Comparison of object-based and pixel-based Random Forest algorithm for wetland vegetation mapping using high spatial resolution GF-1 and SAR data

Vegetation is an integral component of wetland ecosystems. Mapping distribution, quality and quantity of wetland vegetation is important for wetland protection, management and restoration. This study evaluated the performance of object-based and pixel-based Random Forest (RF) algorithms for mapping wetland vegetation using a new Chinese high spatial resolution Gaofen-1 (GF-1) satellite image, L-band PALSAR and C-band Radarsat-2 data. This research utilized the wavelet-principal component analysis (PCA) image fusion technique to integrate multispectral GF-1 and synthetic aperture radar (SAR) images. Comparison of six classification scenarios indicates that the use of additional multi-source datasets achieved higher classification accuracy. The specific conclusions of this study include the followings:(1) the classification of GF-1, Radarsat-2 and PALSAR images found statistically significant difference between pixel-based and object-based methods; (2) object-based and pixel-based RF classifications both achieved greater 80% overall accuracy for both GF-1 and GF-1 fused with SAR images; (3) object-based classifications improved overall accuracy between 3%-10% in all scenarios when compared to pixel-based classifications; (4) object-based classifications produced by the integration of GF-1, Radarsat-2 and PALSAR images outperformed any of the lone datasets, and achieved 89.64% overall accuracy.     

Mapping and monitoring of coastal wetlands of Çukurova Delta in the Eastern Mediterranean region

The aim of this research was to link vegetation characteristics, such as spatial and temporal distribution, and environmental variables, with land cover information derived from remotely sensed satellite images of the Eastern Mediterranean coastal wetlands of Turkey. The research method was based on (i) recording land cover characteristics by means of a vegetation indicator, and (ii) classifying and mapping coastal wetlands utilizing a Landsat Thematic Mapper (TM) image of Çukurova Deltas in Turkey. Vegetation characteristics of various habitats, such as sand dunes, salt marshes, salty plains and afforestation areas, were identified by field surveys. A Landsat TM image of 4 July 1993 was pre-processed and then classified using the Maximum Likelihood (ML) algorithm and Artificial Neural Networks (ANN). As a result of this supervised classification, the land cover types were classified with a largest accuracy of 90.2% by ANN. The classified satellite sensor imagery was linked to vegetation and bird census data, which were available through literature in a Geographical Information System (GIS) environment to determine the spatial distribution of plant and bird biodiversity in this coastal wetland. The resulting data provide an important baseline for further investigations such as monitoring, change detections and designing conservation policies in this coastal ecosystem.     

A Simple Semi-Automatic Approach for Land Cover Classification from Multispectral Remote Sensing Imagery

Land cover data represent a fundamental data source for various types of scientific research. The classification of land cover based on satellite data is a challenging task, and an efficient classification method is needed. In this study, an automatic scheme is proposed for the classification of land use using multispectral remote sensing images based on change detection and a semi-supervised classifier. The satellite image can be automatically classified using only the prior land cover map and existing images; therefore human involvement is reduced to a minimum, ensuring the operability of the method. The method was tested in the Qingpu District of Shanghai, China. Using Environment Satellite 1(HJ-1) images of 2009 with 30 m spatial resolution, the areas were classified into five main types of land cover based on previous land cover data and spectral features. The results agreed on validation of land cover maps well with a Kappa value of 0.79 and statistical area biases in proportion less than 6%. This study proposed a simple semi-automatic approach for land cover classification by using prior maps with satisfied accuracy, which integrated the accuracy of visual interpretation and performance of automatic classification methods. The method can be used for land cover mapping in areas lacking ground reference information or identifying rapid variation of land cover regions (such as rapid urbanization) with convenience.     

An empirical evaluation of spatial value transfer methods for identifying cultural ecosystem services

A significant barrier to the assessment of ecosystem services is a lack of primary data, especially for cultural ecosystem services. Spatial value transfer, also known as benefits transfer, is a method to identify the probable locations of ecosystem services based on empirical spatial associations found in other geographic locations. To date, there has been no systematic evaluation of spatial value transfer methods for cultural ecosystem services identified through participatory mapping methods. This research paper addresses this knowledge gap by examining key variables that influence value transfer for cultural ecosystem services: (1) the geographic setting, (2) the type of ecosystem services, and (3) the land cover data selected for value-transfer. Spatial data from public participation GIS (PPGIS) processes in two regions in Norway were used to evaluate spatial value transfer where the actual mapped distribution of cultural ecosystem values were compared to maps generated using value transfer coefficients. Six cultural ecosystem values were evaluated using two different land cover classification systems GlobCover (300 m resolution) and CORINE (100 m resolution). Value transfer maps based on the distribution of mapped ecosystem values produced strongly correlated results to primary data in both regions. Value transfer for cultural ecosystems appear valid under conditions where the primary data and value transfer regions have similar physical landscapes, the social and cultural values of the human populations are similar, and the primary data sample sizes are large and unbiased. We suggest the use of non-economic value transfer coefficients derived from participatory mapping as the current best approach for estimating the importance and spatial distribution of cultural ecosystem services.     

粤港澳大湾区生态系统服务识别、分类及制图

海岸带生态系统服务识别、分类与制图是合理利用海岸带自然资源,协调海岸带开发与保护矛盾的重要基础。现有生态系统服务分类方法在海岸带应用存在一定的局限性。在前人研究的基础上,以我国城市化和工业化水平较高的粤港澳大湾区为研究区,对该区域海岸带生态系统服务进行识别、分类,并在此基础上使用地图大数据与遥感解译的土地利用数据对海岸带生态系统服务进行了制图。共识别出35种海岸带生态系统服务,并对其中的31种服务进行制图。结果表明,建立的这套方法能较为系统地展示粤港澳大湾区生态系统服务的类型及空间分布特征。具体而言,该区域供给服务和文化服务在城市中心区较为集中,而调节服务多分布于城市周边。对识别的生态系统服务进行综合叠加分析,可将研究区分为文化服务主体区、供给服务主体区、调节服务主体区。建立的海岸带生态系统服务识别、分类体系和制图方法可操作性强,能为我国海岸带生态系统的保育、修复和重建提供科学基础。     

高分辨率影像支持的群落尺度沼泽湿地分类制图

湿地作为众多野生动物和植物的栖息地,具有稳定环境及物种基因保护等重要功能.但是,湿地复杂的水陆交界生境特征及难以进入等客观条件限制给湿地研究造成了很大的困难.因此,遥感技术作为地表生态环境过程参量获取的重要工具,在当今湿地科学领域发挥着重要作用,特别是,当前高空间分辨率影像的性能与应用水平不断得到提高.以自然状态下的黑龙江三江平原洪河国家级自然保护区为研究对象,应用飞艇搭载的空间高分辨率摄像系统获取影像地面分辨率为0.13m的影像数据,主要结合面向对象分类方法,开展了基于湿地植物群落尺度的分类制图研究.结果表明:①因飞艇影像对植物形态、纹理等细致特征的刻画非常充分,沼泽植被型、草甸植被型和各种乔木、灌木植被型,都可以在合适的遥感分类方法下提取出来,总体分类精度能达到91.77％;②通过采用针对高分辨率影像面向对象的分类方法与传统的最大似然比遥感分类方法对比,前者达到很高的精度,而后者效果不理想,说明遥感分类方法的选择对于群落尺度湿地植物分类制图结果非常重要;③遥感分类制图的结果显示出研究区湿地植物群落分布格局受到水分环境梯度和微地貌的共同控制,呈现交替环带状分布规律.     

Linking vegetation type and condition to ecosystem goods and services

Our focus here is on how vegetation management can be used to manipulate the balance of ecosystem services at a landscape scale. Across a landscape, vegetation can be maintained or restored or modified or removed and replaced to meet the changing needs of society, giving mosaics of vegetation types and ‘condition classes’ that can range from intact native ecosystems to highly modified systems. These various classes will produce different levels and types of ecosystem services and the challenge for natural resource management programs and land management decisions is to be able to consider the complex nature of trade-offs between a wide range of ecosystem services. We use vegetation types and their condition classes as a first approximation or surrogate to define and map the underlying ecosystems in terms of their regulating, supporting, provisioning and cultural services. In using vegetation as a surrogate, we believe it is important to describe natural or modified (e.g. agronomic) vegetation classes in terms of structure – which in turn is related to ecosystem function (rooting depth, nutrient recycling, carbon capture, water use, etc.). This approach enables changes in vegetation as a result of land use to be coupled with changes to surface and groundwater resources and other physical and chemical properties of soils.For Australian ecosystems an existing structural classification based on height and cover of all vegetation layers is suggested as the appropriate functional vegetation classification. This classification can be used with a framework for mapping and manipulating vegetation condition classes. These classes are based on the degree of modification to pre-existing vegetation and, in the case of biodiversity, this is the original vegetation. A landscape approach enables a user to visualise and evaluate the trade-offs between economic and environmental objectives at a spatial scale at which the delivery of ecosystem services can meaningfully be influenced and reported. Such trade-offs can be defined using a simple scoring system or, if the ecological and socio-economic data exist in sufficient detail, using process-based models.Existing Australian databases contain information that can be aggregated at the landscape and water catchment scales. The available spatial information includes socio-economic data, terrain, vegetation type and cover, soils and their hydrological properties, groundwater quantity and surface water flows. Our approach supports use of this information to design vegetation management interventions for delivery of an appropriate mix of ecosystem services across landscapes with diverse land uses.     

Structural and functional mapping of geosigmeta in Atlantic coastal marshes (France) using a satellite time series

Geosynphytosociology deals with the study of combinations of vegetation series – or geosigmeta – within landscape. Its main advantage is to assess conservation status based on vegetation dynamics. However, this field-based approach has not been widely applied, because local surveys are not representative of spatio-temporal landscape complexity, which leads to uncertainties and errors for geosigmeta structural and functional mapping. In this context, satellite time series appear as relevant data for monitoring vegetation dynamics. This article aims to assess the contribution of an annual satellite time series for geosigmeta structural and functional mapping. The study area, which focuses on the French Atlantic coast (4630 km²), includes salt, brackish, sub-brackish and fresh marshes. A structural vegetation map was derived from the classification of an annual time series of 38 MODIS images validated with field surveys. The functional vegetation map was derived from the annual Integral of Normalized Difference Vegetation Index (NDVI-I), as an indicator of above-ground net primary production. Results show that geosigmeta were successfully mapped at a scale of 1:250,000 with an overall accuracy of 82.9%. The geosigmeta functional map highlights a strong gradient from the lowest NDVI-I values in salt marshes to the highest values in fresh marshes.     

Classification and Mapping of Coenotic Diversity of Forests

The proposed approach to the study of regularities of spatial variability of plant cover and to mapping forest vegetation is illustrated by the example of European Russia. It is shown that remote sensing and GIS technologies require particular standards of plant cover classification and reflection in maps. The given principles of classification and compilation of explications for maps of forest cover enable an assessment of its status and dynamics and a comparison of materials of different scales. We use the ecological–phytocoenotic approach to classifying forest vegetation. The specified units correspond to the categories of the main classifications of plant cover used in Russian geobotanics. In our classification, we have verified some parameters and the semantics of the mapped units, using satellite images, for their definite identification and interpretation. The elaborated approach to the classification and mapping of forest cover is applied for the study of the diversity of spruce forests under different climatic conditions in two regions, where they occupy about 20% of the total area. The first example characterizes the northern taiga subzone of forests of eastern Fennoscandia in the center of Murmansk oblast, and the second one represents the subzone of broad-leaved–coniferous forest in the southwest of Moscow oblast.     

Landscape-scale regeneration dynamics of disturbed Mediterranean maquis

Abstract. Long-term regeneration dynamics of Mediterranean maquis was investigated by analysing historical aerial photographs of Mt. Carmel, one of the largest protected areas in the Mediterranean region of Israel. Two sets of aerial photographs were processed, one from 1960 (before the area was protected), and the second from 1992 (21 yr after the area was declared a nature reserve). The photographs of each year were classified into three vegetation states based on the percentage cover of trees: open maquis with tree cover < 33.3 %, moderately developed maquis (tree cover 33.3–66.6 %), and closed maquis (tree cover > 66.6 %). Grid maps constructed from the classified images were used to determine probabilities of transition between vegetation states. Closed maquis showed zero probability of transition to either open or moderately developed maquis. Probabilities of ‘forward’ transitions (transitions from low-cover to high-cover classes) were higher on north-facing than on south-facing slopes. On north-facing slopes, the area of open maquis decreased from 87 % to 46 % during the period studied, while that of closed maquis increased from 3 % to 29 %. On south-facing slopes open maquis decreased from 87 % to 69 % while closed maquis increased from 1 % to 8 %. Within a particular aspect, tree cover in 1960 was a reliable predictor of tree cover in 1992. This indicates that micro-scale patterns of tree distribution in 1960 were important in determining the structure of the maquis 32 yr later. Simulations based on the empirically derived transition probabilities suggest that, under current climatic conditions, the process of maquis regeneration on Mt. Carmel may take 500–1000 yr.     

Use of multispectral satellite datasets to improve ecological understanding of the distribution of Invasive Alien Plants in a water-limited catchment,South Africa

Invasive Alien Plants (IAPs) pose major threats to biodiversity, ecosystem functioning and services. The availability of moderate resolution satellite data (e.g. Sentinel-2 Multispectral Instrument and Landsat-8 Operational Land Imager) offers an opportunity to map and monitor the occurrence and spatial distribution of IAPs. The use of two multispectral remote sensing data sets to map and monitor IAPs in the Heuningnes Catchment, South Africa, was therefore investigated using the maximum likelihood classification algorithm. It was possible to identify areas infested with IAPs using remote sensing data. Specifically, IAPs were mapped with a higher overall accuracy of 71%, using Sentinel-2 MSI as compared to using Landsat 8 OLI, which produced 63% accuracy. However, both sensors showed similar patterns in the spatial distribution of IAPs within the hillslopes and riparian zones of the catchment. This work demonstrates the utility of the two multispectral data sets in mapping and monitoring the occurrence and distribution of IAPs, which contributes to improved ecological modelling and thus to improved management of invasions and biodiversity in the catchment.     

Dealing with vagueness in complex forest landscapes: A soft classification approach through a niche-based distribution model

The increasing interest in biodiversity conservation has led to the development of new approaches to facilitate ecologically based conservation policies and management plans. In this context, the development of effective methods for the classification of forest types constitutes a crucial issue as forests represent the most widespread vegetation structure and play a key role in ecosystem functioning. In this study a maximum entropy approach (Maxent) to forest type classification in a complex Mediterranean area, has been investigated. Maxent, a niche-based model of species/habitat distribution, allowed researchers to estimate the potential distribution of four forest types: Holm oak, Mixed oak, Mixed broadleaved and Riparian forests. The Maxent model's internal tests have proved a powerful tool for estimating the model's accuracy and analyzing the effects of the most important variables in the produced models. Moreover the comparison with a spectral response-based fuzzy classification, showed a higher accuracy in the Maxent outputs, demonstrating how the use of environmental variables, combined with spectral information in the classification of natural or semi-natural land cover classes, improves map accuracies. The modeling approach followed by this study, taking into account the uncertainty proper of the natural ecosystems and the use of environmental variables in land cover classification, can represent a useful approach to making more efficient and effective field inventories and to developing effective conservation policies.     

粤港澳大湾区城市群生态系统智能化管理

开展生态系统数字化、信息化、智能化管理,全面提升粤港澳大湾区生态环境质量,是建设国际一流湾区的必然趋势。以城市群生态系统智能化管理为目标,系统整合各类生态环境相关数据资源,形成生态系统管理数据和决策支撑体系,并以此为基础构建生态智能管理平台。研究以生态系统要素和功能管理逻辑为核心,构建了生态系统管理业务流程：（1）精准剖析生态环境问题,确定问题发生的尺度、范围并对其进行分类和定性;（2）确立生态管理目标,制定适宜的管理策略;（3）根据现状与基线进行生态系统服务权衡,通过生态管理恢复工程提升生态系统质量;（4）通过环境物联网监测生态系统变化,及时调整和改进生态系统管理计划。针对城市群生态系统多尺度、多层次、复杂化等特点,在制定管理决策时应充分权衡管理目标和生态服务,兼顾各类生态系统服务效益;需通过示范性生态工程印证管理方案的可行性、适用性和协同性;以趋善化理念为指导思想,不断优化调整生态管理目标;同时在管理活动实施的过程中不断积累、凝练、总结所获得的反馈信息和经验。面向生态管理体制和管理能力的现代化提升需求,融合大数据、地理信息系统（GIS）、全球广域网络（Web）等信息化技术,构建粤港澳大湾区生态管理智能平台,实现多主体信息共享,打破管理决策的"黑箱",为推进生态环境管理现代化提供可靠可行的方案。构建的生态系统管理业务流程和管理策略,将知识充分融入管理决策的制定流程,能服务于粤港澳大湾区的生态文明建设,推动可持续和高质量发展。     

Species classification from hyperspectral leaf information using machine learning approaches

Information on plant species is fundamental to forest ecosystems, in the context of biodiversity monitoring and forest management. Traditional methods for plant species inventories are generally inefficient, in terms of cost and performance, and there is a high demand for a quick and feasible approach to be developed. Of the various attempts, remote sensing has emerged as an active approach for plant species classification, but most studies have concentrated on image processing and only a few of them ever use hyperspectral information, despite the wealth of information it contains. In this study, plant species are classified from hyperspectral leaf information using different machine learning models, coupled with feature reduction and selection methods, and their performance is optimized through Bayesian optimization. The results show that including feature selection and Bayesian optimization increases the classification accuracy of machine learning models. Among these, the Bayesian optimization-based support vector machine (SVM) model, combined with the recursive feature elimination (RFE) feature selection method, yields the best output, with an overall accuracy of 86% and a kappa coefficient of 0.85. Furthermore, the confusion matrix revealed that the number of samples correlates with classification accuracy. The support vector machine with informative bands after Bayesian optimization outperformed in classing plant species. The results of this study facilitate a better understanding of spectral (phenotype) information with plant species (genotype) and help to bridge hyperspectral information with ecosystem functions.     

On ecology of post-fire soil fungi: assessing impact of disturbance using species-abundance models as measure of community organization

Mediterranean ecosystems are among those most significantly modified by fires. Such fires lead to evident disturbance of the above- and below-ground ecosystem components, at various spatial and temporal scales, including soil microfungi. The ecological parameters used to measure the effects of disturbance on soil fungal communities include species-abundance distribution patterns, which reflect changes in the relationships between species numbers and their relative abundance, and serve as a critical measure of community organization. Species-abundance distribution patterns were used to assess the disturbance impact of experimental fires on soil fungal communities in Mediterranean maquis (southern Italy) in the short- to mid-term. The trend in the distribution patterns of heat-stimulated and xerotolerant soil fungi over time, their varying responses to low- and high-intensity fire, the efficiency of the soil germplasm bank, and the pivotal role of Neosartorya spp. in post-fire community structure in Mediterranean burned soils may all be used as tools to accurately assess the effects of fire on soil mycobiota.     

Landscape‐scale determinants of non‐native fish communities

Aim Predicting and preventing invasions depends on knowledge of the factors that make ecosystems susceptible to invasion. Current studies generally rely on non‐native species richness (NNSR) as the sole measure of ecosystem invasibility; however, species identity is a critical consideration, given that different ecosystems may have environmental characteristics suitable to different species. Our aim was to examine whether non‐native freshwater fish community composition was related to ecosystem characteristics at the landscape scale. Location United States. Methods We described spatial patterns in non‐native freshwater fish communities among watersheds in the Mid‐Atlantic region of the United States based on records of establishment in the U.S. Geological Survey’s Nonindigenous Aquatic Species Database. We described general relationships between non‐native species and ecosystem characteristics using canonical correspondence analysis. We clustered watersheds by non‐native fish community and described differences among clusters using indicator species analysis. We then assessed whether non‐native communities could be predicted from ecosystem characteristics using random forest analysis and predicted non‐native communities for uninvaded watersheds. We estimated which ecosystem characteristics were most important for predicting non‐native communities using conditional inference trees. Results We identified four non‐native fish communities, each with distinct indicator species. Non‐native communities were predicted based on ecosystem characteristics with an accuracy of 80.6%, with temperature as the most important variable. Relatively uninvaded watersheds were predicted to be invasible by the most diverse non‐native community. Main conclusions Non‐native species identity is an important consideration when assessing ecosystem invasibility. NNSR alone is an insufficient measure of invasibility because ecosystems with equal NNSR may not be equally invasible by the same species. Our findings can help improve predictions of future invasions and focus management and policy decisions on particular species in highly invasible ecosystems.     

Differentiating geological fertility derived vegetation zones in Kruger National Park,South Africa,using Landsat and MODIS imagery

Spatial technologies present possibilities for producing frequently updated and accurate habitat maps, which are important in biodiversity conservation. Assemblages of vegetation are equivalent to habitats. This study examined the use of satellite imagery in vegetation differentiation in South Africa's Kruger National Park (KNP). A vegetation classification scheme based on dominant tree species but also related to the park's geology was tested, the geology generally consisting of high and low fertility lithology. Currently available multispectral satellite imagery is broadly either of high spatial but low temporal resolution or low spatial but high temporal resolution. Landsat TM/ETM+ and MODIS images were used to represent these broad categories. Rain season dates were selected as the period when discrimination between key habitats in KNP is most likely to be successful. Principal Component Analysis enhanced vegetated areas on the Landsat images, while NDVI vegetation enhancement was employed on the MODIS image. The images were classified into six field sampling derived classes depicting a vegetation density and phenology gradient, with high (about 89%) indicative classification accuracy. The results indicate that, using image processing procedures that enhance vegetation density, image classification can be used to map the park's vegetation at the high versus low geological fertility zone level, to accuracies above 80% on high spatial resolution imagery and slightly lower accuracy on lower spatial resolution imagery. Rainfall just prior to the image date influences herbaceous vegetation and, therefore, success at image scene vegetation mapping, while cloud cover limits image availability. Small scale habitat differentiation using multispectral satellite imagery for large protected savanna areas appears feasible, indicating the potential for use of remote sensing in savanna habitat monitoring. However, factors affecting successful habitat mapping need to be considered. Therefore, adoption of remote sensing in vegetation mapping and monitoring for large protected savanna areas merits consideration by conservation agencies.     

What is the Value of a Good Map? An Example Using High Spatial Resolution Imagery to Aid Riparian Restoration

Riparian areas contain structurally diverse habitats that are challenging to monitor routinely and accurately over broad areas. As the structural variability within riparian areas is often indiscernible using moderate-scale satellite imagery, new mapping techniques are needed. We used high spatial resolution satellite imagery from the QuickBird satellite to map harvested and intact forests in coastal British Columbia, Canada. We distinguished forest structural classes used in riparian restoration planning, each with different restoration costs. To assess the accuracy of high spatial resolution imagery relative to coarser imagery, we coarsened the pixel resolution of the image, repeated the classifications, and compared results. Accuracy assessments produced individual class accuracies ranging from 70 to 90% for most classes; whilst accuracies obtained using coarser scale imagery were lower. We also examined the implications of map error on riparian restoration budgets derived from our classified maps. To do so, we modified the confusion matrix to create a cost error matrix quantifying costs associated with misclassification. High spatial resolution satellite imagery can be useful for riparian mapping; however, errors in restoration budgets attributable to misclassification error can be significant, even when using highly accurate maps. As the spatial resolution of imagery increases, it will be used more routinely in ecosystem ecology. Thus, our ability to evaluate map accuracy in practical, meaningful ways must develop further. The cost error matrix is one method that can be adapted for conservation and planning decisions in many ecosystems.