Nearest neighbor imputation of species-level, plot-scale forest structure attributes from LiDAR data

Meaningful relationships between forest structure attributes measured in representative field plots on the ground and remotely sensed data measured comprehensively across the same forested landscape facilitate the production of maps of forest attributes such as basal area (BA) and tree density (TD). Because imputation methods can efficiently predict multiple response variables simultaneously, they may be usefully applied to map several structural attributes at the species-level. We compared several approaches for imputing the response variables BA and TD, aggregated at the plot-scale and species-level, from topographic and canopy structure predictor variables derived from discrete-return airborne LiDAR data. The predictor and response variables were associated using imputation techniques based on normalized and unnormalized Euclidean distance, Mahalanobis distance, Independent Component Analysis (ICA), Canonical Correlation Analysis (aka Most Similar Neighbor, or MSN), Canonical Correspondence Analysis (aka Gradient Nearest Neighbor, or GNN), and Random Forest (RF). To compare and evaluate these approaches, we computed a scaled Root Mean Square Distance (RMSD) between observed and imputed plot-level BA and TD for 11 conifer species sampled in north-central Idaho. We found that RF produced the best results overall, especially after reducing the number of response variables to the most important species in each plot with regard to BA and TD. We concluded that RF was the most robust and flexible among the imputation methods we tested. We also concluded that canopy structure and topographic metrics derived from LiDAR surveys can be very useful for species-level imputation.     

Mapping U.S. forest biomass using nationwide forest inventory data and moderate resolution information

A spatially explicit dataset of aboveground live forest biomass was made from ground measured inventory plots for the conterminous U.S., Alaska and Puerto Rico. The plot data are from the USDA Forest Service Forest Inventory and Analysis (FIA) program. To scale these plot data to maps, we developed models relating field-measured response variables to plot attributes serving as the predictor variables. The plot attributes came from intersecting plot coordinates with geospatial datasets. Consequently, these models serve as mapping models. The geospatial predictor variables included Moderate Resolution Imaging Spectrometer (MODIS)-derived image composites and percent tree cover; land cover proportions and other data from the National Land Cover Dataset (NLCD); topographic variables; monthly and annual climate parameters; and other ancillary variables. We segmented the mapping models for the U.S. into 65 ecologically similar mapping zones, plus Alaska and Puerto Rico. First, we developed a forest mask by modeling the forest vs. nonforest assignment of field plots as functions of the predictor layers using classification trees in See5©. Secondly, forest biomass models were built within the predicted forest areas using tree-based algorithms in Cubist©. To validate the models, we compared field-measured with model-predicted forest/nonforest classification and biomass from an independent test set, randomly selected from available plot data for each mapping zone. The estimated proportion of correctly classified pixels for the forest mask ranged from 0.79 in Puerto Rico to 0.94 in Alaska. For biomass, model correlation coefficients ranged from a high of 0.73 in the Pacific Northwest, to a low of 0.31 in the Southern region. There was a tendency in all regions for these models to over-predict areas of small biomass and under-predict areas of large biomass, not capturing the full range in variability. Map-based estimates of forest area and forest biomass compared well with traditional plot-based estimates for individual states and for four scales of spatial aggregation. Variable importance analyses revealed that MODIS-derived information could contribute more predictive power than other classes of information when used in isolation. However, the true contribution of each variable is confounded by high correlations. Consequently, excluding any one class of variables resulted in only small effects on overall map accuracy. An estimate of total C pools in live forest biomass of U.S. forests, derived from the nationwide biomass map, also compared well with previously published estimates.     

Estimating biomass carbon stocks for a Mediterranean forest in central Spain using LiDAR height and intensity data

Biomass fractions (total aboveground, branches and foliage) were estimated from a small footprint discrete-return LiDAR system in an unmanaged Mediterranean forest in central Spain. Several biomass estimation models based on LiDAR height, intensity or height combined with intensity data were explored. Raw intensity data were normalized to a standard range in order to remove the range dependence of the intensity signal. In general terms, intensity-based models provided more accurate predictions of the biomass fractions. Height models selected were mainly based on a percentile of the height distribution. Intensity models selected included variables that consider the percentage of the intensity accumulated at different height percentiles, which implicitly take into account the height distribution. The general models derived considering all species together were based on height combined with intensity data. These models yielded R² values greater than 0.58 for the different biomass fractions considered and RMSE values of 28.89, 18.28 and 1.51 Mg ha⁻¹ for aboveground, branch and foliage biomass, respectively. Results greatly improved for species-specific models using the main species present in each plot, with R² values greater than 0.85, 0.70 and 0.90 for black pine, Spanish juniper and Holm oak, respectively, and with lower RMSE for the biomass fractions. Reductions in LiDAR point density had only a small effect on the results obtained, except for those models based on a variation of the Canopy Reflection Sum, which was weighted by the mean point density. Based on the species-specific equations derived, Holm oak dominated plots showed the highest average carbon contained by aboveground biomass and branch biomass 44.66 and 31.42 Mg ha^− 1 respectively, while for foliage biomass carbon, Spanish juniper showed the highest average value (3.04 Mg ha^− 1).     

Estimation of above- and below-ground biomass across regions of the boreal forest zone using airborne laser

Regression models relating variables derived from airborne laser scanning (ALS) to above-ground and below-ground biomass were estimated for 1395 sample plots in young and mature coniferous forest located in ten different areas within the boreal forest zone of Norway. The sample plots were measured as part of large-scale operational forest inventories. Four different ALS instruments were used and point density varied from 0.7 to 1.2 m^− 2. One variable related to canopy height and one related to canopy density were used as independent variables in the regressions. The statistical effects of area and age class were assessed by including dummy variables in the models. Tree species composition was treated as continuous variables. The proportion of explained variability was 88% for above- and 85% for below-ground biomass models. For given combinations of ALS-derived variables, the differences between the areas were up to 32% for above-ground biomass and 38% for below-ground biomass. The proportion of spruce had a significant impact on both the estimated models. The proportion of broadleaves had a significant effect on above-ground biomass only, while the effect of age class was significant only in the below-ground biomass model. Because of local effects on the biomass-ALS data relationships, it is indicated by this study that sample plots distributed over the entire area would be needed when using ALS for regional or national biomass monitoring.     

Predictive relations of tropical forest biomass from Landsat TM data and their transferability between regions

The full realization of the potential of remote sensing as a source of environmental information requires an ability to generalize in space and time. Here, the ability to generalize in space was investigated through an analysis of the transferability of predictive relations for the estimation of tropical forest biomass from Landsat TM data between sites in Brazil, Malaysia and Thailand. The data sets for each test site were acquired and processed in a similar fashion to facilitate the analyses. Three types of predictive relation, based on vegetation indices, multiple regression and feedforward neural networks, were developed for biomass estimation at each site. For each site, the strongest relationships between the biomass predicted and that measured from field survey was obtained with a neural network developed specifically for the site (r>0.71, significant at the 99% level of confidence). However, with each type of approach problems in transferring a relation to another site were observed. In particular, it was apparent that the accuracy of prediction, as indicated by the correlation coefficient between predicted and measured biomass, declined when a relation was transferred to a site other than that upon which it was developed. Part of this problem lies with the observed variation in the relative contribution of the different spectral wavebands to predictive relations for biomass estimation between sites. It was, for example, apparent that the spectral composition of the vegetation indices most strongly related to biomass differed greatly between the sites. Consequently, the relationship between predicted and measured biomass derived from vegetation indices differed markedly in both strength and direction between sites. Although the incorporation of test site location information into an analysis resulted in an increase in the strength of the relationship between predicted and actual biomass, considerable further research is required on the problems associated with transferring predictive relations.     

Estimation of tropical rain forest aboveground biomass with small-footprint lidar and hyperspectral sensors

Tropical forests are an important component of the global carbon balance, yet there is considerable uncertainty in estimates of their carbon stocks and fluxes, which are typically estimated through analysis of aboveground biomass in field plots. Remote sensing technology is critical for assessing fine-scale spatial variability of tropical forest biomass over broad spatial extents. The goal of our study was to evaluate relatively new technology, small-footprint, discrete-return lidar and hyperspectral sensors, for the estimation of aboveground biomass in a Costa Rican tropical rain forest landscape. We derived a suite of predictive metrics for field plots: lidar metrics were calculated from plot vertical height profiles and hyperspectral metrics included fraction of spectral mixing endmembers and narrowband indices that respond to photosynthetic vegetation, structure, senescence, health and water and lignin content. We used single- and two-variable linear regression analyses to relate lidar and hyperspectral metrics to aboveground biomass of plantation, managed parkland and old-growth forest plots. The best model using all 83 biomass plots included two lidar metrics, plot-level mean height and maximum height, with an r² of 0.90 and root-mean-square error (RMSE) of 38.3 Mg/ha. When the analysis was constrained to plantation plots, which had the most accurate field data, the r² of the model increased to 0.96, with RMSE of 10.8 Mg/ha (n = 32). Hyperspectral metrics provided lower accuracy in estimating biomass than lidar metrics, and models with a single lidar and hyperspectral metric were no better than the best model using two lidar metrics. These results should be viewed as an initial assessment of using these combined sensors to estimate tropical forest biomass; hyperspectral data were reduced to nine indices and three spectral mixture fractions, lidar data were limited to first-return canopy height, sensors were flown only once at different seasons, and we explored only linear regression for modeling. However, this study does support conclusions from studies at this and other climate zones that lidar is a premier instrument for mapping biomass (i.e., carbon stocks) across broad spatial scales.     

Depression detection from sMRI and rs-fMRI images using machine learning

Major Depression Disorder (MDD) is a common mental disorder that negatively affects many people’s lives worldwide. Developing an automated method to find useful diagnostic biomarkers from brain imaging data would help clinicians to detect MDD in its early stages. Depression is known to be a brain connectivity disorder problem. In this paper, we present a brain connectivity-based machine learning (ML) workflow that utilizes similarity/dissimilarity of spatial cubes in brain MRI images as features for depression detection. The proposed workflow provides a unified framework applicable to both structural MRI images and resting-state functional MRI images. Several cube similarity measures have been explored, including Pearson or Spearman correlations, Minimum Distance Covariance, or inverse of Minimum Distance Covariance. Discriminative features from the cube similarity matrix are chosen with the Wilcoxon rank-sum test. The extracted features are fed into machine learning classifiers to train MDD prediction models. To address the challenge of data imbalance in MDD detection, oversampling is performed to balance the training data. The proposed workflow is evaluated through experiments on three independent public datasets, all imbalanced, of structural MRI and resting-state fMRI images with depression labels. Experimental results show good performance on all three datasets in terms of prediction accuracy, specificity, sensitivity, and area under the Receiver Operating Characteristic (ROC) curve. The use of features from both structured MRI and resting state functional MRI is also investigated.

     

Estimation of above-ground forest biomass using metrics based on Gaussian decomposition of waveform lidar data

Large-footprint waveform light detection and ranging (lidar) data have been widely used in above-ground forest biomass estimation. Waveform metrics derived from basic statistics (e.g. percentile of energy) of the lidar waveform, such as canopy height and height of median energy, have been applied to biomass estimation in numerous studies. In this study, a set of metrics based on Gaussian decomposition (GD) results were developed and evaluated for forest above-ground biomass estimation using NASA’s laser vegetation imaging sensor (LVIS) data. The GD metrics were designed to explicitly incorporate lidar intensity and vertical structures of canopy layers for biomass estimation. The proposed GD metrics used information related to the above-ground height of each Gaussian centroid and the Gaussian area index (GAI), where GAI is the area covered by a Gaussian function. Two types of novel GD metrics were developed: (1) percentile-height GAI metrics expressing the GAI summation of a subset of Gaussian centroids located within a certain percentile height range; and (2) height-weighted GAI metrics, a summation of GAIs of a waveform weighted by the corresponding heights of their Gaussian centroids. A biomass regression model was built by eight newly developed GD metrics using GAI information and five pre-existing GD-derived metrics that have not previously been used for biomass estimation. The performance of the regression model was then compared to another regression model using 12 previously published metrics (non-GD metrics). The Random Forests (RF) regression algorithm was employed for predicting biomass. The RF out-of-bag results indicated that above-ground biomass estimations using GD metrics achieved significantly better results than those derived from non-GD metrics for deciduous plots (19% lower root mean square error (RMSE), 25% higher coefficient of determination (R²), and marginally better results in coniferous plots (4% lower RSME, 6% higher R²). The combination of GD and non-GD metrics achieved comparable biomass estimation results to the model using exclusively GD metrics. GD metrics also showed strong correlation with forest attributes such as mean diameter at breast height (DBH) and stem density. This study contributes to the usage of GD results for accurate estimation of forest above-ground biomass in large-footprint lidar waveform data in temperate deciduous forests, because temperate deciduous forests have been proved challenging in regard to lidar-derived biomass estimations.     

基于GF-1 WFV影像和机器学习算法的玉米叶面积指数估算

目的 叶面积指数（LAI）是重要的植被生物理化参数,对农作物长势和产量预测具有重要研究意义。基于物理模型和经验模型的LAI估算方法被认为是当前最常用的方法,但两种方法的估算效率和精度有限。近年来,机器学习算法在遥感监测领域广泛应用,算法具有描述非线性数据拟合、融合更多辅助信息的能力,为了评价机器学习算法在玉米LAI遥感估算中的适用性,本文分析比较了随机森林和BP神经网络算法估算玉米LAI的能力,并与传统经验模型进行了比较。方法 以河北省怀来县东花园镇为研究区,基于野外实测玉米LAI数据,结合同时期国产高分卫星（GF1-WFV影像）,首先分析了8种植被指数与LAI的相关性,进而采用保留交叉验证的方式将所有样本数据分为两部分,65%的数据作为模型训练集,35%作为验证集,重复随机分为3组,构建以8种植被指数为自变量,对应LAI值为因变量的RF模型、BP神经网络模型及传统经验模型。采用决定系数R²和均方根误差（RMSE）作为模型评价指标。结果 8种植被指数与LAI的相关性分析表明所有样本数据中,实测LAI值与各植被指数均在（P<0.01）水平下极显著相关,且相关系数均高于0.5;将3组不同样本数据在随机森林、BP神经网络算法中多次训练,并基于验证数据集进行估算精度检验,经验模型采用训练数据集建模,验证数据集检验,结果表明,RF模型表现出了较强的预测能力,LAI预测值与实测值R²分别为0.681、0.757、0.701,均高于BP模型（0.504、0.589、0.605）和经验模型（0.492、0.557、0.531）,对应RMSE分别为0.264、0.292、0.259;均低于BP模型（0.284、0.410、0.283）和经验模型（0.541、0.398、0.306）。结论 研究表明,RF算法能更好地进行玉米LAI遥感估算,为快速准确进行农作物LAI遥感监测提供了技术参考。     

Estimation of tropical forest biomass using Landsat TM imagery and permanent plot data in Xishuangbanna,China

Because of its complexity, it is very difficult to obtain information about distribution of biomass in tropical forests. This article describes the estimation of tropical forest biomass by using Landsat TM and forest plot data in Xishuangbanna, PR China. The method includes several steps. First, the biomass for each forest permanent plot is calculated by using field inventory data. Second, Landsat TM images are geometrically corrected by using topographic maps. Third, a map of the tropical forest is obtained by using data from a variety of sources such as Landsat TM, digital elevation model (DEM), temperature and precipitation layers and expert knowledge. Finally, the biomass and carbon storage of each forest vegetation type in the forest map is calculated by using the tropical forest map and the forest plot biomass GIS database. In the study area, forest area accounts for 57% of the total 1.7?×?10⁶ hectares. The total forest biomass is 2.0?×?10⁸ tonne. It is shown that the forest vegetation map, the forest biomass and the forest carbon storage can be obtained by effectively integrating Landsat TM, ancillary data including DEM, temperature and precipitation, forest permanent plots and knowledge using the method proposed here.     

光学遥感舰船目标识别方法

提出一种基于粗糙集理论和分层判别回归技术的光学遥感舰船目标识别方法。该方法首先提出新的光学遥感舰船识别特征———面积比编码,并与四类特征组合作为备选特征;然后基于粗糙集理论按同可区分度来计算各备选特征的重要性权值,自动选择出对正确识别贡献较大的特征组合;最后根据分层判别回归原理生成分类判决树来识别光学遥感舰船目标。实验结果表明,本文方法在识别精度和速度方面优于最近邻和支持向量机方法,且通用可行。     

Estimation of habit-related information from male voice data using machine learning-based methods

Artificial Life and Robotics - According to a survey on the cause of death among Japanese people, lifestyle-related diseases (such as malignant neoplasms, cardiovascular diseases, and pneumonia)...     

Exploring LiDAR-RaDAR synergy—predicting aboveground biomass in a southwestern ponderosa pine forest using LiDAR, SAR and InSAR

Scanning Light Detecting and Ranging (LiDAR), Synthetic Aperture Radar (SAR) and Interferometric SAR (InSAR) were analyzed to determine (1) which of the three sensor systems most accurately predicted forest biomass, and (2) if LiDAR and SAR/InSAR data sets, jointly considered, produced more accurate, precise results relative to those same data sets considered separately. LiDAR ranging measurements, VHF-SAR cross-sectional returns, and X- and P-band cross-sectional returns and interferometric ranges were regressed with ground-estimated (from dbh) forest biomass in ponderosa pine forests in the southwestern United States. All models were cross-validated. Results indicated that the average canopy height measured by the scanning LiDAR produced the best predictive equation. The simple linear LiDAR equation explained 83% of the biomass variability (n = 52 plots) with a cross-validated root mean square error of 26.0 t/ha. Additional LiDAR metrics were not significant to the model. The GeoSAR P-band (λ = 86 cm) cross-sectional return and the GeoSAR/InSAR canopy height (X-P) captured 30% of the forest biomass variation with an average predictive error of 52.5 t/ha. A second RaDAR-FOPEN collected VHF (λ ∼ 7.8 m) and cross-polarized P-band (λ = 88 cm) cross-sectional returns, none of which proved useful for forest biomass estimation (cross-validated R² = 0.09, RMSE = 63.7 t/ha). Joint consideration of LiDAR and RaDAR measurements produced a statistically significant, albeit small improvement in biomass estimation precision. The cross-validated R² increased from 83% to 84% and the prediction error decreased from 26.0 t/ha to 24.9 t/ha when the GeoSAR X-P interferometric height is considered along with the average LiDAR canopy height. Inclusion of a third LiDAR metric, the 60th decile height, further increased the R² to 85% and decreased the RMSE to 24.1 t/ha. On this 11 km² ponderosa pine study area, LiDAR data proved most useful for predicting forest biomass. RaDAR ranging measurements did not improve the LiDAR estimates.     

Characterizing forest ecological structure using pulse types and heights of airborne laser scanning

Characterizing forest structure is an important part of any comprehensive biodiversity assessment. However, current methods for measuring structural complexity require a laborious process that involves many logistically expensive point based measurements. An automated or semi-automated method would be ideal. In this study, the utility of airborne laser scanning (LiDAR; Light Detection and Ranging) for characterizing the ecological structure of a forest landscape is examined. The innovation of this paper is to use different laser pulse return properties from a full waveform LiDAR to characterize forest ecological structure. First the LiDAR dataset is stratified into four vertical layers: ground, low vegetation (0-1 m from the ground), medium vegetation (1-5 m from the ground) and high vegetation (> 5 m). Subsequently the “Type” of LiDAR return is analysed: Type 1 (singular returns); Type 2 (first of many returns); Type 3 (intermediate returns); and Type 4 (last of many returns). A forest characterization scheme derived from LiDAR point clouds is proposed. A validation of the scheme is then presented using a network of field sites that recorded commonly used metrics of biodiversity. The proposed forest characterization categories allow for quantification of gaps (above bare ground, low vegetation and medium vegetation), canopy cover and its vertical density as well as the presence of various canopy strata (low, medium and high). Regression analysis showed that LiDAR derived variables were good predictors of field recorded variables (R² = 0.82, P < 0.05 between LiDAR derived presence of low vegetation and field derived LAI for low vegetation). The proposed scheme clearly shows the potential of full waveform LiDAR to provide information on the complexity of habitat structure.     

Fast damage mapping in case of earthquakes using multitemporal SAR data

This work shows that earthquake damages in urban areas can be determined with an acceptable accuracy through the exploitation of multitemporal SAR data and ancillary information defining urban blocks. In this article, two different methodologies are presented: an unsupervised statistical analysis of the parameters of the models representing backscatterer intensity or coherence values for each block of the urban area under analysis, and a supervised approach which involves a multi-band/multi-temporal classification, performed using a Markov Random Field (MRF) classifier or a spatial Fuzzy ARTMAP (FA) classifier. The two procedures are compared by using ERS images acquired before and after the earthquake of Turkey in 1999.

结合Landsat时序遥感干扰检测的福建将乐县森林林龄估算与评价

高精度的森林林龄可以改善森林生物量、蓄积量、碳储存量等的估算精度。为提高频繁发生干扰区域森林年龄估算精度,以森林干扰强度较大的福建省将乐县为例,通过构建将乐县1987～2019年Landsat时序数据集,利用LandTrendr算法获得森林干扰开始时间节点特征,与林龄建模,实现干扰区林龄估算;接着利用GF-1号影像的波段、植被指数、纹理以及地形因子特征,通过递归特征消除的随机森林算法,与林龄建模,实现非干扰区林龄估算;最后将两部分的林龄合并,得到研究区2019年森林年龄。结果表明：(1)将乐县森林干扰总面积为346.37 km2,其中,针叶林、阔叶林干扰面积占比75.06%;(2)利用LandTrendr算法的干扰开始时间节点估算的林龄误差（RMSE=1.91 a）较小,模型精度（R2=0.94）较高;(3)通过递归特征消除的随机森林算法估算的针叶林、阔叶林林龄的R2和RMSE分别为0.64、0.48和4.71 a、12.71 a。研究表明：结合长时间序列的干扰算法可以有效提高干扰区森林年龄估算精度,为亚热带山区的区域尺度上进行森林林龄估计提供参考。     

基于深度网络的遥感图像目标检测与识别

本文针对卫星遥感图像中的典型目标检测与识别问题,采用了基于深度神经网络的回归算法,利用深度网络所学习出来的特征结构针对多分类复杂场景具有很好的识别效果,同时将检测、分类与识别一同进行回归处理。可以有效缩短训练时间,并提升处理效率,保证目标检测的准确度,在实际应用中前景广泛。     

Structural analysis and detection of android botnets using machine learning techniques

Nowadays, smartphone devices are an integral part of our lives since they enable us to access a large variety of services from personal to banking. The worldwide popularity and adoption of smartphone devices continue to approach the capabilities of traditional computing environments. The computer malware like botnets is becoming an emerging threat to users and network operators, especially on popular platform such as android. Due to the rapid growth of botnet applications, there is a pressing need to develop an effective solution to detect them. Most of the existing detection techniques can detect only malicious android applications, but it cannot detect android botnet applications. In this paper, we propose a structural analysis-based learning framework, which adopts machine learning techniques to classify botnets and benign applications using the botnet characteristics-related unique patterns of requested permissions and used features. The experimental evaluation based on real-world benchmark datasets shows that the selected patterns can achieve high detection accuracy with low false positive rate. The experimental and statistical tests show that the support vector machine classifier performs well compared to other classification algorithms.     

Remote sensing estimates of boreal and temperate forest woody biomass: carbon pools, sources, and sinks

The relation between satellite measurements of the normalized difference vegetation index (NDVI), cumulated over the growing season, and inventory estimates of forest woody biomass carbon is estimated statistically with data from 167 provinces and states in six countries (Canada, Finland, Norway, Russia and the USA for a single time period and Sweden for two periods). Statistical tests indicate that the regression model can be used to represent the relation between forest biomass and NDVI across spatial, temporal and ecological scales for relatively long time scales. For the 1.42 billion ha of boreal and temperate forests in the Northern Hemisphere, the woody biomass carbon pools and sinks are estimated at a relatively high spatial resolution (8×8 km). We estimate the carbon pool to be 61±20 gigatons (10⁹) carbon (Gt C) during the late 1990s and the biomass sink to be 0.68±0.34 Gt C/year between the 1982 and 1999. The geographic detail of carbon sinks provided here can contribute to a potential monitoring program for greenhouse gas emission reduction commitments under the Kyoto Protocol.