基于Stacked ConvLSTM的时间序列森林火烧迹地检测

确定森林火烧迹地的准确时间点以及空间范围对于森林的受损评价、管理、碳核算以及森林恢复的管理有重要意义。由于森林火烧迹地在空间分布上具有一定的连续性,现有的森林火烧迹地提取方法大都采用先分类再后处理的两步处理策略来抑制虚警像素的影响。本文提出将时空检测方法Stacked ConvLSTM用于时间序列森林火烧迹地的检测,在保持结果具有较好空间连续性的基础上避免了具有主观性的后处理操作,实现端到端提取森林火烧迹地信息,提升了森林火烧迹地的提取精度。采用MODIS时间序列数据,基于2001年—2008年以及2001年—2016年的黑龙江沾河林业局伊南河林场和内蒙古自治区毕拉河林业局北大河林场两个区域的历史时间序列,分别对这两个区域2009年以及2017年发生的特大火灾区域进行火烧迹地检测,利用Stacked ConvLSTM、Stacked LSTM以及bfast算法在两个区域的MODIS时间序列中提取森林火烧迹地,并将火烧迹地检测结果与ESA发布的Fire_CCI 5.1火烧迹地产品进行对比分析。结果表明：首先,从目视效果来看,在研究区域Ⅰ,Stacked ConvLSTM检测的结果比Stacked LSTM和bfast算法错误检测点少,并且在空间分布也保持较高连续性;在研究区域Ⅱ,Stacked ConvLSTM检测到了较完整的火烧迹地区域。其次,在定量的精度评价指标上,在研究区域Ⅰ,Stacked ConvLSTM的精确度比Stacked LSTM和bfast算法分别高出0.120和0.405,并且召回率、准确度和F1-score也更高,Fire_CCI 5.1召回率虽更高,由于错检区域较大,其他精度指标远低于Stacked ConvLSTM;在研究区域Ⅱ,Stacked ConvLSTM精确度达0.924,召回率、准确度和F1-score相比Stacked LSTM和bfast算法以及Fire_CCI 5.1更高。     

基于多源遥感数据的森林火灾监测研究

以2019年5月13日黑龙江省佳木斯市同江市洪河国家级自然保护区为研究对象,基于静止气象卫星Himawari-8(H8)和极轨气象卫星MODIS数据对火灾发生经过进行还原,基于高分一号(GF-1)和高分六号(GF-6)数据进行火烧迹地识别.试验结果表明,Himawari-8及MODIS数据都可较好地反演出火灾发生地点,但基于固定阈值算法对Himawari-8数据进行火灾反演的实效性更强,可实现10 min级监测,可以弥补MODIS卫星时间分辨率低的不足,其可以对林业火灾监测提供较好的技术支撑.基于GEMI模型进行的火烧迹地提取可以为灾情损失及灾后重建提供本地数据.     

基于MODIS影像的森林火灾火线检测方法

结合归一化火灾差异指数NDBR(normalized difference burn ratio)和MODIS多波段影像梯度边缘分析手段检测火线, 应用B样条函数拟合火线并确定火势蔓延方向。为对比验证, 基于火线的Kriging插值实现火灾外推预测, 与30min后的火灾参考数据目视对比与统计：火线的预测变化与参考影像基本保持一致, 火灾外推影像的均值和熵约为参考影像的86%和81%, 火迹地检测的Kappa系数达80.2%。试验表明, 提出的森林火线特征自动检测方法在动态火灾监测中是可行、有效的。     

基于中分辨率成像光谱仪数据的建筑指数研究

中分辨率成像光谱仪(MODIS)数据是研究大尺度土地利用/土地覆盖的有效数据手段。针对现有建设用地提取方法的不足,文章通过分析MODIS数据的地物光谱特征,给出适合MODIS数据的归一化建筑指数,并以京津冀城市群为研究区提取建设用地,最后从位置和面积两方面对提取结果进行精度验证。结果表明:该文构建的归一化建筑指数不但加强了MODIS数据中的建设用地,还成功抑制了裸地、荒地和已收割耕地等对建设用地提取的干扰。     

用于时间序列聚类分析的小波变换和特征量提取方法

利用小波变换的多尺度分辨功能,对时间序列数据进行多次Haar小波变换,将时间序列分解为尺度分量和细节分量;通过保留尺度分量,提取时间序列的趋势信息,并结合统计特征量计算,将时间序列的趋势信息与几种统计特征量组合在一起,构成SOM神经网络的输入向量,对时间序列进行聚类分析。通过在几类模拟时间序列数据上进行实验分析,取得了较好的实验效果。并将此方法应用到基于MODIS遥感影像的林地植被提取中,获得了较高的提取精度。     

多时相MODIS指数提取东北三省水稻种植面积

为准确提取水稻面积,以东北为研究区域,采用多时相16d合成MODIS增强型植被指数数据和8d合成MODIS地表反射率数据提取水稻种植分布。选取水稻代表样点利用IDL编程提取物候曲线,利用归一化植被指数(NDVI)将水稻与其他明显地类区分,然后建立水稻增强型植被指数(EVI)、地表水体指数(LSWI)之间的相关关系,结合最新2015年土地利用数据提取东北三省2015年水稻种植面积。同时运用运筹学理论建立省级尺度水稻判别条件最优化模型,分析其在空间分布上的差异性和相关性,并将结果与统计年鉴进行对比分析,分析表明MODIS数据适合大区域省级范围水稻面积的提取,精度可达90%以上。由此得出,MODIS数据在省级尺度提取水稻种植面积上有着较大的优势。     

基于MODIS EVI时序数据的江汉平原油菜种植分布信息提取

油菜是我国主要的食用油料作物。及时、准确地获取油菜种植分布信息对油菜长势监测、估产以及灾情评估具有十分重要的意义。以江汉平原为研究区,利用250 m空间分辨率的MODIS EVI时序数据,以TM数据作为野外采样数据与MODIS EVI数据之间的过渡数据,间接提取MODIS EVI数据农作物的训练样本;通过分析江汉平原油菜和冬小麦的EVI光谱特征及物候信息,建立油菜种植面积提取模型;采用多次阈值比较法提取2014—2015年间江汉平原油菜种植分布信息。研究结果表明,2014年和2015年油菜面积遥感提取结果与农业局统计数据相比,总体提取精度分别达到95.22%和91.29%;2014年MODIS数据与TM数据提取的油菜面积一致性为88.61%;基于时间序列MODIS EVI数据,结合EVI光谱特征和物候信息,利用该方法可以有效提取江汉平原油菜种植分布信息。     

自动赋初值的LSB-Snake模型的道路半自动提取方法研究

LSB-Snake模型是一个半自动提取线状地物的有效方法,但是在提取道路时需要人工输入道路特征（与背景相比道路的亮暗、宽度）作为初始值,在初始种子点稀少的时候不够稳健,且降低了LSB-Snake模型的效率。因此,本文提出了自动赋初值的LSB-Snake模型,采用自适应模板匹配方法为LSB-Snake模型提供初始特征,同时在初始种子点的基础上自动生成可靠、密集的点,共同作为LSB-Snake模型的种子点参与道路提取。对IKONOS和QuickBird进行道路提取的试验证明：在给定同样数目的初始种子点的情况下,提出的自动赋初值LSB-Snake模型比LSB-Snake模型更加稳健;由于无需人工输入道路特征,自动化程度更高;在道路附近地物阴影、遮挡较多的情况下,抗干扰能力更强。     

车载激光扫描数据中高速公路路面点滤波

提出了一种从车载激光扫描数据中自动提取路面的方法。通过分析车载激光扫描点云的空间特征,提出运用近似平面约束法、有序最小二乘坡度估计法和多尺度窗口迭代分析法进行初始路面种子点提取;然后基于局部坡度滤波方法提取所有的路面点;最后选择两组实际点云数据进行实验。结果表明,该方法能快速准确地提取高速公路路面点云,实验数据的提取准确率为95.74%,完整率为98.11%。     

一种利用点云邻域信息的建筑物屋顶面高精度自动提取方法

从LiDAR数据中高精度地提取建筑物屋顶面是构建屋顶面拓扑关系、实现建筑物三维模型重建的关键。本文针对现有算法提取复杂建筑物屋顶面适应性较差、精度较低等问题,提出了一种利用点云邻域信息的建筑物屋顶面高精度自动提取方法。通过主成分分析计算点云特征,构建特征直方图,选取可靠种子点;利用提出的局部点云法向量分布密度聚类算法聚类种子点,快速准确地提取初始屋顶面片;构建基于邻域信息的投票模型,有效地解决屋顶面竞争现象。试验结果表明,本文方法可自动、高精度地提取屋顶面,对不同复杂程度的建筑物具有较好的适应性,能为建筑物三维模型重建提供可靠的屋顶面信息。     

A comparative analysis of a fixed thresholding vs. a classification tree approach for operational burn scar detection and mapping

The scope of this paper is to demonstrate, evaluate and compare two burn scar mapping (BSM) approaches developed and applied operationally in the framework of the RISK-EOS service element project within the Global Monitoring for Environment and Security (GMES) program funded by ESA (http://www.risk-eos.com). The first method is the BSM_NOA, a fixed thresholding method using a set of specifically designed and combined image enhancements, whilst the second one is the BSM_ITF, a decision tree classification approach based on a wide range of biophysical parameters. The two methods were deployed and compared in the framework of operational mapping conditions set by RISK-EOS standards, based either on sets of uni- or multi-temporal satellite images acquired by Landsat 5 TM and SPOT 4 HRV. The evaluation of the performance of the two methods showed that either in uni- or multi-temporal acquisition mode, the two methods reach high detection capability rates ranging from 80% to 91%. At the same time, the minimum burnt area detected was of 0.9–1.0 ha, despite the coarser spatial resolution of Landsat 5 TM sensor. Among the advantages of the satellite-based approaches compared to conventional burn scar mapping, are cost-efficiency, repeatability, flexibility, and high spatial and thematic accuracy from local to country level. Following the catastrophic fire season of 2007, burn scar maps were generated using BSM_NOA for the entirety of Greece and BSM_ITF for south France in the framework of the RISK-EOS/GMES Services Element project.     

Modeling of multi-strata forest fire severity using Landsat TM Data

Most of fire severity studies use field measures of composite burn index (CBI) to represent forest fire severity and fit the relationships between CBI and Landsat imagery derived differenced normalized burn ratio (dNBR) to predict and map fire severity at unsampled locations. However, less attention has been paid on the multi-strata forest fire severity, which represents fire activities and ecological responses at different forest layers. In this study, using field measured fire severity across five forest strata of dominant tree, intermediate-sized tree, shrub, herb, substrate layers, and the aggregated measure of CBI as response variables, we fit statistical models with predictors of Landsat TM bands, Landsat derived NBR or dNBR, image differencing, and image ratioing data. We model multi-strata forest fire in the historical recorded largest wildfire in California, the Big Sur Basin Complex fire. We explore the potential contributions of the post-fire Landsat bands, image differencing, image ratioing to fire severity modeling and compare with the widely used NBR and dNBR. Models using combinations of post-fire Landsat bands perform much better than NBR, dNBR, image differencing, and image ratioing. We predict and map multi-strata forest fire severity across the whole Big Sur fire areas, and find that the overall measure CBI is not optimal to represent multi-strata forest fire severity.     

Correcting the Fire Scar Perimeter of a 1983 Wildfire Using USGS-Archived Landsat Satellite Data

In July 1983, a large wildfire occurred on the island of Karpathos in Greece. However, only a general sketch of the burn perimeter was available and this lacked detailed spatial information, particularly for unburned patches within the fire scar perimeter. A study was undertaken to correctly map the area burned using USGS-archived Landsat data by applying several digital image processing techniques. This paper summarizes and discusses the main findings of that study and provides some general recommendations on the use of remote sensing and archived Landsat data for reconstructing fire history. Remote sensing along with geographic information systems can provide an excellent framework for fast, reliable data capture, measurement, and synthesis, all of which are essential for thorough eco-environmental analysis. Satellite data of multiple types offer an unlimited source of information due to their rich spectral and spatial information content. Satellite mapping of burned areas is considered a standard technique in creating maps of fire scars at multiple scales as a function of the satellite sensor's geometric resolution.     

Influences of forest roads and their edge effects on the spatial pattern of burn severity

Previous research has shown that forest roads are an important feature in many landscapes and have significant effects on wildfire ignition and cessation. However, forest road effects on burn severity have not been studied at the landscape level. Therefore, the overarching goal of our study is to identify the influences of road edge effects on the spatial patterns of burn severity. We analyzed six fires within the Okanogan–Wenatchee National Forest on the eastern slope of the Cascades mountain range of central Washington.We generated two categories for assessing road variables: (1) Primary Road Effect Zone (area within 150 m of the nearest road) and (2) Secondary Road Effect Zone (area from 150 m to 300 m to the nearest road). A regular sampling grid including one out of every 9 cells was created for each fire.These grids were intersected with burn severity data in the form of the Relative Differenced Normalized Burn Ratio (RdNBR), road distance category, stream distance, elevation, slope, terrain shape index, heat load index, canopy cover, and fuel type. We fit spatial regression models with RdNBR as the dependent variable.We found that high burn severity is less likely to occur in the Primary Road Effect Zone for most fires, although one fire exhibited the opposite relationship. Forest road edge effects were hypothesized to be an important determinant of burn severity because fragmentation created by roads alters the roadside fuel profile and environment and because road corridors create barriers to fire spread. Recognizing roadside effects on burn severity patterns highlights the need for further study of the range of effects that roads have on fuels and the fire environment and the potential for incorporating road effects into landscape-level assessments of fire risk.     

Wildfire Potential Mapping over the State of Mississippi: A Land Surface Modeling Approach

A relationship between the likelihood of wildfires and various drought metrics (soil moisture-based fire potential indices) were examined over the southern part of Mississippi. The following three indices were tested and used to simulate spatial and temporal wildfire probability changes: (1) the accumulated difference between daily precipitation and potential evapotranspiration (P - E); (2) simulated moisture content of the top 10 cm of soil; and (3) the Keetch-Byram Drought Index (KBDI). These indices were estimated from gridded meterological data and Mosaic-simulated soil moisture data available from the North American Land Data Assimilation System (NLDAS-2). The relationships between normalized fire potential index deviations and the probability of at least one fire occurring during the following five consecutive days were evaluated using a 23-year (1986-2008) forest fire record for an evenly spaced grid (0.25° x 0.25°) across the state of Mississippi's coastal plain. Two periods were selected and examined (January-mid June and mid September-December). There was good agreement between the observed and logistic model-fitted fire probabilities over the study area during both seasons. The fire potential indices based on the top 10 cm soil moisture and KBDI had the largest impact on wildfire odds, increasing it by almost 2 times in response to each unit change of the corresponding fire potential index during January-mid-June period and by nearly 1.5 times during mid-September-December. These results suggest that soil moisture-based fire potential indices are good indicators of fire occurrence probability across this region.     

FY-3D/MERSI-II全球火点监测产品及其应用

FY-3D/MERSI-II全球火点监测产品主要包括全球范围内的火点位置、亚像元火点面积和火点强度等信息,可用于实时监测全球范围的森林草原火灾、秸秆焚烧等生物质燃烧状况。火点判识算法主要根据中红外通道对高温热源的敏感特性,即含有火点的中红外通道像元辐亮度和亮温较远红外通道的辐亮度和亮温偏高,同时较周边非火点的中红外像元偏高,建立合适的阈值可探测含有火点的像元。亚像元火点面积估算主要使用中红外单通道估算,根据亚像元火点面积估算结果对火点强度进行分级,不同的级别表示不同程度的火点辐射强度。基于全球火点自动判识结果,每日生成0.01°分辨率的卫星遥感日全球火点产品,每月生产0.25°×0.25°格点的全球月火点密度图。在利用FY-3D/MERSI-II火点产品开展的全球火点监测应用中,对多起全球重大野火事件进行了监测,为防灾减灾、全球气候变化研究、生态环境保护等方面提供卫星遥感信息支持。     

Mapping Wildfire Burn Severity in Southern California Forests and Shrublands Using Enhanced Thematic Mapper Imagery

Abstract

Wildfire is a major disturbance agent in Mediterranean Type Ecosystems (MTEs). Providing reliable, quantitative information on the area of burns and the level of damage caused is therefore important both for guiding resource management and global change monitoring. Previous studies have successfully mapped burn severity using remote sensing, but reliable accuracy has yet to be gained using standard methods over different vegetation types. The objective of this research was to classify burn severity across several vegetation types using Landsat ETM imagery in two areas affected by wildfire in southern California in June 1999. Spectral mixture analysis (SMA) using four reference endmembers (vegetation, soil, shade, non‐photosynthetic vegetation) and a single (charcoal‐ash) image endmember were used to enhance imagery prior to burn severity classification using decision trees. SMA provided a robust technique for enhancing fire‐affected areas due to its ability to extract sub‐pixel information and minimize the effects of topography on single date satellite data. Overall kappa classification accuracy results were high (0.71 and 0.85, respectively) for the burned areas, using five canopy consumption classes. Individual severity class accuracies ranged from 0.5 to 0.94.     

基于决策树的CBERS遥感影像分类及分析评价

以江苏省徐州市为研究区,以城市土地利用遥感分类为目标,采用CBERS多光谱数据的近红外波段、全球环境监测植被指数(GEMI)、归一化植被指数(NDVI)及主成分分析得出的第一和第二主成分作为分类的特征数据,基于先验知识和统计分析构建层次分类决策树,进而发展和改进了决策树交互式构建算法,实现了城市土地利用遥感分类。通过与最大似然分类器(MLC)和支持向量机分类器(SVM)分类结果的比较分析,表明基于多种特征的决策树分类器能够有效应用于CBERS遥感数据分类,在研究区具有良好的推广性。     

运用NDVI-Albedo特征空间提取石漠化信息

石漠化是西南喀斯特地貌地区面临的最主要生态环境问题之一,对石漠化进行监测是其防治的一项重要工作。本文以威宁西部典型的石漠化研究区斗古乡为例,基于Landsat 8 OLI遥感数据,计算了研究区归一化植被指数(NDVI)和地表反照率(Albedo),通过NDVI-Albedo特征空间构建石漠化差值指数(RSDDI),对石漠化信息进行提取并对其进行精度验证。研究表明:基于NDVI-Albedo特征空间法构建的石漠化差值指数能够较为准确且便捷地对石漠化信息进行提取与分级,在中度石漠化及重度石漠化的制图精度均达到89%以上,提取效果较好,有利于西南喀斯特地貌地区对石漠化的定量评估与监测。     

一种新的增强型不透水面指数

针对西北干旱地区城市不透水面提取存在的局限性以及阈值确定的繁琐性等问题,该文提出了一种新的增强型不透水面指数(ENDISI)。基于Landsat8_OLI影像,以兰州市建成区为例进行不透水面信息提取,总体精度达到88.5%,结果较为理想。相比于已有的不透水面指数,ENDISI可以有效避免西北干旱区沙土、裸露山体的影响,适用性更强;采用"0"作为提取不透水面的阈值,简单、客观并且提取精度高。综上表明,ENDISI可用于西北干旱地区城市不透水面信息的高效提取。