Vegetation vulnerability to hydrometeorological stresses in water-scarce areas using machine learning and remote sensing techniques

The quantitative understanding of vegetation vulnerability as a major example of terrestrial ecosystems under hydrometeorological stress is essential for environmental risk preparedness and mitigation strategies. The aim of this study was to develop a new quantitative vegetation vulnerability map using benchmark and standalone machine learning (ML) algorithms (e.g., RF, SVM and Maxent), as well as influencing variables (evaporation, rainfall, maximum temperature, slope degree, elevation, topographic wetness index, distance from river, aspect, land use), in the South Baluchistan basin, Iran. An ensemble model was developed based on selected standalone ML algorithms. A vegetation vulnerability index (VVI), based on remote sensing indices (NDVI, VCI, LST, and TCI), was used to monitor vegetation conditions and changes. Five evaluation metrics for the confusion matrix (accuracy, precision, bias, Probability of Detection (POD), False Alarm Ratio (FAR)) and ROC-AUC were used to measure the predictive performance of the ensemble model and VVI. The optimum values for accuracy, precision, bias, POD, FAR, and ROC-AUC were obtained as 0.89, 0.88, 1.02, 0.91, 0.11, and 0.946, respectively for the ensemble model. Based on remote sensing data, VVI achieved a 0.923 prediction rate in vegetation vulnerability mapping (the efficiency of the ensembled model was somewhat better than VVI). Based on the results obtained from the ensemble model, precipitation (PRD = 20.61), maximum temperature (PRD = 12.31), evaporation (PRD = 5.53), and distance from the river (PRD = 2.62) were found to be the most important variables. The methodology as presented in this study provides valuable information in a large area and can be easily modified for other case studies by adding different influencing variables.     

Machine learning based wildfire susceptibility mapping using remotely sensed fire data and GIS: A case study of Adana and Mersin provinces,Turkey

In recent years, the number of wildfires has increased all over the world. Therefore, mapping wildfire susceptibility is crucial for prevention, early detection, and supporting wildfire management decisions. This study aims to generate Machine Learning (ML) based wildfire susceptibility maps for Adana and Mersin provinces, which are located in the Mediterranean Region of Turkey. To generate a wildfire inventory, this study uses active fire pixels derived from MODIS monthly MCD14ML composites. Furthermore, as a sub aim, the performance of seven ML approaches, namely, stand-alone Logistic Regression (LR), Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), and ensemble algorithms, namely Random Forest (RF), Gradient Boosting (GB), eXtreme Gradient Boosting (XGB), and AdaBoost (AB), was evaluated based on wildfire susceptibility mapping. The capabilities of the corresponding ML methods were assessed using thirteen wildfire conditioning factors, which can be grouped into four main categories: topographical, meteorological, vegetation, and anthropogenic factors. The Information Gain (IG) approach was used to assess their importance scores. A multicollinearity analysis was also performed to assess the relationship between conditioning factors. To compare the predictive performances of ML algorithms, five performance metrics, namely average accuracy, precision, recall, F1 score, and area under the curve, were used. To test the significance of the generated wildfire susceptibility maps and to detect similarities and differences among the output of these ML algorithms, McNemar's test was implemented. In the end, the ML-based models were locally interpreted using the Shapley Additive exPlanations (SHAP) technique. The AUC values of seven methods varied from 0.817 to 0.879, and the accuracy scores ranged between 0.734 and 0.812. The results showed that the RF model provided the best results considering all performance metrics. The accuracy score and AUC values of the RF model were equal to 0.812 and 0.879, respectively. On the other hand, stand-alone algorithms (LDA, SVM, and LR) represented lower performance than tree-based ensemble methods. Both the IG and SHAP analyses showed that elevation, temperature, and slope factors were the most contributing factors. The RF model classifier found that 7.20% of the study area has very high wildfire susceptibility, and the majority of the wildfire samples (68.84%) correspond to the very high susceptible areas in the RF model. The outcomes of this study are likely to provide decision-makers with a better understanding of wildfires in the Eastern Mediterranean Region of Turkey.     

Monitoring <i>Thosea sinensis</i> Walker in Tea Plantations Based on UAV Multi-Spectral Image

Thosea sinensis Walker (TSW) rapidly spreads and severely damages the tea plants. Therefore, finding a reliable operational method for identifying the TSW-damaged areas via remote sensing has been a focus of a research community. Such methods also enable us to calculate the precise application of pesticides and prevent the subsequent spread of the pests. In this work, based on the unmanned aerial vehicle (UAV) platform, five band images of multispectral red-edge camera were obtained and used for monitoring the TSW in tea plantations. By combining the minimum redundancy maximum relevance (mRMR) with the selected spectral features, a comprehensive spectral selection strategy was proposed. Then, based on the selected spectral features, three classic machine learning algorithms, including random forest (RF), support vector machine (SVM), and k-nearest neighbors (KNN) were used to construct the pest monitoring model and were evaluated and compared. The results showed that the strategy proposed in this work obtained ideal monitoring accuracy by only using the combination of a few optimized features (2 or 4). In order to differentiate the healthy and TSW-damaged areas (2-class model), the monitoring accuracies of all the three models were computed, which were above 96%. The RF model used the least number of features, including only SAVI and Band_red. In order to further discriminate the pest incidence levels (3-class model), the monitoring accuracies of all the three models were computed, which were above 80%, among which the RF algorithm based on SAVI, Band_red, VARI_green, and Band_{red_edge} features achieve the highest accuracy (OAA of 87%, and Kappa of 0.79). Considering the computational cost and model accuracy, this work recommends the RF model based on a few optimal feature combinations to monitor and distinguish the severity of TSW in tea plantations. According to the UAV remote sensing mapping results, the TSW infestation exhibited an aggregated distribution pattern. The spatial information of occurrence and severity can offer effective guidance for precise control of the pest. In addition, the relevant methods provide a reference for monitoring other leaf-eating pests, effectively improving the management level of plant protection in tea plantations, and guaranting the yield and quality of tea plantations.     

基于景观及微地形特征的丘陵区土壤属性预测

为探讨小流域尺度丘陵区的高分辨率数字土壤制图方法,通过对景观相分类的探索,配合应用不同尺度的Geomorphons(GM)微地形特征数据构成分类变量组参与高分辨率土壤pH、黏粒含量和阳离子交换量的预测制图,并与传统数字高程模型衍生变量和遥感变量进行组合与比较分析。此外,采用支持向量机、偏最小二乘回归和随机森林3种机器学习模型择优与残差回归克里金复合参与预测模型的构建与评价。结果表明: 景观及多尺度微地形分类变量组的应用分别提高小流域尺度丘陵地貌区pH、黏粒含量和阳离子交换量预测精度的18.8%、8.2%和8.7%。包含植被信息的景观相分类图相比土地利用数据有更高的模型贡献度;5 m分辨率的GM微地形分类图相比低分辨率的分类图更适宜高精度的预测制图。黏粒含量使用随机森林复合模型有最高的预测精度,而pH和阳离子交换量则不适宜在随机森林模型的基础上加入残差回归克里金模型。景观-多尺度微地形分类变量、数字高程模型衍生变量和遥感变量三者结合的模型预测表现最佳,表明多元变量在起伏地形区域相比单一数据源能够包含更多的土壤有效信息。由GM数据和地表景观数据组成的景观分类变量组作为主要变量能够解释小流域丘陵区部分土壤属性约40%的空间变异。在同类型土壤预测制图研究中,多分辨率GM及景观分类数据有潜力作为环境变量参与预测模型的构建。     

An algorithm for early rice area mapping from satellite remote sensing data in southwestern Guangdong in China based on feature optimization and random Forest

Estimating of rice areas using images obtained from satellite remote sensing is important for guiding operators. The object of this study was the Sentinel-2A/B image data of the rice planting demonstration regions in southwestern Guangdong, China. We designed an algorithm for early rice area mapping based on feature optimization and random forest (RF). For modeling, we selected 35 common remote sensing features and applied out-of-bag (OOB) to construct 7 feature combinations. The results showed that the overall accuracy (OA) and Kappa coefficient of the RF with the best combination were 91.23% and 87.55%, respectively. Compared with support vector machine (SVM) and back propagation neural network (BPNN), the model result of RF was also the best among the three. Additionally, the maximum error of the rice area was less than 16% when the model was transferred to other regions in Guangdong. The feature optimization and RF-based algorithm proposed in this study can effectively map the early rice region. It can be applied to estimate rice area based on satellite remote sensing image data and reveal the ecological status of rice cultivation in southwestern Guangdong.     

Background sampling for multi-scale ensemble habitat selection modeling: Does the number of points matter?

Ensemble habitat selection modeling is becoming a popular approach among ecologists to answer different questions. Since we are still in the early stages of development and application of ensemble modeling, there remain many questions regarding performance and parameterization. One important gap, which this paper addresses, is how the number of background points used to train models influences the performance of the ensemble model. We used an empirical presence-only dataset and three different selections of background points to train scale-optimized habitat selection models using six modeling algorithms (GLM, GAM, MARS, ANN, Random Forest, and MaxEnt). We tested four ensemble models using different combinations of the component models: (a) equal numbers of background points and presences, (b) background points equaled ten times the number of presences, (c) 10,000 background points, and (d) optimized background points for each component model. Among regression-based approaches, MARS performed best when built with 10,000 background points. Among machine learning models, RF performed the best when built with equal presences and background points. Among the four ensemble models, AUC indicated that the best performing model was the ensemble with each component model including the optimized number of background points, while TSS increased as the number of background points models increased. We found that an ensemble of models, each trained with an optimal number of background points, outperformed ensembles of models trained with the same number of background points, although differences in performance were slight. When using a single modeling method, RF with equal number of presences and background points can perform better than an ensemble model, but the performance fluctuates when the number of background points is not properly selected. On the other hand, ensemble modeling provides consistently high accuracy regardless of background point sampling approach. Further, optimizing the number of background points for each component model within an ensemble model can provide the best model improvement. We suggest evaluating more models across multiple species to investigate how background point selection might affect ensemble models in different scenarios.     

Knowledge discovery of Middle East dust sources using Apriori spatial data mining algorithm

Identifying the areas susceptible to dust storm formation is one effective way of dealing with this destructive environmental phenomenon. This study is the first attempt to employ the Apriori spatial data mining algorithm to dust source susceptibility mapping (DSSM). The research process was based on extracting association rules between spatial-temporal patterns of dust drivers (including soil, vegetation, and climate parameters) in the Middle East's hotspots dust sources (HDSs). For this purpose, HDSs were identified using visual interpretation of sub-daily MODIS-Terra/Aqua RGB images from 2000 to 2021. The Middle East's HDSs mainly correspond to desert areas with poor vegetation cover and ephemeral/dried-up water bodies. A total of three million rules were extracted by running the Apriori algorithm. Accordingly, bare and non-vegetated lands, high soil thickness, low soil moisture, very high wind speed, and high temperature were identified as the most common features of HDSs. Using three measures including support, confidence, and lift, 54 frequent, reliable, and logical rules were selected, and the related maps were generated. Then, the susceptible dust sources (SDSs) map of the Middle East was produced in five classes of extreme (13% of the areas), high (14%), moderate (16%), low (17%), and no (40%) susceptibility through the weighted linear combination of the rule maps. The accuracy of the identified SDSs was estimated at 83.7% using the verification points. A sensitivity analysis was performed using the leave-one-out method to determine the isolated effect of the selected rules on the produced SDSs map. The model uncertainty varied between 15.7% and 16.8% for different rules. The variation range of uncertainty was 1.1%, demonstrating that a single rule does not significantly affect the model's performance; however, some rules have a more influential role. Our results revealed that Apriori's ability to provide generalizable association rules is a robust algorithm for DSSM.     

普洱季风常绿阔叶林乔木物种多样性高分辨率遥感估测

季风常绿阔叶林是我国南亚热带典型的地带性植被,也是云南省普洱地区重要森林类型。季风常绿阔叶林乔木物种多样性遥感估测对研究区域尺度生物多样性格局及其规律具有重要作用。根据光谱异质性假说和环境异质性假说,首先使用1m空间分辨率的机载高光谱数据和激光雷达数据提取了光谱多样性特征和垂直结构特征。然后利用基于随机森林算法的递归特征消除方法选择对研究区森林乔木物种多样性指数具有较好解释能力的遥感特征,并对Shannon-Winner物种多样性指数进行建模、制图。研究结果表明：(1)基于机载LiDAR数据提取的垂直结构特征和机载高光谱数据提取的光谱多样性特征均对研究区森林乔木物种多样性具有较好的解释能力,随机森林模型估测结果分别为R²=0.48,RMSE=0.46和R²=0.5,RMSE=0.45;两种数据源融合可以进一步提高遥感数据的森林乔木物种多样性估测精度,随机森林估测模型R²和RMSE分别为0.69和0.37。(2)机载激光雷达数据对研究区针阔混交林乔木物种多样性的估测能力优于机载高光谱数据。(3)机器学习方法有助于从高维遥感...     

A comparison of absolute performance of different correlative and mechanistic species distribution models in an independent area

To investigate the comparative abilities of six different bioclimatic models in an independent area, utilizing the distribution of eight different species available at a global scale and in Australia. Global scale and Australia. We tested a variety of bioclimatic models for eight different plant species employing five discriminatory correlative species distribution models (SDMs) including Generalized Linear Model (GLM), MaxEnt, Random Forest (RF), Boosted Regression Tree (BRT), Bioclim, together with CLIMEX (CL) as a mechanistic niche model. These models were fitted using a training dataset of available global data, but with the exclusion of Australian locations. The capabilities of these techniques in projecting suitable climate, based on independent records for these species in Australia, were compared. Thus, Australia is not used to calibrate the models and therefore it is as an independent area regarding geographic locations. To assess and compare performance, we utilized the area under the receiver operating characteristic (ROC) curves (AUC), true skill statistic (TSS), and fractional predicted areas for all SDMs. In addition, we assessed satisfactory agreements between the outputs of the six different bioclimatic models, for all eight species in Australia. The modeling method impacted on potential distribution predictions under current climate. However, the utilization of sensitivity and the fractional predicted areas showed that GLM, MaxEnt, Bioclim, and CL had the highest sensitivity for Australian climate conditions. Bioclim calculated the highest fractional predicted area of an independent area, while RF and BRT were poor. For many applications, it is difficult to decide which bioclimatic model to use. This research shows that variable results are obtained using different SDMs in an independent area. This research also shows that the SDMs produce different results for different species; for example, Bioclim may not be good for one species but works better for other species. Also, when projecting a “large” number of species into novel environments or in an independent area, the selection of the “best” model/technique is often less reliable than an ensemble modeling approach. In addition, it is vital to understand the accuracy of SDMs' predictions. Further, while TSS, together with fractional predicted areas, are appropriate tools for the measurement of accuracy between model results, particularly when undertaking projections on an independent area, AUC has been proved not to be. Our study highlights that each one of these models (CL, Bioclim, GLM, MaxEnt, BRT, and RF) provides slightly different results on projections and that it may be safer to use an ensemble of models.     

气候变化对物种分布影响模拟中的不确定性组分分割与制图——以油松为例

物种分布模型是预测评估气候变化对物种分布影响的主要工具。为了降低物种分布模型在预测过程中的不确定性,近期有学者提出了采用组合预测的新方法,即采用多套建模数据、模型技术,模型参数,以及环境情景数据对物种分布进行预测,构成物种分布预测集合。但是,组合预测中各组分对变异的贡献还知之甚少,因此有必要把变异组分来源进行分割,以更有效地利用组合预测方法来降低模型预测中的不确定性。以油松为例,采用8个生态位模型,9套模型训练数据,3个GCM模型和一个SRES(A2)排放情景,模型分析了油松当前(1961-1990年)和未来气候条件下3个时间段(2010-2039年,2040-2069年,2070-2099年)的潜在分布。共计得到当前分布预测数据72套,未来每个时间段分布数据216套。采用开发的ClimateChina软件进行当前和未来气候数据的降尺度处理。采用Kappa、真实技巧统计方法(TSS)和接收机工作特征曲线下的面积(AUC)对模型预测能力进行评估。结果表明,随机森林(RF)、广义线性模型(GLM),广义加法模型(GAM)、多元自适应样条函数(MARS)以及助推法(GBM)预测效果较好,几乎不受建模数据之间差异的影响。混合判别分析模型(MDA)对建模数据之间的差异非常敏感,甚至出现建模失败现象。采用三因素方差分析方法对组合预测中的不确定性来源进行变异分割,结果表明,模型之间的差异对模拟预测结果不确定性的贡献最大且所占比例极高,而建模数据之间的差异贡献最小,GCM贡献居中。研究将有助于加深对物种分布模拟预测中不确定性的认识。     

联合GF-6和Sentinel-2红边波段的森林地上生物量反演

光谱反射率能反映地物差异,是森林地上生物量（Aboveground Biomass,AGB）遥感反演的理论基础。红边波段处于近红外与红光波段交界处快速变化的区域,能对植被冠层结构和叶绿素含量的微小变化做出快速反应,对植被生长状况较敏感。研究以GF-6和Sentinel-2多光谱影像作为数据源,结合野外调查AGB数据,构建落叶松和樟子松AGB线性和非线性估测模型,通过比较模型精度选择最优模型进行森林AGB反演和空间分布制图。结果表明：GF-6和Sentinel-2影像红边波段反射率与落叶松、樟子松AGB均呈显著相关（P<0.05）,红边波段对AGB估测较敏感。多变量估测模型整体估测效果优于单变量模型,所有模型中多元线性回归模型取得了最优的决定系数（落叶松R²=0.66,樟子松R²=0.65）和最低的均方根误差（落叶松RMSE=31.45 t/hm²,樟子松RMSE=54.77 t/hm²）。相比单个数据源,联合GF-6和Sentinel-2影像构建的多元线性回归模型估测效果得到了显著提升,模型RMSE对于落叶松和樟子松AGB估测分别最大降低了22.9%和11.2%。增加红边波段进行AGB估测能显著提高模型估测精度,三组数据源分别加入红边波段信息后进行建模,模型RMSE得到了显著降低。GF-6拥有800 km观测幅宽和高效的重访周期,可以快速地提供大尺度时间序列数据,在森林地上生物量反演和动态监测方面有着很大潜力。     

Not seeing the forest for the trees: Generalised linear model out-performs random forest in species distribution modelling for Southeast Asian felids

Species Distribution Models (SDMs) are a powerful tool to derive habitat suitability predictions relating species occurrence data with habitat features. Two of the most frequently applied algorithms to model species-habitat relationships are Generalised Linear Models (GLM) and Random Forest (RF). The former is a parametric regression model providing functional models with direct interpretability. The latter is a machine learning non-parametric algorithm, more tolerant than other approaches in its assumptions, which has often been shown to outperform parametric algorithms. Other approaches have been developed to produce robust SDMs, like training data bootstrapping and spatial scale optimisation. Using felid presence-absence data from three study regions in Southeast Asia (mainland, Borneo and Sumatra), we tested the performances of SDMs by implementing four modelling frameworks: GLM and RF with bootstrapped and non-bootstrapped training data. With Mantel and ANOVA tests we explored how the four combinations of algorithms and bootstrapping influenced SDMs and their predictive performances. Additionally, we tested how scale-optimisation responded to species' size, taxonomic associations (species and genus), study area and algorithm. We found that choice of algorithm had strong effect in determining the differences between SDMs' spatial predictions, while bootstrapping had no effect. Additionally, algorithm followed by study area and species, were the main factors driving differences in the spatial scales identified. SDMs trained with GLM showed higher predictive performance, however, ANOVA tests revealed that algorithm had significant effect only in explaining the variance observed in sensitivity and specificity and, when interacting with bootstrapping, in Percent Correctly Classified (PCC). Bootstrapping significantly explained the variance in specificity, PCC and True Skills Statistics (TSS). Our results suggest that there are systematic differences in the scales identified and in the predictions produced by GLM vs. RF, but that neither approach was consistently better than the other. The divergent predictions and inconsistent predictive abilities suggest that analysts should not assume machine learning is inherently superior and should test multiple methods. Our results have strong implications for SDM development, revealing the inconsistencies introduced by the choice of algorithm on scale optimisation, with GLM selecting broader scales than RF.     

Sea-level rise effects on macrozoobenthos distribution within an estuarine gradient using Species Distribution Modeling

The sea-level rise induced by climate change has caused impacts (e.g., floods and saline intrusion) in estuaries. In this work, we used monitoring data (salinity, sediment and taxa occurrence), simulated saline intrusion and Species Distribution Model to predict the spatial distribution of families in the estuary at two levels of SLR (0.5 m and 1 m) for two scenarios (moderate and extreme). For the simulation, we used the ensemble method applied to five models (MARS, GLM, GAM, RF and BRT). High AUC and TSS values indicated “good” to “excellent” accuracy. RF and GLM obtained the best and worst values, respectively. The model predicted local extinctions and new colonization in the upper estuarine zones. With the effects of climate change intensifying, it is extremely important that managers consider the use of predictive tools to anticipate the impacts of climate change on a local scale on species migration.     

Spectral mapping tools from the earth sciences applied to spectral microscopy data.

BACKGROUND: Spectral imaging, originating from the field of earth remote sensing, is a powerful tool that is being increasingly used in a wide variety of applications for material identification. Several workers have used techniques like linear spectral unmixing (LSU) to discriminate materials in images derived from spectral microscopy. However, many spectral analysis algorithms rely on assumptions that are often violated in microscopy applications. This study explores algorithms originally developed as improvements on early earth imaging techniques that can be easily translated for use with spectral microscopy. METHODS: To best demonstrate the application of earth remote sensing spectral analysis tools to spectral microscopy data, earth imaging software was used to analyze data acquired with a Leica confocal microscope with mechanical spectral scanning. For this study, spectral training signatures (often referred to as endmembers) were selected with the ENVI (ITT Visual Information Solutions, Boulder, CO) "spectral hourglass" processing flow, a series of tools that use the spectrally over-determined nature of hyperspectral data to find the most spectrally pure (or spectrally unique) pixels within the data set. This set of endmember signatures was then used in the full range of mapping algorithms available in ENVI to determine locations, and in some cases subpixel abundances of endmembers. RESULTS: Mapping and abundance images showed a broad agreement between the spectral analysis algorithms, supported through visual assessment of output classification images and through statistical analysis of the distribution of pixels within each endmember class. CONCLUSIONS: The powerful spectral analysis algorithms available in COTS software, the result of decades of research in earth imaging, are easily translated to new sources of spectral data. Although the scale between earth imagery and spectral microscopy is radically different, the problem is the same: mapping material locations and abundances based on unique spectral signatures.     

Identifying sand and dust storm sources using spatial-temporal analysis of remote sensing data in Central Iran

Sand and dust storms (SDS) are common meteorological phenomena in arid and semi-arid regions caused by natural or anthropogenic factors. Central Iran, covers a large area on the Iranian plateau, and SDS has been known as a prevalent phenomenon in certain parts of the region since ancient times. The frequency and severity of SDS have increased over the last two decades due to population growth and mismanagement of natural resources. Identifying SDS sources is the first step to combating this phenomenon and reducing its destructive impacts. Accordingly, this study employed a remote sensing approach based on the modeling of environmental parameters to identify high potential SDS sources in Central Iran. The proposed model was implemented through a multi-step masking procedure using 20-year time-series datasets of MODIS and TerraClimate products. According to the results, 5.3% of Central Iran is identified as high potential SDS sources. Among these, sandy sources have the largest share in terms of area (60.9%) and frequency of SDS occurrence (>50%). The highest seasonal frequency of SDS (76%) was in spring and summer. The highest yearly frequency of SDS was observed in 2008, which was 120% higher than the 20-year average (2000−2020). In sandy and salt plain sources, SDS formation is predominantly associated with natural factors. However, in lakes and alluvial sources, anthropogenic activities have been directly linked with variations in SDS frequency and extent. The occurrence of severe droughts has intensified the frequency of SDS emerging from all types of high-potential sources in Central Iran.     

Remote sensing imagery in vegetation mapping: a review

Aims Mapping vegetation through remotely sensed images involves various considerations, processes and techniques. Increasing availability of remotely sensed images due to the rapid advancement of remote sensing technology expands the horizon of our choices of imagery sources. Various sources of imagery are known for their differences in spectral, spatial, radioactive and temporal characteristics and thus are suitable for different purposes of vegetation mapping. Generally, it needs to develop a vegetation classification at first for classifying and mapping vegetation cover from remote sensed images either at a community level or species level. Then, correlations of the vegetation types (communities or species) within this classification system with discernible spectral characteristics of remote sensed imagery have to be identified. These spectral classes of the imagery are finally translated into the vegetation types in the image interpretation process, which is also called image processing. This paper presents an overview of how to use remote sensing imagery to classify and map vegetation cover.Methods Specifically, this paper focuses on the comparisons of popular remote sensing sensors, commonly adopted image processing methods and prevailing classification accuracy assessments.Important findings The basic concepts, available imagery sources and classification techniques of remote sensing imagery related to vegetation mapping were introduced, analyzed and compared. The advantages and limitations of using remote sensing imagery for vegetation cover mapping were provided to iterate the importance of thorough understanding of the related concepts and careful design of the technical procedures, which can be utilized to study vegetation cover from remote sensed images.     

基于多源遥感数据的大豆叶面积指数估测精度对比

近年来遥感技术的革新促使遥感源越来越丰富.为分析多源遥感数据的叶面积指数(LAI)估测精度,本文以大豆为研究对象,利用比值植被指数(RVI)、归一化植被指数(NDVI)、土壤调整植被指数(SAVI)、差值植被指数(DVI)、三角植被指数(TVI)5种植被指数,结合地面实测LAI构建经验回归模型,比较3类遥感数据(地面高光谱数据、无人机多光谱影像以及高分一号WFV影像)对大豆LAI的估测能力,并从传感器几何位置和光谱响应特性以及像元空间分辨率三方面分析讨论了3类遥感数据的LAI反演差异.结果表明： 地面高光谱数据模型和无人机多光谱数据模型都可以准确预测大豆LAI(在α=0.01显著水平下,R²均>0.69,RMSE均<0.40);地面高光谱RVI对数模型的LAI预测能力优于无人机多光谱NDVI线性模型,但两者差异不大(E_A相差0.3%,R²相差0.04,RMSE相差0.006);高分一号WFV数据模型对研究区内大豆LAI的预测效果不理想(R²<0.30,RMSE>0.70).针对星、机、地三类遥感信息源,地面高光谱数据在反演LAI方面较传统多光谱数据有优势但不突出;16 m空间分辨率的高分一号WFV影像无法满足田块尺度作物长势监测的需求;在保证获得高精度大豆LAI预测值和高工作效率的前提条件下,基于无人机遥感的农情信息获取技术不失为一种最佳试验方案.在当今可用遥感信息源越来越多的情况下,农业无人机遥感信息可成为指导田块精细尺度作物管理的重要依据,为精准农业研究提供更科学准确的信息.     

基于无人机多源遥感数据的亚热带森林树种分类

树种多样性是生态学研究的重要内容,树木的种类和空间分布信息可有效服务于可持续森林管理。但在复杂林分条件下,获取高精度分类结果的难度大。而无人机遥感可获取局域超精细数据,为树种分类精度的提高提供了可能。基于可见光、高光谱、激光雷达等多源无人机遥感数据,探究其在亚热带林分条件下的树种分类潜力。研究发现:(1)随机森林分类器总体精度和各树种的F1分数最高,适合亚热带多树种的分类制图,其区分13种类别(8乔木,4草本)的总体精度为95.63%,Kappa系数为0.948;(2)多源数据的使用可以显著提高分类精度,全特征模型精度最高,且高光谱和激光雷达数据显著影响全特征模型分类精度,可见光纹理数据作用较小;(3)分类特征重要性从大到小排序为结构信息,植被指数,纹理信息,最小噪声变换分量。     

Predicting alpha diversity of African rain forests: models based on climate and satellite‐derived data do not perform better than a purely spatial model

Aim Our aim was to evaluate the extent to which we can predict and map tree alpha diversity across broad spatial scales either by using climate and remote sensing data or by exploiting spatial autocorrelation patterns. Location Tropical rain forest, West Africa and Atlantic Central Africa. Methods Alpha diversity estimates were compiled for trees with diameter at breast height ≥ 10 cm in 573 inventory plots. Linear regression (ordinary least squares, OLS) and random forest (RF) statistical techniques were used to project alpha diversity estimates at unsampled locations using climate data and remote sensing data [Moderate Resolution Imaging Spectroradiometer (MODIS), normalized difference vegetation index (NDVI), Quick Scatterometer (QSCAT), tree cover, elevation]. The prediction reliabilities of OLS and RF models were evaluated using a novel approach and compared to that of a kriging model based on geographic location alone. Results The predictive power of the kriging model was comparable to that of OLS and RF models based on climatic and remote sensing data. The three models provided congruent predictions of alpha diversity in well‐sampled areas but not in poorly inventoried locations. The reliability of the predictions of all three models declined markedly with distance from points with inventory data, becoming very low at distances > 50 km. According to inventory data, Atlantic Central African forests display a higher mean alpha diversity than do West African forests. Main conclusions The lower tree alpha diversity in West Africa than in Atlantic Central Africa may reflect a richer regional species pool in the latter. Our results emphasize and illustrate the need to test model predictions in a spatially explicit manner. Good OLS or RF model predictions from inventory data at short distance largely result from the strong spatial autocorrelation displayed by both the alpha diversity and the predictive variables rather than necessarily from causal relationships. Our results suggest that alpha diversity is driven by history rather than by the contemporary environment. Given the low predictive power of models, we call for a major effort to broaden the geographical extent and intensity of forest assessments to expand our knowledge of African rain forest diversity.     

Predicting the cover and richness of intertidal macroalgae in remote areas: a case study in the Antarctic Peninsula

Antarctica is an iconic region for scientific explorations as it is remote and a critical component of the global climate system. Recent climate change causes a dramatic retreat of ice in Antarctica with associated impacts to its coastal ecosystem. These anthropogenic impacts have a potential to increase habitat availability for Antarctic intertidal assemblages. Assessing the extent and ecological consequences of these changes requires us to develop accurate biotic baselines and quantitative predictive tools. In this study, we demonstrated that satellite‐based remote sensing, when used jointly with in situ ground‐truthing and machine learning algorithms, provides a powerful tool to predict the cover and richness of intertidal macroalgae. The salient finding was that the Sentinel‐based remote sensing described a significant proportion of variability in the cover and richness of Antarctic macroalgae. The highest performing models were for macroalgal richness and the cover of green algae as opposed to the model of brown and red algal cover. When expanding the geographical range of the ground‐truthing, even involving only a few sample points, it becomes possible to potentially map other Antarctic intertidal macroalgal habitats and monitor their dynamics. This is a significant milestone as logistical constraints are an integral part of the Antarctic expeditions. The method has also a potential in other remote coastal areas where extensive in situ mapping is not feasible.