Fast R-CNN深度学习和无人机遥感相结合在松材线虫病监测中的初步应用研究

松材线虫病因其破坏性强、传播速度快和防治难度大等特点,严重威胁着我国的松林资源.及时发现、定位和清理病死松树是控制松材线虫病蔓延的有效手段.本研究利用小型无人机获得松材线虫病疫点的可见光和多光谱的航摄影像.根据松树针叶颜色变化,将松材线虫Bursaphelenchus xylophilus侵染的松树分为病树和枯死树两种...     

利用大面积无人机航拍影像研究松材线虫病病死树林间分布规律

本研究使用固定翼无人机拍摄4 200 ha林地,从中选取了广东省河源市和平县阳明镇、紫金县紫城镇、东源县义合镇共3个样地的3 500 ha林地的航拍影像进行分析,用以探究松材线虫Bursaphelenchus xylophilus病死树的空间分布情况,及不同立地因子对疫情的影响,为松材线虫病监测预报提供解决途径。通过Pix4Dmapper软件对航拍的图像进行拼接生成正射影像图(DOM)等成果,然后使用eCognition(易康)软件对影像成果进行分割、分类和信息提取,最后借助ArcGIS平台进行病死树数量统计并获取方位、坡向、坡度、海拔等立地因子信息。结果表明,松材线虫病死树分布均呈聚集分布。使用双对角线法、平行线法、“Z”字法、五点法等不同抽样方法调查发现,仅五点法所得平均数与总体平均数无明显差异(P<0.05)。松材线虫病死树在不同立地因子下均有差异：主要分布在西坡、南坡和东南坡,西坡最多为25.94%,其次是南坡23.57%;主要分布在半阳坡和阳坡,半阳坡占36.54%,阳坡占34.09%;主要分布在凸坡,但随着疫情的发展,凹坡病死树数量逐渐超过凸坡;主要分布海拔区间在30...     

基于多色彩空间的YOLOv5松枯死树检测方法

【目的】针对在松枯死树监测实践中,从无人机航拍RGB影像中自动识别松枯死树漏检率高的问题,提出了一种生产应用场景下基于多色彩空间的YOLOv5松枯死树高精度自动识别新方法。【方法】利用无人机采集大面积松材线虫病发生林分的RGB图像,用Pix4Dmapper软件拼接,用LabelImg开源软件建立VOC格式的松枯死树数据集,分别用Faster R-CNN、YOLOv3、YOLOv4、YOLOv5、SSD和EfficientDet等6种基于深度学习的目标检测算法对数据集进行训练和测试,以精确率、召回率、平均准确率以及F₁分数作为评价指标筛选出最优目标检测算法;然后将采集的RGB图像转换成LAB和HSV色彩空间图像,再将这3个色彩空间的图像分别用最优目标检测算法进行训练,得到目标在每个色彩空间的边界框,使用非极大值抑制算法对这些边界框进行处理,得到最优边界框实现松枯死树自动识别。【结果】6种算法均取得良好效果,其中YOLOv5模型为最优算法,其精准率、平均查准率和F₁分数在6种算法中均最高,分别达到97.58%、82.40%和0.85。通过3个色彩空间融合后,反映漏检情况的召回率由74.54%提高到98.99%,平均准确率提升至98.39%。【结论】基于多色彩空间的YOLOv5模型能够显著提高从无人机航拍RGB影像中检测松枯死树的精度,为松枯死树监测提供了有力工具,也有助于松材线虫病的防治。     

江西省赣州市南康区松材线虫病发生特征

松材线虫病是我国南方森林面临的主要灾害之一。本文基于野外调查和高分一号(GF-1)卫星WFV影像数据,采用随机森林模型构建松材线虫病空间识别模型,探究地形、人类活动和林分因子等对病害发生的影响,监测病害空间分布,并采用空间自相关性分析评估江西省赣州市南康区松材线虫病发生特征。结果表明：构建模型对松材线虫病的识别效果良好(AUC值=0.99,总体精度=0.96),可以实现对区域松材线虫病空间分布情况的有效监测;归一化差异绿度指数(NDGI)、距高速公路的距离、归一化植被指数(NDVI)是重要的建模因子;空间自相关性分析表明,松材线虫病的发生存在明显的空间正相关性即空间聚集性特征;南康区松材线虫病高发生区集中于赤土乡、朱坊镇和十八塘乡,低发生区集中于蓉江街道附近;分析变量的边际效应发现,离高速公路远、离县道近的低海拔地段是松材线虫病易发区域。研究结果可服务于区域松材线虫病分布的快速监测,对该病害防治和管理具有一定的指导意义。     

顾及无人机影像点云特征的绿地信息分类方法

采用无人机影像进行绿地信息分类时,常利用影像光谱、纹理、形状等分类特征,忽视了通过无人机影像生成点云构建的数字表面模型(Digital surface model, DSM)和数字高程模型(Digital elevation model, DEM)差异特征。基于此,提出一种顾及无人机影像点云特征的绿地信息分类方法。方法首先基于摄影测量理论对研究区无人机影像进行空三计算,并生成点云,在此基础上构建DSM、DEM和数字正射影像(Digital Orthophoto Map, DOM);然后,利用DSM和DEM模型构建地物高度差异模型(normalizedDigitalSurfaceModel,nDSM);最后,利用可见光波段差异植被指数(Visible-band difference vegetation index, VDVI)对DOM进行植被与非植被分类,并结合nDSM对植被进行分类。以昆明市呈贡区白龙潭公园为研究区进行绿地信息分类, Kappa系数精度达到0.862,实验表明本文的方法对城市绿地调查具有实际意义。     

基于多光谱卫星影像与机器学习算法的松材线虫病受害林分识别研究

松材线虫病(Pine Wilt Disease, PWD)被称为“松树癌症”,具有高传染率和高死亡率,对我国森林资源构成了严重的威胁,对我国的经济、社会和生态造成了重大损失。及时发现并清理疫木是遏制松材线虫病蔓延的有效手段,精准监测疫木是防控松材线虫病的前提,但是现阶段缺少大面积识别松材线虫病疫木的技术方法。本文旨在探索哨兵-2号与Landsat-8遥感卫星影像对受害松林的识别能力,采用随机森林(Random Forest, RF)、支持向量机(Support Vector Machine, SVM)、决策树(Decision Tree, DT)和极端梯度提升(Extreme Gradient Boosting, XGBoost)等4种机器学习算法建立了松材线虫病监测模型。结果表明：基于哨兵-2号影像数据建立的监测模型对受害松林的识别准确率高于Landsat-8遥感卫星影像,其中基于10 m分辨率的影像数据建立的监测模型识别准确率最高,随机森林、决策树、支持向量机和极端梯度提升等算法建立模型的准确率分别达到了79.3%、76.2%、78.7%和78.9%。在3种不同的影像数据集中,RF...     

松材线虫病发生及防控现状

松材线虫在20世纪30年代就由美国的Steiner和Buhrer作为新种报道,但直至1971年才在日本被确认是引起松树枯死的原因。目前松材线虫在世界上分布还不很普遍,但对世界松林构成严重威胁。本文介绍了松材线虫发生及防控现状,着重介绍了松材线虫在世界的分布、寄主植物、传媒昆虫及管理现状。针对我国松材线虫病的发生和防控现状,探讨了我国松材线虫病防控对策和提出展望。     

基于无人机遥感监测滩涂湿地入侵种互花米草植被覆盖度

为了实现对沿海滩涂湿地资源与生态环境的有效管理与利用,需要对区域内的入侵种互花米草进行高精度的监测与分析.本文以福建三沙湾为试验区,以低空无人机获取的可见光和多光谱影像为数据源,对互花米草植被覆盖度进行监测与分析,通过NDVI指数模型获取了多光谱影像的植被覆盖度信息,以可见光影像为参考进行了精度检验.结果表明: 影像区互花米草植被覆盖度以40%～60%和60%～80%的中高和高等级覆盖度为主.NDVI模型估算值与真实值之间的均方根误差RMSE为0.06,决定系数R²为0.92,两者具有较好的一致性.     

基于多源遥感数据的大豆叶面积指数估测精度对比

近年来遥感技术的革新促使遥感源越来越丰富.为分析多源遥感数据的叶面积指数(LAI)估测精度,本文以大豆为研究对象,利用比值植被指数(RVI)、归一化植被指数(NDVI)、土壤调整植被指数(SAVI)、差值植被指数(DVI)、三角植被指数(TVI)5种植被指数,结合地面实测LAI构建经验回归模型,比较3类遥感数据(地面高光谱数据、无人机多光谱影像以及高分一号WFV影像)对大豆LAI的估测能力,并从传感器几何位置和光谱响应特性以及像元空间分辨率三方面分析讨论了3类遥感数据的LAI反演差异.结果表明： 地面高光谱数据模型和无人机多光谱数据模型都可以准确预测大豆LAI(在α=0.01显著水平下,R²均>0.69,RMSE均<0.40);地面高光谱RVI对数模型的LAI预测能力优于无人机多光谱NDVI线性模型,但两者差异不大(E_A相差0.3%,R²相差0.04,RMSE相差0.006);高分一号WFV数据模型对研究区内大豆LAI的预测效果不理想(R²<0.30,RMSE>0.70).针对星、机、地三类遥感信息源,地面高光谱数据在反演LAI方面较传统多光谱数据有优势但不突出;16 m空间分辨率的高分一号WFV影像无法满足田块尺度作物长势监测的需求;在保证获得高精度大豆LAI预测值和高工作效率的前提条件下,基于无人机遥感的农情信息获取技术不失为一种最佳试验方案.在当今可用遥感信息源越来越多的情况下,农业无人机遥感信息可成为指导田块精细尺度作物管理的重要依据,为精准农业研究提供更科学准确的信息.     

中国亚热带4个森林动态监测样地无人机可见光遥感影像数据集

常绿阔叶林是我国亚热带地区的地带性植被类型。由于亚热带森林植物群落垂直结构复杂、林冠郁闭度高, 对常绿阔叶林冠层的研究尚缺乏高质量的监测数据。本数据集包含浙江天童山、浙江百山祖、广东车八岭、广东鼎湖山4个大于20 ha的森林动态监测样地2014年8月或2016年9月采集的无人机可见光遥感影像。本数据集是通过将无人机影像、地面控制点和地面调查数据相结合而获得的。每个样地的数据集包括4个文件: ~5 cm空间分辨率的正射影像图和数字表面模型、1 m空间分辨率的森林冠层高度数据和正射影像质量报告。本数据集可为常绿阔叶林的林冠生态学、生物多样性、生态系统功能等研究提供数据支撑。     

Influence of induced water stress in ponderosa pine on pine sawflies

Summary Three levels of water stress were induced on pole-size ponderosa pine (Pinus ponderosa) to determine the influence of plant moisture stress on oviposition, survival, and growth of two species of pine sawfly (Neodiprion fulviceps and N. autumnalis). It was found that water stress affected oviposition and subsequent egg survival but not larval development or survival. Stress had a negative effect on early season oviposition (N. fulviceps) and a positive effect on late season oviposition (N. autumnalis). Egg hatch was different between species and years and among treatment levels. Larval development, feeding, and survival were not affected by water stress. Overall, the effect of stress was not sufficient to explain population outbreaks of sawflies. Several hypotheses are presented as possible explanations for the observed experimental results.     

A dispersal-aggregation model for mountain pine beetle in lodgepole pine stands

In this paper a dispersal-attack theory for bark beetle attacking trees is developed from a set of simple assumptions, and the resulting theoretical model is fit to data from four epidemic studies. Implications of the theory are discussed in relation to the dynamics of lodgepole pine-mountain pine beetle interactions.     

Influence of watering and trenching ponderosa pine on a pine sawfly

Summary Neodiprion autumnalis (Smith) larvae were caged for two successive years on root-trenched, watered, and untreated ponderosa pine (Pinus ponderosa Doug. ex Laws.) to determine effects of host moisture stress on larval feeding. Levels of moisture stress (as measured by the Scholander pressure chamber) differed significantly among treatment levels during 1984 and 1985 larval feeding periods. Differences in larval feeding success were not detected in 1984. In 1985, however, larvae on trenched (stressed) trees clipped and rejected more foliage, consumed more needles, had lower pupal weights, lower survival, and a longer feeding period than larvae on watered or untreated trees. Frass production did not differ among treatment levels. The length of the feeding period was shorter for larvae on watered trees than for larvae on untreated trees, but other measures of feeding success did not significantly differ between watered and untreated trees.     

Development of secondary pine forests after pine wilt disease in western Japan

Abstract. The development of secondary Pinus densiflora (Japanese red pine) forests after pine wilt disease was studied through phytosociological analysis, estimation of forest structure before disease and size-structure, tree ring and stem analyses. Following the end of the disease, the growth of previously suppressed small oak trees was accelerated. This is quite different from the development of forests following fire, which starts with the establishment of pine seedlings. Pine wilt disease shifted the dominance of secondary forests from Pinus densiflora to Quercus serrata oak forest. In pine forests, disturbance by fire is important for forest maintenance. In contrast, disturbance by pine wilt disease leads to an acceleration of succession from pine forest to oak forest.     

Effects of spring-killed pine trees on the epidemics of pine wilt disease

Pine wilt disease (PWD) is caused by the pinewood nematode (PWN), Bursaphelenchus xylophilus (Steiner et Buhrer) Nickle, and is transmitted by cerambycid beetles. In some pine trees infected with the PWNs in Japan, foliage changes from green to brown in summer to autumn of a nematode infection year (summer- autumn-killed trees) and the others in the following spring of a nematode infection year (spring-killed trees). The vector beetles require 1 or 2?years for development in cool summer areas and 1?year in warm summer areas. To evaluate the effects of the spring-killed trees and vectors with a long developmental time on the PWD epidemics, we presented simple mechanistic mathematical models. The models showed that it was possible for spring-killed trees to cause PWD epidemics when the transmission rate was high, and the efficacy of spring-killed trees as infection source was similar to that of summer?C autumn-killed trees. Spring-killed trees and vector beetles with a developmental time of 2?years harbored in summer- autumn-killed trees delayed epidemic timing by 3?C10?years or actually suppressed epidemics.     

Acid-Fastness of pine pollen

This is a study of the acid-fast staining characteristics of pine pollen and an investigation of factors causing loss of acid-fastness after pine pollen has been in contact with tisues or mucous membranes. Intact loblolly pine pollen was readily stained with cold carbol-fuchsin, and retained its acid-fastness after decolorization with 3% HC1 in 95% ethyl alcohol for 2 min, followed by methylene blue counterstain. Pine pollen resembles spermatozoa in ease of staining and resistance to decolorization. Acid-fastness was destroyed by crushing or by germination of the pollen grain, and by contact for several hours with serum or saline solutions, but was unchanged by exposure to 0.1% solution of streptomycin in water. Nonviable pine pollen did not lose acid-fastness after suspension for several days in serum or water. When counterstain was omitted, crushed or germinated pollen appeared red to pink after staining with carbol-fuchsin and decolorization with acid alcohol, thus indicating that lipids of an acid-fast nature were still present. The large size and biologic properties of pine pollen provide a unique means of studying the chemical and physical aspects of the acid-fast phenomenon.     

Modeling the dynamics of mountain pine beetle aggregation in a lodgepole pine stand

Summary At least once a year the mountain pine beetle searches for lodgepole pines that provide a suitable habitat for a new brood. After attacking females feed, they produce an attractant pheromone that causes beetles to aggregate and, during outbreaks, to usually mass attack the focus tree. Near the completion of mass attack, incoming beetles are repelled and initiate attacks on adjacent recipient trees. An understanding of this switching process is useful for prescribing measures that minimize beetle damage.A mathematical model was developed to (1) describe beetle aggregation, (2) predict the relation of tree susceptibility and switching to changes in beetle density, (3) provide a structure for current knowledge, and (4) pose questions for further research. The model indicates that a high population density ensures mass aggregation and consequently successful tree colonization and switching. The model also indicates that the number of beetles attracted per attacking beetle differs from tree to tree, possibly depending on resin quality and production and/or the local flying density of beetles. Field and model results indicate that tree size appears to affect the repellence of beetles, suggesting that the attack density or the visual attractiveness of large trees is a factor. Further research could be directed at our assumptions on host resistance, repellence, pheromone emission rates, threshold concentrations, navigation, and pheromone dispersion. Return Address: Center for Quantitative Sciences, University of Washington, Seattle, Washington, 98195, USA     

Loblolly pine grown under elevated CO2 affects early instar pine sawfly performance

Seedlings of loblolly pine Pinus taeda (L.), were grown in open-topped field chambers under three CO₂ regimes: ambient, 150 l l^–1 CO₂ above ambient, and 300 l l^–1 CO₂ above ambient. A fourth, non-chambered ambient treatment was included to assess chamber effects. Needles were used in 96 h feeding trials to determine the performance of young, second instar larvae of loblolly pine's principal leaf herbivore, red-headed pine sawfly, Neodiprion lecontei (Fitch). The relative consumption rate of larvae significantly increased on plants grown under elevated CO₂, and needles grown in the highest CO₂ regime were consumed 21% more rapidly than needles grown in ambient CO₂. Both the significant decline in leaf nitrogen content and the substantial increase in leaf starch content contributed to a significant increase in the starch:nitrogen ratio in plants grown in elevated CO₂. Insect consumption rate was negatively related to leaf nitrogen content and positively related to the starch:nitrogen ratio. Of the four volatile leaf monoterpenes measured, only -pinene exhibited a significant CO₂ effect and declined in plants grown in elevated CO₂. Although consumption changed, the relative growth rates of larvae were not different among CO₂ treatments. Despite lower nitrogen consumption rates by larvae feeding on the plants grown in elevated CO₂, nitrogen accumulation rates were the same for all treatments due to a significant increase in nitrogen utilization efficiency. The ability of this insect to respond at an early, potentially susceptible larval stage to poorer food quality and declining levels of a leaf monoterpene suggest that changes in needle quality within pines in future elevated-CO₂ atmospheres may not especially affect young insects and that tree-feeding sawflies may respond in a manner similar to herb-feeding lepidopterans.