本身进行正则化。所提出的方法可以有效地实现为迭代线性问题，并最终得出给定几何形状的火灾前和火灾后反射率的估计值。为了解释火灾引起的反射率变化，我们假设火灾会引起落叶（因此暴露在土壤中）及其组合炭和灰。我们使用一个简单的线性混合模型，其中假定后燃信号是前燃反射率与非典型燃烧信号的混合，通常是干燥土壤，炭和灰的混合。可以通过约束二次函数在光谱上模拟此刻录信号。所提出的模型也是线性的。利用MODIS数据，我们对所提出的方法进行了说明。该方法不依赖于传感器类型，并考察了其有用性，传感器间数据融合的机会以及实际的复杂性和局限性。

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