Zoning adjustments are a key method of improving the conservation efficiency of a nature reserve. Existing studies typically consider the one-period programming method and ignore dynamic ecological changes during the programming of a nature reserve. In this study, a scientific method for nature reserve (NR) programming, namely the multiperiod dynamic programming (MDP) algorithm, is proposed. The MDP algorithm designs an NR over three periods and does so by using ecological suitability values for each grid area. Ecological suitability values for each period were determined based on existing data on rare aquatic animals with Maxent software and cellular automata (CA). CA were used to determine the actual protection effect and to adjust each period’s ecological suitability values through comparisons with the sites’ surroundings. The maximization of ecological suitability values was used as an objective function; these values were assumed to indicate protection benefits. The objective function of the MDP also includes grid perimeters and numerical minimization for spatial compactness. Moreover, we designed three MDP constraints for the dynamic programming, including base constraints, distinguishing constraints, and multiperiod constraints. In the base and distinguishing constraints, we require a grid square to be a core, buffer, or unselected square, and we require the core and buffer grids to be spatially connected. For the multiperiod constraints, we used virtual points to ensure spatial continuity in different periods while attaining high ecological suitability. Our main contributions are as follows: (1) the novel MDP algorithm combining ecological attributes and multiperiod dynamic planning to optimize NR planning; (2) the use of virtual points to avoid selecting invalid grids and to ensure spatial continuity with significant protection benefits; and (3) the definition of ecological suitability values and use of CA to simulate dynamic changes over the three periods. The results reveal that the MDP algorithm results in a reserve with greater protection benefits than current reserves with superior spatial distribution due to multiperiod programming. The proposed MDP algorithm is a novel method for the scientific optimization and adjustment of nature reserves.

[目的]传统的森林生物多样性保育价值缺乏统一的评估标准,不便于结果比较.目前,物种保育价值模型主要通过林分丰富度核算森林当前保育价值,但未考虑土地利用类型对树种生长的潜在影响,不能体现森林的长期保育价值.近年来,逐步完善的生物多样性价值评价新理论中,尚未出现适用于森林生物多样性保育价值评估的方法.[方法]通过分析福建土地利用情况计算生境质量并划分等级,首次提出生境质量调整系数.利用生境质量和森林分布之间的地理对应关系,构建出基于生境质量的森林生物多样性保育价值模型(HQ-BPV模型),由此将土地利用的影响融入福建省森林生物多样性保育价值评估中.[结果](1)福建省生境质量分布整体呈现内北高南低、内陆高沿海低和自然保护区高非自然保护区低的态势.(2)福建主要植被类型的平均生境质量按大小排序为针阔混交林＞竹林＞阔叶混交林＞阔叶林＞灌丛＞针叶林＞经济林.(3)调整后栎类和栲类的物种单位面积保育价值最高,为3.58万元/hm2,调整后各优势树种(组)的物种保育总值中阔叶混交林价值最高,为658.42亿元,构建模型核算的福建省森林生物多样性保育价值为2056.39亿元.[结论]基于能够量化土地利用影响的生境质量指数,构建出HQ-BPV模型用于评价福建省森林生物多样性保育价值,为森林生物多样性的保育价值核算提供新思路,有利于森林资源监测和结构优化.

使用低空遥感图像进行图像识别为森林调查和监测提供了新的技术契机.基于无人机低空航拍光学图像,以福建省安溪县崩岗区为研究区,建立FC-DenseNet模型进行树种识别.首先,利用Dense模块提取树种图像特征并增强深层网络信息,透过下采样模块降低图像维度,凸显图像的纹理特征和光谱特征;然后,使用上采样模块还原预测图至原始图像大小,并融合浅层Dense模块信息的丰富特征;最后,采用Softmax分类器实现像素分类,完成树种识别.结果显示,基于低空航拍光学图像,FC-DenseNet模型能够准确区分植被与非植被,定位其空间分布特征,其中,FC-DenseNet-103模型的二分类识别精度为92.1％,表明FC-DenseNet模型加深网络深度后具有较好的识别效果;将植被与非植被细分为13类,FC-DenseNet-103模型的平均识别正确率达到75.67％.研究结果表明,基于低空航拍光学图像建立的FC-DenseNet模型具有较高的树种分类精度.由于低空航拍光学图像的成本较低,数据获取费用小,时间周期短,可便于森林资源调查和森林树种检测,为深度学习在树种识别领域的应用提供了新思路.

科学合理的自然保护区功能分区有利物种长期生存繁衍和自然保护区的管理.现阶段的自然保护区功能分区方法中没有考虑自然保护区的空间特性和生物多样性的保护效益,尚未形成一套系统的保护区分区模型.为此,本研究基于Maxent模型模拟珍稀物种的分布,提出基于网络流的保护区功能分区模型,划分出一个连续和紧实的自然保护区功能区,同时考虑保护区保护效益最大化.为验证模型的可行性,将泉州湾河口湿地自然保护区划分为5969个候选单元,以珍稀水生动物为保护对象,选出方格的最优组合作为自然保护区规划结果,并结合原保护区提出保护区优化方案.模型规划核心区和缓冲区具有连续性和紧实性,且核心区被缓冲区所包围,其中,核心区选中491个候选单元,其生境总适宜值远高于原保护区;结合原保护区进行优化,以原保护区为中心,向正东方向拓展,核心区候选单元数拓展至614个,缓冲区候选单元数为184个.研究设计了新的功能分区整数规划模型,结合泉州湾河口湿地自然保护区提出新的优化建议,由原保护区向正东方向拓展,覆盖珍稀水生动物的集中区域,有效保护珍稀水生动物及其生境,为我国自然保护区功能分区理论和实践提供新的思路.

Plant classification is a science that is used to assess the quality of forest resources and has been studied extensively. In this paper, we proposed a novel convolutional neural network known as the Fourier Dense Network (FDN) which is a data-driven method to classify the optical aerial images of plants. To efficiently classify plants, the FDN learns and extracts the features of plants in time and frequency domains from the optical images captured using an unmanned aerial vehicle (UAV). In FDN, we designed a fast Fourier dense block (FF-dense block) that describes the features of the plants by using the magnitude and phase information in the frequency domain. Moreover, we used a transition layer to reduce the dimension of feature maps between two FF-dense blocks in the time domain. The primary contributions of this study are as follows: (1) an FF-dense block that considers frequency information and transfers the information into various layers of a block was designed; and (2) the characteristics between the time and frequency domains were repeatedly extracted and combined to more effectively describe the characteristics of tree species. To evaluate our study, we established a novel dataset comprising the UAV-based optical images of plants-vegetational optical aerial image dataset-for conducting plant classification and information retrieval. The dataset contains more than 21863 images of 12 plants. To the best of our knowledge, this is the largest publicly available dataset of the UAV-based optical images of plants. The experimental results demonstrated that the FDN can achieve state-of-the-art performance in terms of plant classification.

Establishing the conservation reserves is an efficient approach to protect various species, in which maintaining the spatial characteristics of the habit is an important issue. A continuous and compact reserve is vital for long-term survival and reproduction of various species. However, the existing methods for nature reserve design have the problem of the selected sites which are sparse in space. The sparseness refers to the isolated regions which are the unselected (selected) sites and surrounded by the selected (unselected) sites in the designed nature reserve and thus making the management of protected areas and exchange of information between species difficult. In this paper, we propose the concept of a dual-flow mechanism which is called the dual-flow structure (DFS) to address the problem of sparseness in sites selected for the reserve network design. The DFS constructs two consecutive arcs: one is for selected sites and the other is for unselected sites, simultaneously. Those two arcs ensure that each site has at least one adjacent site which has the same status (selected/unselected) with the competitive mechanism. The concept of two consecutive arcs can solve the problem of sparseness and achieve spatial continuity. The DFS further utilizes the perimeter in dual-flow structure to achieve the spatial characteristic of compactness. The main contributions of this study are: (1) we design a novel mathematical model which solves the problem of sparseness in the space, including selected and un-selected sites; and (2) the DFS generates the connectivity and compactness of the reserve network with the scientific and reasonable techniques. In the experiments, we take the Daiyun Mountains as a study area to demonstrate the feasibility and effectiveness of the proposed model.

利用无人机航拍获得光学影像数据,结合深度学习理论,建立树种识别模型,以期为大规模树种识别提供一种新的方式.首先以福建安溪县为例,采用无人机获取20 m及40 m高度的航拍影像.其次,以树种为对象,对航拍影像进行分割,获得12种树种影像.最后,结合深度学习理论,采用DenseNet卷积神经网络建立树种识别模型,探讨不同航拍高度以及不同网络深度对树种识别的影响.结果表明:不同航拍高度的树种识别模型,其分类精度均达80％以上,最高精度为87.54％.从航拍影像解析度分析,随着航拍影像解析度的下降,模型识别精度呈现下降趋势,以20 m航拍影像数据建构的树种识别模型,其分类精度高于40 m模型;从模型网络深度分析,随着模型网络层数的增加,模型分类精度出现下降现象,DenseNet121模型分类精度高于DenseNet169模型分类精度.综上所述,基于无人机航拍影像,结合深度卷积神经网络,提出了新的树种识别方式,并以安溪县森林树种识别为例证明了该分类框架的有效性.

使用无人机进行低空航拍,快速取得大范围的植被图像,结合多元HoG特征进行植被类型识别.首先,利用Gabor滤波器提取图像的纹理信息,HSV和Lab颜色空间转化提取图像的颜色信息.其次,将图像分割为N个单元格(cell),基于纹理与颜色信息计算每个单元格的方向梯度直方图(HoG)特征,形成多元HoG特征.最后,以单元格为分类单位,结合随机森林机器学习算法,建立植被类型识别模型.以福建省安溪县山区为研究区域,结果表明:利用无人机低空航拍的光学影像结合多元HoG特征进行植被类型识别是可行的;对于植被与非植被识别,其最高分类正确率达到96.04％;20 m航拍下,植被类型识别率最高,为82.44％,随着航拍高度的升高,模型识别效果呈现下降趋势.进一步采集福建省长汀县山区的植被航拍影像为测试数据,证明模型对于不同地区植被类型识别的稳定性,其识别精度最高可达73.31％,正确率无显著差异.本研究采用无人机载光学相机获取植被光学图像数据,数据获取方便且所需费用较低;提出的植被类型识别模型具有较高的精度;对于不同地区的植被类型识别具有较好的稳健性,可方便应用于野外森林树种监控与管理.根据不同高度模型识别结果,航拍高度不宜过高,航拍高度以20 m为宜.

[目的]保护区规划设计是自然保护区研究的重要内容,以数学模型为工具研究保护区设计成为新趋势.集合覆盖物种模型(SCSP)和最大集合物种模型(MCSP)等传统数学模型,仅考虑以最小代价进行保护区规划及设计,未考虑保护区的空间特征,导致选择的保护区区域过于分散,且仅考虑土地市场价格,不能完全体现保护区生态价值.本研究以生态值为衡量指标,结合连续性和紧实性的空间特征,构建有效合理的保护区规划模型,以期为自然保护区建立提供科学依据.[方法]以福建省戴云山国家级自然保护区为例,将戴云山区域划分为567个规则地块,每个地块面积为2 km×2 km,且任意地块都包含若干个小班.依据生态值赋分标准和重心算法,先后计算戴云山小班和其所属地块生态值.结合特殊空间特征与生态值,构建基于尾长法的空间集合覆盖模型(SSCP),并探讨不同紧实性权重和物种保护比例组合对保护区规划结果的影响.最后,以传统SCSP模型、系统保护规划工具Marxan模型和Zonation模型的规划结果为对照组,验证SSCP模型的有效性.[结果]从空间分布来看,用SCSP模型、Marxan模型和Zonation模型求解的保护区设计结果表现为地块分布离散,破碎度高,SSCP模型的设计结果表现出更好的连续性和紧实性;从选地数量来看,SSCP模型的选地数量与物种保护比例及紧实性权重正相关,即物种保护比例上升,或紧实性权重增大,选中的地块数增加.[结论]研究设计了新型保护区数学规划模型——SSCP模型,并以戴云山自然保护区为例验证了算法的合理性,提出规划建议:向其西北方向和东南方向扩展,其中西北方向纵向扩展6 km,横向扩展18 km,东南方向纵向扩展9 km,横向扩展8 km.本研究方法的提出为我国保护区规划设计理论与实践提供新思路.

基于SEEA 2012体系提出符合中国国情的生态系统生产总值(GEP)核算体系,结合调研数据,通过灰色预测模型,对漳江口红树林保护区2004—2014年和2015—2019年的生态系统生产总值进行核算与估算.结果表明:漳江口红树林保护区的红树林面积逐年增加,2004—2014年生态系统生产总值保持上升趋势,生态系统更加稳定;2015—2019年漳江口红树林保护区生态系统生产总值有明显上升趋势.

为有效实施植被信息获取及监测,亟需分类准确及易于推广的植被信息识别技术.本文利用无人机航拍获取植被光学影像,利用深度语义分割技术建构植被种类识别模型,为植被变化动态监测提供准确的植被类别信息.首先,基于安溪县龙门镇崩岗区的采样点,获取20 m航拍高度的无人机影像,构建FCN-VGG19植被识别模型,探讨不同特征融合结构对FCN-VGG19识别性能的影响,测算出各植被的覆盖面积;其次,取安溪县另一取样点的无人机影像作为验证集,分析FCN-VGG19的迁移学习能力,验证模型稳健性.结果表明:(1)基于20 m高度的无人机影像建立的FCN-VGG19-8s模型识别正确率最高,为86.30％;(2)FCN-VGG19-8s识别精度高于FCN-VGG19-32s;并从测试集中随机抽取一张图,测算该测试图的马尾松覆盖面积为78.38 m2,芒萁覆盖面积为12.77 m2,柠檬桉覆盖面积为0.89 m2;(3)在模型的迁移学习能力试验分析中,当A数据集占训练集的比例下降时,对模型识别B数据集的影响不大;当B数据集的数据量减少时,其识别精度稍有下降,仍有84.5％.本文基于无人机光学影像,结合深度语义分割模型进行植被识别,以福建安溪县为例验证模型稳健性,分析模型在测算植被覆盖面积的适用性,旨在为植被识别研究提供新思路.

Fujian Province is the first pilot area of ecological civilization in China. One of its important tasks is to promote the ecologically fragile area to construct the ecological compensation mechanism. Mangrove wetland is a coastal ecologically critical area and its natural distribution is shrinking. Therefore, it had great challenge in wetland protection and restoration and required the government to have a higher management function. Taking Fujian Zhangjiangkou mangrove as an example, the significance in the exploration on mangrove wetland ecological compensation mechanism was very important. In this paper, we analyzed the concept of the wetland ecological compensation mechanism from three aspects: 1 ) Analysis of the ecological compensation mechanism was based on DPSIR model. 2 ) After field investigation, the residents' basic status and the willingness of ecological compensation in Zhangjiangkou mangrove wetland were surveyed by contingent valuation method( CVM) , and the influencing factors were analyzed by establishing the Tobit model. 3)Three kinds of ecological compensation standard scheme for different regions were discussed. The results showed that the surrounding residents had a low level of education and their environmental awareness was weak. These residents were preferable to select land compensation, employment arrangements and infrastructure construction on the premise of non-cash compensation. The willingness compensation between different villages had notable difference, and that was very significantly affected by respondents on education, population of family, acknowledge of mangrove wetland ecological benefit and the satisfaction of an ecosystem.