基于两种无人机航拍影像的林窗和林冠提取分析

森林冠层和林窗的结构及其时空变化是理解森林生态系统格局、动态变化过程的重要基础。在当前生物多样性监测倍受关注的契机下,如何以合适的手段准确描述林窗面积、分布等特征,并与森林固定样地监测数据有效地结合,更好地回答群落构建的理论问题,使森林群落物种多样性维持机制得到更全面的认识,是目前亟待解决的问题。以鼎湖山南亚热带常绿阔叶林20hm2固定监测样地为研究对象,基于不同遥感影像提取方法对其林窗和林冠表层数据进行提取分析。结果表明:基于监督分类的提取方法适合RGB波段航片林窗的提取,在林窗分类中,应首先确定林窗高度、边界木与最小面积,不同分类方法差异主要表现在林冠分类中,林窗分类生产者精度和用户精度表现都较为一致。无人机航拍识别率受地形因素影响较大,在地形复杂林地应按坡度分区域进行飞行以降低误差。相对于地面调查,MD4-1000无人机航片的林窗识别率为98.7%;大疆Phantom4无人机航片的林窗识别率为72.3%,影像后期处理数据量小,同样适用于森林林窗定量研究,符合生态学、林业等从业人员对大型样地林窗长期监测的要求。无人机航拍南亚热带森林物种识别难度较大,基于MD4-1000无人机搭载的高分辨率相机,在地势平缓区域优选的4 hm2样地中可识别林冠表层物种数17种,共2 706个个体。搭载高分辨率无人机在降低飞行高度的基础上可进行部分物种识别。应用无人机近地面遥感对森林固定样地进行林冠监测,可为后期群落构建研究提供数据基础,有望从新的研究角度探讨森林群落物种多样性维持机制。     

基于随机森林的光子计数激光雷达点云滤波

新一代星载激光雷达卫星ICESat-2首次采用了微脉冲光子计数激光雷达技术,由于单光子探测的灵敏性导致数据在大气和地表下层产生了大量噪声,因此对光子计数激光雷达点云数据实现信号和噪声的分离是开展进一步应用研究的前提和基础。本文选择美国俄勒冈州和弗吉尼亚州2个研究区,采用MATLAS数据,根据光子点云数据的特点构造了12个光子点云特征,对所构造的特征利用随机森林进行变量筛选,用机器学习方法对光子点云进行分类,并将建立好的模型推广到整个研究区。研究结果表明,本文构建的分类器分类总精度达到了96.79%,Kappa系数为0.94,平均生产者精度和用户精度分别为97.1%和96.8%。在相对弱噪声、平坦地形区域和强噪声、复杂地形区域都取得较好的分类结果。本文结果显示了基于少量样本通过机器学习的方法构建模型,可以推广到较大范围区域的光子点云分类应用中。     

基于深度学习的钨钼找矿靶区预测方法研究

随着矿产勘查工作由浅部矿向深部隐伏矿、由易识别矿向难识别矿发展,找矿难度日益增大,地质专家越来越重视新理论、新方法、新技术的应用。深度学习作为人工智能的前沿领域/技术,对于实现矿产资源预测“智能化预测评价”具有得天独厚的优势。本文以陕西省镇安县西部钨钼矿集区单元素化探异常原始数据为基础,提出了基于深度学习的钨钼矿产评价方法。该方法以归一化地球化学数据作为模型训练数据,通过深度学习中深度自编码网络方法实现异常值提取进而识别重点成矿有利地段,实现矿产资源找矿远景区定性预测。研究结果表明,在对957条单元素化探异常原始数据分类且做好模型标签后,整个过程在计算机的“黑盒子”中自动完成学习和预测,相较于传统预测研究方法,本文方法具有自动化程度高和客观性强的特征。此外,本文利用已知矿点构建训练数据集,采用随机森林方法对预测区进行矿产资源找矿靶区预测圈定,为进一步缩小找矿靶区范围提供科学依据。     

南方亚热带季风区将乐县森林植被动态变化及其对气候变化的响应

在全球气候变化背景下,植被动态变化以及植被对气候变化的响应方式已经成为生态学和地理学领域的热点。本文对比分析了南方亚热带季风区将乐县不同类型森林植被对不同时间尺度的干旱响应的差别。基于2000-2017年MODIS-EVI数据及气象站点数据,用最大值合成法、趋势分析法以及相关分析法,分析了森林植被及气象因子的动态变化特征,并对比不同森林植被对气候变化响应的差别。研究表明：① 2000-2017年,研究区植被覆盖度、EVI和降水均显著增加,区域内湿度增加,森林长势渐趋良好;② EVI在生长季初期和末期与同期的降水、温度均显著正相关（P<0.1）,初期森林受降水因子的影响更大,末期受温度因子的影响大;③ 1-3月和周年的气候变化对森林的生长至关重要,长时间尺度的湿度增加对森林生长具有显著的促进作用,SPEI的时间尺度越长与EVI的相关性也越大;④ 针阔混交林与同期温度、降水的相关系数最高,并且与不同时间尺度的SPEI相关性均比较高,属于气候敏感型林型,在生产经营中要谨慎预防气候变化对该林型带来的伤害;⑤ 森林覆盖度变化与降水和SPEI_24的相关性极显著,长时间尺度的降水变化是影响森林植被覆盖率变化的重要因素之一。     

A protocol for classifying ecologically relevant marine zones,a statistical approach

Mapping ecologically relevant zones in the marine environment has become increasingly important. Biological data are however often scarce and alternatives are being sought in optimal classifications of abiotic variables. The concept of ‘marine landscapes’ is based on a hierarchical classification of geological, hydrographic and other physical data. This approach is however subject to many assumptions and subjective decisions.     

Trophic significance of the kelp Laminaria digitata (Lamour.) for the associated food web: a between-sites comparison

This study aimed at establishing the trophic significance of the kelp Laminaria digitata for consumers inhabiting two rocky shores of Northern Brittany (France), displaying contrasted ecological conditions. The general trophic structure did not vary between these two sites, with a wide diversity of filter-feeders and predators, and only 14% of the species sampled belonging to the grazers' trophic group. The diversity of food sources fueling the food web appeared also similar. The food webs comprised four trophic levels and the prevalence of omnivory appeared relatively low compared to previous studies in the same area. Conversely, to the food web structure, which did not differ, the biochemical composition of L. digitata differed between the two sites, and was correlated to a larger diversity of grazers feeding on this kelp in sheltered conditions. This indicated that the spatial variability occurring in the nutritive value of L. digitata is likely to deeply affect the functioning of kelp-associated food webs. The contribution of L. digitata-derived organic matter to the diet of filter-feeders inhabiting these two environments was assessed using the mixing model Isosource, which showed the higher contribution of kelp matter in sheltered conditions. These results highlight the spatial variability that may occur in the functioning of kelp-associated food webs. Moreover, this suggests that hydrodynamics is likely to control the availability of kelp-derived organic matter to local filter-feeders, probably through an increase of detritus export in exposed areas.     

Surface sediment diatoms from the western Pacific marginal seas and their correlation to environmental variables

Diatom data of 192 surface sediment samples from the marginal seas in the western Pacific together with modern summer and winter sea surface temperature and salinity data were analyzed.The results of canonical correspondence analysis show that summer sea-surface salinity(SSS) is highly positively correlated with winter SSS and so is summer sea-surface temperature(SST) with winter SST.The correlations between SSSs and SSTs are less positively correlated,which may be due to interactions of regional current pa...     

The spectroscopic period of the Cepheid SU Cassiopeiae

The spectroscopic period of the short periodic Cepheid SU Cas was determined from the radial velocity measurements by ABT, GIEREN and HÄUPL. The calculated period is 1.949312 days. Radial velocity differences between measurements by GIEREN and HÄUPL are explained as velocity zero point of the used spectrographs. The zero point of the Tautenburg Coudé spectrograph was determined by measuring the radial velocity standard star ϵ Leo. The difference to the standard value was -0.7 km/s. No significant differences are found between the spectroscopic period and light curve of SU Cas in the last 35 years.     

西安高陵人工林土壤干层与含水量季节变化研究

通过野外调查和室内测定,利用烘干称重法对高陵地区丰水年前后不同人工林下0~6 m土壤含水量及土壤水分的季节变化进行研究。结果表明,2002年高陵田家村中国梧桐林和杨树林下160~400 cm范围内均已发育了土壤干层。经过2003年丰水年充沛的降水补给,2004年高陵团庄槐树林、杏树林0~6 m土层均未出现土壤干层,说明水分在丰水年得到很好恢复。丰水年后梨、杏、槐三种人工林160~400、410~600 cm层位土壤含水量均显示春季最高,夏季次之,秋季降到最低或略微上升。     

ICESat-2植被冠层高度和地表高程数据产品用于森林高度提取的效果评价

由美国宇航局(NASA)研制的新一代冰、云和陆地高程卫星(ICESat-2)于2018-09-15发射成功,其搭载的先进地形激光测高系统(ATLAS)采用微脉冲多波束光子计数激光雷达技术,可用于全球高程线采样数据的获取.目前公开发布了9种数据产品,其中包括植被冠层高度和地表高程数据产品(ATL08),为全球森林结构参数...