采用全卷积神经网络与Stacking算法的湿地分类方法

高精度湿地制图对湿地生态保护与精细管理具有重要的支撑作用。针对传统湿地分类方法的精度不高和泛化能力弱等问题，提出了一种联合全卷积神经网路（fully convolutional neural network，FCN）与集成学习的湿地分类方法。首先利用全卷积神经网络（SegNet、UNet及RefineNet）对GF-6影像的语义特征进行提取与融合，然后利用Stacking集成算法对融合后的特征进行判别和分类。结果表明，联合全卷积神经网络与Stacking算法能有效提取湿地信息，总体分类精度为88.16%，Kappa系数为0.85。与联合全卷积神经网络与单一机器学习RF、SVM与kNN算法相比，该文提出的湿地分类方法在总体分类精度上分别提高了4.87%，5.31%和5.08%；与联合单一全卷积神经网络（RefineNet、SegNet、UNet）与Stacking算法下的湿地分类结果，该文提出的湿地分类方法在总体分类精度上分别提高了2.78%，4.48%与4.91%；该文方法一方面能通过卷积神经网络提取遥感影像深层的语义特征，另一方面通过集成学习根据各分类器的表征性能进行合理的选择并重组，从而提高分类精度及其泛化能力。该方法能为湿地信息提取及土地覆盖分类方法的研究提供参考。     

基于GF-1卫星数据与面向对象分类的达里诺尔湿地自然保护区土地覆盖信息提取

以达里诺尔湿地自然保护区为研究区，基于国产GF-1遥感影像，采用面向对象和传统目视解译的分类方法对研究区土地覆盖遥感信息进行提取，并对其结果进行对比分析，采取混淆矩阵对面向对象分类结果进行精度验证。结果表明：（1）充分利用了GF-1遥感影像的光谱信息，面向对象分类采取试错法确定最优分割尺度为550，形状和紧致度因子分别为0.6和0.5，各波段权重均为1。（2）面向对象分类总体分类精度达98.22％，KAPPA系数为0.96；（3）面向对象分类方法可快速准确提取类型较为复杂的土地覆盖信息，为内陆湿地精准快速提取研究区土地覆盖分类信息提供参考，以期为湿地遥感业务化监测提供技术规范。     

宝鸡金渭湖湿地小气候效应初探

以宝鸡市金渭湖湿地为研究对象,在金渭湖拦河闸桥附近进行气象及大气颗粒物观测,将拦河闸桥看作湿区,公路旁看作干区,对比分析湿地小气候特征的结果表明:宝鸡市金渭湖湿地具有明显的降温增湿作用,距离湿区越远,温度越高,相对湿度越低;冬季的相对湿度大于夏季,冬夏季的湿区温度始终低于干区,干湿两区的温度和相对湿度的日变化规律基本一致;湿区的PM2.5质量浓度小于干区,最多可小47%,表明金渭湖湿地对局地大气环境中的细颗粒物质量浓度具有显著削减作用。     

游客视角的洋湖湿地公园游憩空间优化设计

湿地公园不仅在生态环境保护方面发挥着重要的作用,同时还能为人类提供休闲游乐场地,目前对湿地公园游憩空间的研究越来越重要。通过问卷调查研究设计了洋湖湿地公园游憩空间评价指标体系,采用主成分分析法进行因子分析,对问卷调查结果进行配对样本T检验,结果显示游客对于洋湖湿地公园游憩空间体系中17个指标的重要性和满意度评价存在明显的差异性。运用IPA修正模型将洋湖湿地公园游憩空间建设项目分为4种类型,并确定17个指标在4种类型中的定位。最后依据IPA修正模型的分析结果,提出相应的游憩空间优化策略。     

关于建立庄河口湿地自然保护区的意见

在考察庄河口湿地现状基础上，分析湿地生态系统存在着管理、过度开发、环境污染等问题，阐述湿地生态系统保护意义。从保护类型、对象、级别、功能区划分、工程建设等方面提出拟建保护区的基本框架规划。     

Effects of silvicultural activity on ecological processes in floodplain forests of the southern United States: a review of existing reports

Activities associated with timber harvesting have occurred within floodplain forests in the southern United States for nearly two hundred years. However, it is only in the last ten years that any information has become available about the effects of harvesting on the ecological functions of this valuable resource. Hydrology is the driving influence behind all ecological processes in floodplains, and timber harvesting alone usually has little long-term effect on hydroperiod. However, logging roads, built in association with harvest sites, can sometimes alter hydroperiod to the extent that vegetation productivity is raised or lowered. There is no evidence that harvesting followed by natural regeneration represents a threat to ground or surface water quality on flood plain sites, as long as “best management practices” are followed. Harvested floodplains may increase or have little effect on decomposition rates of surface organic matter. The nature of the effect seems to be controlled by site wetness. Data from recently harvested sites (i.e. within the last ten years) suggest that vegetation productivity is maintained at levels similar to those observed prior to harvests. During the early stages of stand development, tree species composition is heavily influenced by harvest method. Similarly, amphibian populations (monitored as bioindicators of ecosystem recovery) seem to rebound rapidly following harvests, although species composition may be different from that of unharvested stands.     

3S技术在中国湿地研究、调查与管理中的应用

This paper introduces the state of waterlands in China and discribes the applications of Remote Sensing （RS）, Geographical Information System （G1S） and Global Positioning System （GPS） in wetland research, including land-coverclassification and change detection, wetland evolutionary processes, landscape-change analyses, channel migration, flood and wetlands resource monitoring and spatial quantitative analyses/modeling, ecosystem service evaluation, ecological processes and risk assessments, disease control, water quality monitoring/modeling, pollution monitoring/modeling, wetlands hydrology, wetland information systems and WebGIS. The limitations and needs for optimal use of these technologies are discussed, such as the limited advanced technical knowledge and skills, low awareness and capacity, unclear link between GIS output and policy making, lack of supporting policies and standards, lack of a wetlands geo-information networklimite, and the use of these techniques in wetland research. It is suggested that for realising true applications of RS, GIS and GPS technologies, the availability, accessibility, reliability, homogeneity, and continuity of wetlands-related geo-information enabling environment, policies and standards, and funding are needed.     

Sedimentation associated with forest road surfacing in a bottomland hardwood ecosystem

Access systems are a necessary element of resource production in bottomland hardwood sites. However, road building may have a detrimental effect on hydrologic function of the site. This report describes initial results of a study designed to examine the effect of different road surfacing treatments on water quality.

Four surfacing treatments installed on two test roads included native soil, native soil with vegetative stabilization, 6 cm of gravel, and 15 cm of gravel over geotextile. During the first flooding season periodic sampling measured floodwater suspended sediments and location of erosion and sediment deposition within the road prism. Initial results suggest that sediment movement was confined to the road right-of-way, with no statistically significant sedimentation effects detected beyond the clearing limits of the road. The study is continuing for another field season.     

鄱阳湖水位特征与湿地生态保护

从分析江湖关系入手，研究鄱阳湖的水位及湿地生态特征，并提出人工控湖等加强湿地生态保护的对策建议。     

Factors influencing stump sprouting by pondcypress (Taxodium distichum var. nutans (Ait.) Sweet)

Sprouting can be an important mechanism of regeneration in harvested cypress (Taxodium sp.) wetlands; however, limited information exists on factors influencing sprouting. We examined the influence of harvesting practices, tree size, and site conditions on stump sprouting in eight pondcypress (Taxodium distichum var. nutans (Ait.) Sweet) wetlands in Florida, USA. Stumps were cut at three heights: 20 cm below high water level, at high water, and 40 cm above high water. Half of each site was harvested using a sawhead feller and the other half using shearhead. We monitored sprouting for 2 years post-harvest. Stump height affected first year sprouting, but the effect disappeared during year 2. Felling equipment did not affect sprouting; however, the shearhead caused greater stump damage. Sprouting declined with increasing parent tree size; percentage of stumps with sprouts after 2 years was 68% for trees 10–14 cm dbh and 28% for trees 20–24 cm dbh. Stump sprouting ranged among sites from 23 to 54% after 2 years. Results indicate that sprouting may be an important, but inadequate, form of regeneration. At sites with short hydroperiods, dense understories, and/or predominance of trees greater than 20 cm dbh, reduced sprouting may occur, and planting seedlings may be beneficial.