Support vector machines for tree species identification using LiDAR-derived structure and intensity variables   总被引：1，自引：0，他引：1  

Zhenyu Zhang 《国际地球制图》2013,28(4):364-378

Tree species identification and forest type classification are critical for sustainable forest management and native forest conservation. Recent success in forest classification and tree species identification using LiDAR (light detection and ranging)-derived variables has been reported in many studies. However, there is still considerable scope for further improvement in classification accuracy. It has driven research into more efficient classifiers such as support vector machines (SVMs) to take maximum advantage of the information extracted from LiDAR data for potential increases in the accuracy of tree species classification. This study demonstrated the success of the SVMs for the identification of the Myrtle Beech (the dominant species of the Australian cool temperate rainforest in the study area) and adjacent tree species – notably, the Silver Wattle at individual tree level using LiDAR-derived structure and intensity variables. An overall accuracy of 92.8% was achieved from the SVM approach, showing significant advantages of the SVMs over the traditional classification methods such as linear discriminant analysis in terms of classification accuracy.  相似文献   

Comparing the hydrology of grassed and cultivated catchments in the semi‐arid Canadian prairies     

G. van der Kamp  M. Hayashi  D. Galln 《水文研究》2003,17(3):559-575

At the St Denis National Wildlife Area in the prairie region of southern Saskatchewan, Canada, water levels in wetlands have been monitored since 1968. In 1980 and 1983 a total of about one‐third of the 4 km² area was converted from cultivation to an undisturbed cover of brome grass. A few years after this conversion all the wetlands within the area of grass dried out; they have remained dry since, whereas wetlands in adjacent cultivated lands have held water as before. Field measurements show that introduction of undisturbed grass reduces water input to the wetlands mainly through a combination of efficient snow trapping and enhanced infiltration into frozen soil. In winter, the tall brome grass traps most of the snowfall, whereas in the cultivated fields more wind transport of snow occurs, especially for short stubble and fallow fields. Single‐ring infiltration tests were conducted during snowmelt, while the soil was still frozen, and again in summer. The infiltrability of the frozen soil in the grassland is high enough to absorb most or all of the snowmelt, whereas in the cultivated fields the infiltration into the frozen soil is limited and significant runoff occurs. In summer, the infiltrability increases for the cultivated fields, but the grassland retains a much higher infiltrability than the cultivated land. The development of enhanced infiltrability takes several years after the conversion from cultivation to grass, and is likely due to the gradual development of macropores, such as root holes, desiccation cracks, and animal burrows. Copyright ©2002 Crown in the right of Canada. Published by John Wiley & Sons, Ltd.  相似文献   

Sedimentary processes in a submarine canyon excavated into a temperate‐carbonate ramp (Granada Basin,southern Spain)     

ÁNGEL PUGA‐BERNABÉU  JOSÉ M. MARTÍN  JUAN C. BRAGA 《Sedimentology》2008,55(5):1449-1466

During the Late Tortonian, shallow‐water temperate carbonates were deposited in a small bay on a gentle ramp linked to a small island (Alhama de Granada area, Granada Basin, southern Spain). A submarine canyon (the ‘Alhama Submarine Canyon’) developed close to the shoreline, cross‐cutting the temperate‐carbonate ramp. The Alhama Submarine Canyon had an irregular profile and steep slopes (10° to 30°). It was excavated in two phases reflected by two major erosion surfaces, the lowermost of which was incised at least 50 m into the ramp. Wedge‐shaped and trough‐shaped, concave‐up beds of calcareous (terrigenous) deposits overlie these erosional surfaces and filled the canyon. A combination of processes connected to sea‐level changes is proposed to explain the evolution of the Alhama Submarine Canyon. During sea‐level fall, part of the carbonate ramp became exposed and a river valley was excavated. As sea‐level rose, river flows continued along the submerged, former river‐channel, eroding and deepening the valley and creating a submarine canyon. At this stage, only some of the transported conglomerates were deposited locally. As sea‐level continued to rise, the river mouth became detached from the canyon head; littoral sediments, transported by longshore and storm currents, were now captured inside the canyon, generating erosive flows that contributed to its excavation. Most of the canyon infilling took place later, during sea‐level highstand. Longshore‐transported well‐sorted calcarenites/fine‐grained calcirudites derived from longshore‐drift sandwaves poured into and fed the canyon from the south. Coarse‐grained, bioclastic calcirudites derived from a poorly sorted, bioclastic ‘factory facies’ cascaded into the canyon from the north during storms.  相似文献   

三江源区毒杂草型退化草地植被光谱特征分析     

周伟  李浩然  石佩琪  谢利娟  杨晗 《地球信息科学学报》2020,22(8):1735-1742

毒草型退化草地具有群落演替特点,通过高光谱遥感技术反演毒杂草分布与退化草地群落结构能对该类退化草地进行有效监测,而光谱特征分析是毒杂草与优良牧草遥感识别的基础。本文选取了三江源区毒草型退化草地的8种典型毒杂草和4种优良牧草的地面实测高光谱数据作为研究样本,经过SG平滑、包络线去除、导数变换和光谱参量化对毒杂草种和优良牧草种的光谱特征进行了分析,并通过马氏距离法提取其特征识别波段。结果表明：① 8种典型毒杂草和4种优良牧草的 “近红外峰值”差异较大,其中鹅绒萎陵菜的“近红外峰值”达到60.07%,而最小者早熟禾仅为17.53%;② 经包络线去除处理后,植被光谱曲线中吸收谷和反射峰光谱差异更加明显,且可减少环境背景对植被光谱的影响,如沼泽草甸的鹅绒委陵菜和驴蹄草,其“绿峰幅值”分别为6.46%和6.89%,经处理后其“绿峰指数”分别为0.2866和0.3671,而在2种环境下生长的同一草种（狼毒草1和狼毒草2）的峰谷特征差异不明显;③ 基于马氏距离法提取的毒杂草与优良牧草的敏感识别波段主要分布在680~750 nm和900~1000 nm波长范围内,以醉马草与矮嵩草为例,其基于反射率的敏感识别波段为713.1~737.1 nm和934.6~965.6 nm。该研究可为利用高光谱遥感进行大面积毒杂草草种识别和植被群落生长监测提供重要科学依据,对于三江源区毒杂草的监测防治和畜牧业的可持续发展具有重要意义。  相似文献   

基于稀疏样点的蒙古国产草量估算方法研究     

王艳杰  王卷乐  魏海硕  Altansukh Ochir  Davaadorj Davaasuren  Sonomdagva Chonokhuu 《地球信息科学学报》2020,22(9):1814-1822

产草量是衡量草原生产力和诊断草原健康状况的指标,是草地资源管理的重要依据。近年来,遥感数据结合地面实测数据建模已成为产草量估算的重要手段。充足的实测样点信息是产草量遥感建模估算的基础。受境外采样多重因素的制约,蒙古国产草量估算研究中无法获取足够且分布均匀的实测样点,估产模型的精度受到影响,这一问题目前尚未发现有好的解决方法。本研究选取中蒙铁路沿线（蒙古段）两侧200 km缓冲区作为研究区,针对产草量遥感估算中野外样点稀少且分布不均的问题,引入P-BSHADE方法,基于多年NDVI数据和获取的少量地面实测样点数据,考虑草地分布的非均匀性以及样点之间的相关性,对均匀分布的模拟样点处的产草量数据进行插值实验。结果显示,P-BSHADE法的插值效果优于Kriging法,可得到均匀分布于研究区的样点。基于以上实测样点和插值样点,结合NDVI、EVI、PsnNet 3种植被指数进行遥感建模,最优模型精度达到80%,精度优于已有相关研究。选取其中最优的基于NDVI的指数模型对研究区2000—2019年产草量进行反演,获得的产草量空间格局与年际变化与已有研究结果趋势吻合,进一步印证了结果的可靠性和插值方法的可行性。本研究通过插值的方式改善数据源从而提高估算模型精度是一种全新的思路与尝试,对于“一带一路”等境外区域资源环境监测具有借鉴意义。  相似文献   

1982~2013年青藏高原高寒草地覆盖变化及与气候之间的关系   总被引：7，自引：2，他引：5  

陆晴  吴绍洪  赵东升 《地理科学》2017,37(2):292-300

利用GIMMS NDVI数据和地面气象站台观测数据,对青藏高原1982~2013年高寒草地覆盖时空变化及其对气象因素的响应进行研究,结果表明：青藏高原高寒草地生长季NDVI表现为从东南到西北逐渐减少的趋势,近32 a来,整个高原草地生长季NDVI呈上升趋势,增加速率为0.000 3/a (p<0.05);高寒草地生长季NDVI年际变化具有空间异质性,整体为增加趋势,呈增加趋势的面积约占研究区域面积的75.3%,其中显著增加的占26.0% （p<0.05）,类型主要为分布在青藏高原东北部地区的高寒草甸;比例为4.7%,草地类型主要为高寒草原,主要分布在高原西部地区;基于生态地理分区的分析显示,青藏高原草地与降水、温度的相关关系具有明显的空间差异,高寒草地生长季NDVI均值与降水呈显著正相关,对降水的滞后效应显著;高原东北部温度较高,热量条件较好,降水为高寒草地生长季NDVI变化的主导因子;东中部地区降水充沛,温度则为高寒草地生长的制约因子;南部地区降水和温度都较适宜,均与高寒草地生长季NDVI相关性显著(p< 0.05),共同作用于草地的生长;中部和西部地区,气候因子与高寒草地生长季NDVI关系均不显著。  相似文献   

基于高光谱分析的草地叶绿素含量估算研究进展   总被引：3，自引：0，他引：3  

马文勇  王训明 《地理科学进展》2016,35(1):25-34

叶绿素是草地进行光合作用最重要的色素,与氮素、蛋白质、水分等其他植被生化参数均有着密切关系,是草地光合能力及生理状况的良好指示剂。利用高光谱数据建模分析是实现大面积草地叶绿素含量估算的一种重要手段。本文将基于高光谱分析估算草地叶绿素含量的方法总结为：基于红边位置及光谱指数的经验模型和辐射传输模型两类。经验模型通过建立叶绿素含量与红边位置、光谱指数之间统计关系来估算叶绿素含量,参数简单,实用性较强;但光谱指数构造形式多样且与草地叶绿素含量关系复杂,在一定程度上影响了叶绿素的估算精度。辐射传输模型以叶绿素含量与辐射能量的作用过程作为其理论基础,模型中参数较多且对估算尺度敏感,有待于进一步完善。目前草地叶绿素估算的研究相对薄弱,专门用于估算的模型较少。未来的工作一方面应致力于发展和改进适宜于草地的光谱指数,同时确定合适的辐射传输模型参数以改进模型对草地的监测效果;另一方面,如何由叶片尺度拓展到冠层尺度进而拓展到像元尺度,从而更好地实现大面积草地叶绿素含量估算,是一项既具有重要意义又有挑战性的工作。  相似文献   

Throughfall and its spatial variability in a temperate rainforest of simple structure     

Jennifer Styger  Jamie B. Kirkpatrick  Greg Unwin 《自然地理学》2016,37(3-4):251-263

There are few throughfall data from southern hemisphere closed-forest, and none from Tasmanian callidendrous cool-temperate rainforest, which has a simpler structure than most primary rainforests. We determined throughfall, measured its local spatial variation, and tested its relationships with rainfall, rainfall intensity, wind speed, canopy dryness, canopy cover, and other structural variables in a cool-temperate callidendrous rainforest in Tasmania. Eighty-two percent of the precipitation was measured as throughfall, which occurred after 2.3 mm of rain fell on dry canopies. The cumulative rainfall in 25 randomly located funnel rain gauges on the forest floor varied from 160 to 567 mm. Canopy cover and other structural variables did not predict the spatial pattern of throughfall. While throughfall in rainfall events was related to rainfall amount and intensity, wind speed did not affect throughfall as a percentage of rainfall. Percentages of throughfall to rainfall over 100 for many low rainfall events may indicate a contribution of fog drip to precipitation on the forest floor. The high local spatial variability in throughfall indicates the mean moisture conditions on the forest floor may not be a good indicator of the potential for localised fire damage.  相似文献   

Oceanographic controls on shallow‐water temperate carbonate sedimentation: Spencer Gulf,South Australia          下载免费PDF全文

Laura G. O'Connell  Noel P. James  Mark Doubell  John F. Middleton  John Luick  David R. Currie  Yvonne Bone 《Sedimentology》2016,63(1):105-135

Spencer Gulf is a large (ca 22 000 km²), shallow (<60 m water depth) embayment with active heterozoan carbonate sedimentation. Gulf waters are metahaline (salinities 39 to 47‰) and warm‐temperate (ca 12 to ?28°C) with inverse estuarine circulation. The integrated approach of facies analysis paired with high‐resolution, monthly oceanographic data sets is used to pinpoint controls on sedimentation patterns with more confidence than heretofore possible for temperate systems. Biofragments – mainly bivalves, benthic foraminifera, bryozoans, coralline algae and echinoids – accumulate in five benthic environments: luxuriant seagrass meadows, patchy seagrass sand flats, rhodolith pavements, open gravel/sand plains and muddy seafloors. The biotic diversity of Spencer Gulf is remarkably high, considering the elevated seawater salinities. Echinoids and coralline algae (traditionally considered stenohaline organisms) are ubiquitous. Euphotic zone depth is interpreted as the primary control on environmental distribution, whereas seawater salinity, temperature, hydrodynamics and nutrient availability are viewed as secondary controls. Luxuriant seagrass meadows with carbonate muddy sands dominate brightly lit seafloors where waters have relatively low nutrient concentrations (ca 0 to 1 mg Chl‐a m^?3). Low‐diversity bivalve‐dominated deposits occur in meadows with highest seawater salinities and temperatures (43 to 47‰, up to 28°C). Patchy seagrass sand flats cover less‐illuminated seafloors. Open gravel/sand plains contain coarse bivalve–bryozoan sediments, interpreted as subphotic deposits, in waters with near normal marine salinities and moderate trophic resources (0·5 to 1·6 mg Chl‐a m^?3) to support diverse suspension feeders. Rhodolith pavements (coralline algal gravels) form where seagrass growth is arrested, either because of decreased water clarity due to elevated nutrients and associated phytoplankton growth (0·6 to 2 mg Chl‐a m^?3), or bottom waters that are too energetic for seagrasses (currents up to 2 m sec^?1). Muddy seafloors occur in low‐energy areas below the euphotic zone. The relationships between oceanographic influences and depositional patterns outlined in Spencer Gulf are valuable for environmental interpretations of other recent and ancient (particularly Neogene) high‐salinity and temperate carbonate systems worldwide.  相似文献   

Modelling evapotranspiration in an alpine grassland ecosystem on Qinghai‐Tibetan plateau          下载免费PDF全文

Gaofeng Zhu  Yonghong Su  Xin Li  Kun Zhang  Changbin Li  Na Ning 《水文研究》2014,28(3):610-619

Thus far, measurements and estimations of actual evapotranspiration (ET) from high‐altitude grassland ecosystems in remote areas like the Qinghai‐Tibetan plateau are still insufficient. To address these issues, a comparison between the results of the eddy covariance (EC) measurements and the estimates, considering the Katerji and Perrier (KP), the Todorovic (TD) and the Priestley–Taylor (PT) models, was carried out over an alpine grassland (38^o03'1.7'' N, 100^o 27’ 26'' E; 3032 m a.s.l.) during the growing seasons in 2008 and 2009. The results indicated that the KP model after a particularly simple calibration gave the most effective ET values in different time scales, the PT model slightly underestimate ET at night and the TD model significantly overestimated ET at noon. In addition, the canopy resistance calculated by the TD model was completely different from that calculated using the inverted EC‐measured data and the KP model, which may be due to some unrealistic assumptions made by the TD model. The KP parameters were a = 0.17 and b = 1.50 for the alpine grassland and appeared to be interannually stable. However, the PT parameter showed some interannual variations (α = 0.83 and 0.74 for 2008 and 2009, respectively). Therefore, the KP model was preferred to estimate the actual ET at both hourly and daily time scales. The PT model, being the simplest approach and field condition dependent, was recommended when available weather data were rare. On the contrary, the TD model always overestimated the actual ET and should be avoided in case of the alpine grassland ecosystems. Copyright © 2012 John Wiley & Sons, Ltd.  相似文献