Modelling forest cover attributes as continuous variables in a regional context with Thematic Mapper data

We modelled forest vegetation attributes as continuous variables across western Oregon using a multi-image mosaic of Thematic Mapper (TM) data. Four specific attributes were modelled using regression analysis: percent green vegetation cover, percent conifer cover, conifer crown diameter, and conifer stand age. Reference data for the cover and diameter attributes were derived from airphotos, and existing agency polygon databases were used for stand age. We developed and applied a new method for regional mapping called applied radiometric normalization. The method involved development of a set of models for a centrally located 'source' scene which were then extended to 'destination' scenes (neighboring scenes in the TM mosaic). Use of airphotos and existing digital databases in combination with applied radiometric normalization translates to a cost-effective procedure for regional mapping with TM data. Modelling forest attributes as continuous variables enables creation of a flexible forest cover information base, containing important fundamental building blocks for a variety of related classification schemes.     

Estimating proportional change in forest cover as a continuous variable from multi-year MODIS data

This article describes a series of fundamental analyses designed to test and compare the utility of various MODIS data and products for detecting land cover change over a large area of the tropics. The approach for estimating proportional forest cover change as a continuous variable was based on a reduced major axis regression model. The model relates multispectral and multi-temporal MODIS data, transformed to optimize the spectral detection of vegetation changes, to reference change data sets derived from a Landsat data record for several study sites across the Central American region. Three MODIS data sets with diverse attributes were evaluated on model consistency, prediction accuracy and practical utility in estimating change in forest cover over multiple time intervals and spatial extents.A spectral index based on short-wave infrared information (normalized difference moisture index), calculated from half-kilometer Calibrated Radiances data sets, generally showed the best relationships with the reference data and the lowest model prediction errors at individual study areas and time intervals. However, spectral indices based on atmospherically corrected surface reflectance data, as with the Vegetation Indices and Nadir Bidirectional Reflectance Distribution Function - Adjusted Reflectance (NBAR) data sets, produced consistent model parameters and accurate forest cover change estimates when modeling over multiple time intervals. Models based on anniversary date acquisitions of the one-kilometer resolution NBAR product proved to be the most consistent and practical to implement. Linear regression models based on spectral indices that correlate with change in the brightness, greenness and wetness spectral domains of these data estimated proportional change in forest cover with less than 10% prediction error over the full spatial and temporal extent of this study.     

Estimating light absorption by chlorophyll, leaf and canopy in a deciduous broadleaf forest using MODIS data and a radiative transfer model

In this paper, we present a theoretical and modeling framework to estimate the fractions of photosynthetically active radiation (PAR) absorbed by vegetation canopy (FAPAR_canopy), leaf (FAPAR_leaf ), and chlorophyll (FAPAR_chl), respectively. FAPAR_canopy is an important biophysical variable and has been used to estimate gross and net primary production. However, only PAR absorbed by chlorophyll is used for photosynthesis, and therefore there is a need to quantify FAPAR_chl. We modified and coupled a leaf radiative transfer model (PROSPECT) and a canopy radiative transfer model (SAIL-2), and incorporated a Markov Chain Monte Carlo (MCMC) method (the Metropolis algorithm) for model inversion, which provides probability distributions of the retrieved variables. Our two-step procedure is: (1) to retrieve biophysical and biochemical variables using coupled PROSPECT + SAIL-2 model (PROSAIL-2), combined with multiple daily images (five spectral bands) from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor; and (2) to calculate FAPAR_canopy, FAPAR_leaf and FAPAR_chl with the estimated model variables from the first step. We evaluated our approach for a temperate forest area in the Northeastern US, using MODIS data from 2001 to 2003. The inverted PROSAIL-2 fit the observed MODIS reflectance data well for the five MODIS spectral bands. The estimated leaf area index (LAI) values are within the range of field measured data. Significant differences between FAPAR_canopy and FAPAR_chl are found for this test case. Our study demonstrates the potential for using a model such as PROSAIL-2, combined with an inverse approach, for quantifying FAPAR_chl, FAPAR_leaf, FAPAR_canopy, biophysical variables, and biochemical variables for deciduous broadleaf forests at leaf- and canopy-levels over time.     

Estimating forest canopy density of tropical mixed deciduous vegetation using Landsat data: a comparison of three classification approaches

Although a number of image classification approaches are available to estimate forest canopy density (FCD) using satellite data, assessment of their relative performances with tropical mixed deciduous vegetation is lacking. This study compared three image classification approaches – maximum likelihood classification (MLC), multiple linear regression (MLR) and FCD Mapper – in estimating the FCD of mixed deciduous forest in Myanmar. The application of MLC and MLR was based on spectral reflectance of vegetation, whereas FCD Mapper was operated on integrating the biophysical indices derived from the reflectance of the vegetation. The FCD was classified into four categories: closed canopy forest (CCF; FCD ≥ 70%), medium canopy forest (MCF; 40% ≥ FCD < 70%), open canopy forest (OCF; 10% ≥ FCD < 40%) and non-forest (NF; FCD < 10%). In the three classification approaches, producer's and user's accuracies were higher for more homogeneous vegetation such as NF and CCF than for heterogeneous vegetation density (VD) such as OCF and MCF. FCD Mapper produced the best overall accuracy and kappa coefficient. This study revealed that only spectral reflectance is not enough to get good results in estimating FCD in tropical mixed deciduous vegetation. This study indicates that FCD Mapper, an inexpensive approach because it requires only validation data and thus saves time, can be applied to monitor tropical mixed deciduous vegetation over time at lower cost than alternative methods.     

融合随机森林的偏最小二乘法及其中医药数据分析

偏最小二乘方法在多元线性回归建模中存在着诸多优势,但其本质还是线性回归,难以满足中医药非线性的特性。而随机森林构建的回归模型是由多个多元线性片段构成,对非线性数据有良好的拟合效果。本文提出了一种融合随机森林的偏最小二乘方法,该方法主要是利用PLS不断提取主成分并累计,利用随机森林算法将这些主成分分别与原始被解释变量不断构建多棵决策树,直到满足精度条件为止。分别采用麻杏石甘汤君药止咳、平喘和UCI数据集的数据进行分析处理,实验结果表明,融合随机森林的偏最小二乘分析方法对中医药数据有很好的适应性。     

Characterizing temperate forest structural and spectral diversity with Hyperion EO-1 data

Forest structural diversity can serve as an important indicator of biodiversity. The relationship between spaceborne hyperspectral remotely sensed data and several measures of forest structure was explored over a 625 km² coastal temperate forest landscape on Vancouver Island, British Columbia, Canada. Thirteen Hyperion bands were selected for analysis based on the documented and hypothesized importance of various spectral wavelengths to forest characterization. To aid in understanding spectral trends, measures of forest stand structural diversity (projected age, projected height, and stand species composition complexity) were derived from forest inventory data. The spectral distance between the stand mean and standard deviation of reflectance and related expectations from global equivalents for each of the 13 bands were used to relate measures of spectral diversity (N = 801 forest inventory stands).Canonical correlation analysis was then used to determine the independent and shared relationships between these selected measures of forest structural diversity (dependent variables) and spectral diversity (independent variables). The dependent variables that were most strongly correlated with the first canonical variate were projected age and projected height, with canonical loadings of 0.973 and 0.979, respectively. In contrast, stand species composition complexity had a weak, negative correlation with spectral diversity (canonical loading = − 0.025). The wavelengths contributing the most to the canonical function included: 681-740 nm, 551-680 nm, and 1401-2400 nm. There have been few studies that attempt to directly link spectral and species diversity in temperate forest environments. From this initial investigation, we posit that the complex spectral response of coastal temperate forests may confound efforts to directly link spectral and species diversity across a range of site conditions.Our results, which are constrained by the spectral and spatial resolution of the data used, our target environment, and the metrics selected for measuring forest structure, suggest that attributes that characterize forest structural conditions may have a more meaningful relationship with spectral diversity than measures of species diversity alone, and that future studies in coastal temperate forests that seek to link spectral diversity with biodiversity should include measures of forest structural diversity, in addition to measures of species diversity.     

Comparing annual MODIS and PRODES forest cover change data for advancing monitoring of Brazilian forest cover

Annual forest cover loss indicator maps for the humid tropics from 2000 to 2005 derived from time-series 500 m data from the MODerate Resolution Imaging Spectroradiometer (MODIS) were compared with annual deforestation data from the PRODES (Amazon Deforestation Monitoring Project) data set produced by the Brazilian National Institute for Space Research (INPE). The annual PRODES data were used to calibrate the MODIS annual change indicator data in estimating forest loss for Brazil. Results indicate that MODIS data may be useful in providing a first estimate of national forest cover change on an annual basis for Brazil. When directly compared with PRODES change at the MODIS grid scale for all years of the analysis, MODIS change indicator maps accounted for 75% of the PRODES change. This ratio was used to scale the MODIS change indicators to the PRODES area estimates. A sliding threshold of percent PRODES forest and 2000 to 2005 deforestation classes per MODIS grid cell was used to match the scaled MODIS to the official PRODES change estimates, and then to differentiate MODIS change within various sub-areas of the PRODES analysis. Results indicate significant change outside of the PRODES-defined intact forest class. Total scaled MODIS change area within the PRODES historical deforestation and forest area of study is 120% of the official PRODES estimate. Results emphasize the importance of synoptic monitoring of all forest change dynamics, including the cover dynamics of intact humid forest, regrowth, plantations, and cerrado tree cover assemblages. Results also indicate that operational MODIS-only forest cover loss algorithms may be useful in providing near-real time areal estimates of annual change within the Amazon Basin.     

Characterization of seasonal variation of forest canopy in a temperate deciduous broadleaf forest, using daily MODIS data

In this paper, we present an improved procedure for collecting no or little atmosphere- and snow-contaminated observations from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor. The resultant time series of daily MODIS data of a temperate deciduous broadleaf forest (the Bartlett Experimental Forest) in 2004 show strong seasonal dynamics of surface reflectance of green, near infrared and shortwave infrared bands, and clearly delineate leaf phenology and length of plant growing season. We also estimate the fractions of photosynthetically active radiation (PAR) absorbed by vegetation canopy (FAPAR_canopy), leaf (FAPAR_leaf), and chlorophyll (FAPAR_chl), respectively, using a coupled leaf-canopy radiative transfer model (PROSAIL-2) and daily MODIS data. The Markov Chain Monte Carlo (MCMC) method (the Metropolis algorithm) is used for model inversion, which provides probability distributions of the retrieved variables. A two-step procedure is used to estimate the fractions of absorbed PAR: (1) to retrieve biophysical and biochemical variables from MODIS images using the PROSAIL-2 model; and (2) to calculate the fractions with the estimated model variables from the first step. Inversion and forward simulations of the PROSAIL-2 model are carried out for the temperate deciduous broadleaf forest during day of year (DOY) 184 to 201 in 2005. The reproduced reflectance values from the PROSAIL-2 model agree well with the observed MODIS reflectance for the five spectral bands (green, red, NIR₁, NIR₂, and SWIR₁). The estimated leaf area index, leaf dry matter, leaf chlorophyll content and FAPAR_canopy values are close to field measurements at the site. The results also showed significant differences between FAPAR_canopy and FAPAR_chl at the site. Our results show that MODIS imagery provides important information on biophysical and biochemical variables at both leaf and canopy levels.     

Estimating aboveground biomass of grassland having a high canopy cover: an exploratory analysis of in situ hyperspectral data

To improve the estimation of aboveground biomass of grassland having a high canopy cover based on remotely sensed data, we measured in situ hyperspectral reflectance and the aboveground green biomass of 42 quadrats in an alpine meadow ecosystem on the Qinghai–Tibetan Plateau. We examined the relationship between aboveground green biomass and the spectral features of original reflectance, first-order derivative reflectance (FDR), and band-depth indices by partial least squares (PLS) regression, as well as the relationship between the aboveground biomass and narrow-band vegetation indices by linear and nonlinear regression analyses. The major findings are as follows. (1) The effective portions of spectra for estimating aboveground biomass of a high-cover meadow were within the red-edge and near infrared (NIR) regions. (2) The band-depth ratio (BDR) feature, using NIR region bands (760–950 nm) in combination with the red-edge bands, yields the best predictive accuracy (RMSE?=?40.0 g m^?2) for estimating biomass among all the spectral features used as independent variables in the partial least squares regression method. (3) The ratio vegetation index (RVI2) and the normalized difference vegetation index (NDVI2) proposed by Mutanga and Skidmore (Mutanga, O. and Skidmore, A.K., 2004a Mutanga, O. and Skidmore, A. K. 2004a. Narrow band vegetation indices solve the saturation problem in biomass estimation. International Journal of Remote Sensing, 25: 1–16.  [Google Scholar], Narrow band vegetation indices solve the saturation problem in biomass estimation. International Journal of Remote Sensing, 25, pp. 1–6) are better correlated to the aboveground biomass than other VIs (R ²?=?0.27 for NDVI2 and 0.26 for RVI2), while RDVI, TVI and MTV1 predicted biomass with higher accuracy (RMSE?=?37.2 g m^?2, 39.9 g m^?2 and 39.8 g m^?2, respectively). Although all of the models developed in this study are probably acceptable, the models developed in this study still have low accuracy, indicating the urgent need for further efforts.     

Mapping deciduous forest ice storm damage using Landsat and environmental data

Sugar maple (Acer Saccharum Marsh.) damage resulting from a severe ice storm was modeled and mapped over eastern Ontario using pre- and post-storm Landsat 5 imagery and environmental data. Visual damage estimates in 104 plots and corresponding reflectance and environmental data were divided into multiple, mutually exclusive training and reference datasets for damage classification evaluation. Damage classification accuracy was compared among four methods: multiple regression, linear discriminant analysis, maximum likelihood, and neural networks. Using the best classifier, various stratification methods were assessed for potential inflationary effects on classification accuracy due to spatial proximity between training and reference data. Of the classifiers that were evaluated, neural networks performed best. Neural networks ‘learn’ training data accurately (94% overall), but classify proximate reference data less accurately (65%), and distant, spatially independent reference data least accurately (55%). Results indicate that, while remotely sensed and environmental data cannot discriminate among many levels of deciduous ice storm damage, they can by considered useful for differentiating areas of low to medium damage from areas of severe damage (69% accuracy). Such classification methods can provide regional damage maps more objectively than point-based visual estimates or aerial sketch mapping and aid in identification of areas of severe damage where management intervention may be advantageous.     

Estimating stand structural details using nearest neighbor analyses to link ground data, forest cover maps, and Landsat imagery

Forest information over a landscape is often represented as a spatial mosaic of polygons, separated by differences in species composition, height, age, crown closure, productivity, and other variables. These polygons are commonly delineated on medium-scale photography (e.g., 1:15,000) by a photo-interpreter familiar with the inventory area, and displayed and stored in a Geographic Information System (GIS) layer as a forest cover map. Forest cover maps are used for multiple purposes including timber and habitat supply analyses, and carbon inventories, at a regional or management unit level, and for parks planning, operational planning, and selection of stands for many purposes at a local level. Attribute data for each polygon commonly include the variables used to delineate the polygon, and other variables that can be measured or estimated using these medium-scale photographs. Additional measures that can only be obtained via expensive ground measures or possibly on high resolution photographs (e.g., volume per unit area, biomass components per unit area, tree-list of species and diameters) are available only for a sample of polygons, or may have been gathered independently using a sample survey over the land area. Improved linkages over a variety of data sources may help to support landscape level analyses. This study presents an approach to combine information from a systematic (grid) ground survey, forest cover (polygon) data, and Landsat Thematic Mapper (TM) imagery using variable-space nearest neighbor methods to estimate (i) mean ground-measured attributes for each polygon, in particular, volume per ha (m³/ha), stems per ha, and quadratic mean diameter for each polygon; and (ii) variation of these ground attributes within polygons. The approach was initially evaluated using Monte Carlo simulations with known measures of these attributes. Nearest neighbor methods were then applied to an approximate 5000 ha area (about 1000 polygons) of high productivity, mountainous forests located near the Pacific Coast of British Columbia, Canada. Based on the simulation results, the use of Landsat pixel reflectances to estimate volume per ha, average tree size (i.e., quadratic mean diameter), and stems per ha did not show great promise in improving estimates for each polygon over using forest cover data alone. However, in application, the use of remotely sensed data provided estimates of within-polygon variability. At the same time, the estimated means of these three imputed variables over the entire study area were very similar to the representative sample estimates using the ground data only. Extensions to other variables such as ranges of diameters and numbers of snags may also be possible providing useful data for habitat and forest growth analysis.     

Estimating fractional cover of plant functional types in African savannah from harmonic analysis of MODIS time-series data

Assessments of tree/grass fractional cover in savannahs using remote sensing are challenging due to the heterogeneous mixture of the two plant functional types. Time-series decomposition models can be used to characterize vegetation phenology from satellite data, but have rarely been used for attributing phenological signal components to different plant functional types. Here, tree/grass dynamics are assessed in savannah ecosystems using time-series decomposition of 14 years of Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index data acquired from 2002 to 2015. The decomposition method uses harmonic analysis and tests the individual harmonic terms for statistical significance. Field data of fractional cover of trees and grasses were collected for 28 plots in Kruger National Park, South Africa. Matching MODIS pixels were analysed for their tree/grass phenological signals. Tree/grass annual and interannual variability were then assessed based on the harmonic models. In most harmonic cycles, grass-dominated sites had higher amplitudes than tree-dominated sites, while the tree green-up started earlier than grasses, before the start of the wet season. While changes in tree phenology are gradual, grasses present higher variability over time. Tree cover showed a significant correlation with the amplitude (r (correlation coefficient) = ?0.59, p = 0.001) and phase of the first harmonic term (r = ?0.73, p = 0.0001) and the number of cycles of the second harmonic term (r = 0. 56, p = 0.002). Grass cover was also significantly correlated with the amplitude (r = 0. 51, p = 0.005) and phase of the first harmonic term (r = 0.55, p = 0.002) and the number of cycles of the second harmonic term (r = ?0.52, p = 0.005). The positive correlation of grass cover with phase and negative correlation with number of cycles is indicating a late greening period and higher variability, respectively. Tree cover estimated from the phase of the strongest harmonic term showed a positive correlation with field-measured tree cover (R² (coefficient of determination) = 0.55, p < 0.01, slope = 0.93, root mean square error = 13.26%). The estimated tree cover also had a strong correlation with the woody cover map (r = 0.78, p < 0.01) produced by Bucini. The results show that MODIS time-series data can be used to estimate the fractional tree cover in heterogeneous savannahs from the phase of the plant functional type’s phenological behaviour. This study shows that harmonic analysis is able to discriminate between fractional cover by trees and grasses in savannahs. The quantitative analysis of tree/grass phenology from satellite time-series data enables a better understanding of the dynamics of the tree/grass competition and coexistence.     

Subpixel measurement of tropical forest cover using AVHRR data

Abstract

Tropical forest assessment using data from the Advanced Very High Resolution Radiometer (AVHRR) may lead to inaccurate estimates of forest cover in regions of small subpixel forest or non-forest patches and in regions where the pattern of clearance is particularly convoluted. Test sites typifying these two patterns were chosen in Ghana and Rondonia, respectively. To capture the subpixel proportions of forest cover, a linear mixture model was applied to two AVHRR test images over the test sites. The model produced image outputs in which pixel intensities indicated the proporton of forest cover per km². For comparison, supervised maximum likelihood classifications were also performed. The outputs were assessed against classified Landsat TM scenes, converted to proportions maps and coregistered to the AVHRR images. An empirical method was applied for determining the critical forest cover per km² needed for an AVHRR pixel to be classified as forest. The critical values exceeded 50 per cent, indicating a tendency for AVHRR classification to underestimate forest cover. This was confirmed by comparing estimates of total forest cover obtained from the AVHRR and TM classifications. In the case of Ghana, a more accurate estimate of forest cover was obtained from the AVHRR mixture model than from the classification. Both mixture model outputs were found to be well correlated with those from Landsat TM. Further work should test the robustness of the approach adopted here when applied to much larger areas.     

Estimating field-plot data of forest stands using airborne laser scanning and SPOT HRG data

Airborne laser scanner systems provide detailed forest information that can be used for important improvements in forest management decisions. Planning systems under development use plot-survey data to represent forest stands in large forest holdings which enables new flexible methods to model the forest and optimize selection of silvicultural treatments. In Sweden today, only averages of forest stand variables are used, and the survey methods used do not provide plot-survey data for all stands in large forest holdings. This is a task possibly solved using airborne laser scanner data. Various measures can be derived from laser data, each describing different forest variables, such as tree height distribution, vegetation density and vertical tree crown structure. Here, imputation of field plot (10 m radius) data using measures derived from airborne laser scanner data (TopEye) and optical image data (SPOT 5 HRG satellite sensor) were evaluated as a method to provide data for new long-term management planning systems. In addition to commonly applied measures, the semivariogram of laser measurements was evaluated as a new measure to extract spatial characteristics of the forest. The study used data from 870, 10 m radius field plots (0 to 812 m³ ha^− 1) surveyed for a 1200 ha large forest estate in the south of Sweden. At the best, combining measures derived from laser scanner data and SPOT 5 data, stand mean volume was estimated with a root mean square error (RMSE) of 20% of the sample mean and stem density with 22% RMSE. Bias of stem density estimates was 5%, and stand stem volume 4%. Although these accuracies are sufficient for operational application, estimates of tree species proportions and within-stand variation were clearly not.     

Subpixel forest cover in central Africa from multisensor, multitemporal data

Seven Landsat Multispectral Scanner (MSS) scenes in central Africa were coregistered with 8 km resolution data from the 1987 AVHRR Pathfinder Land data set. Percent forest cover in each 8 km grid cell was derived from the classified MSS scenes. Linear relationships between percent forest cover and 30 multitemporal metrics derived from all AVHRR optical and thermal channels were determined. Correlations were strongest for the mean annual normalized difference vegetation index (NDVI) and mean annual brightness temperature (AVHRR Channel 3) and weakest for those metrics, besides NDVI, based on near-infrared reflectances (AVHRR Channel 2). The relationships were used to estimate percent forest cover in various locations in the study area using multiple linear regression and regression trees. Overall, the multiple linear regression provided more accurate results. Predicted percent forest cover estimates were within 20% of the “actual” percent forest cover (derived from the MSS data) for approximately 90% of the grid cells. The RMS error for the prediction was 12% forest cover. RMS errors above 18% forest cover were obtained when using AVHRR data from a single month to derive predictive relationships. The results demonstrate that multitemporal data reflecting vegetation phenology can be used to estimate subpixel forest cover at coarse spatial resolutions.     

The method of attributes for data flow analysis

Summary The importance of producing data flow information on demand is discussed. The method of attributes is applied to the demand analysis of live variables.Part I of this paper described the method of attributes, which is a technique for high level data flow analysis. In that paper, the method was applied to two well-known problems: analysis of dead variables and analysis of available expressions. Both of these analyses are called exhaustive because they uncover information for all program points.In this part, we apply the method of attributes to a problem in demand data flow analysis.     

Retrieval of forest canopy attributes based on a geometric-optical model using airborne LiDAR and optical remote-sensing data

In the retrieval of forest canopy attributes using a geometric-optical model, the spectral scene reflectance of each component should be known as prior knowledge. Generally, these reflectances were acquired by a foregone survey using an analytical spectral device. This article purposed to retrieve the forest structure parameters using light detection and ranging (LiDAR) data, and used a linear spectrum decomposition model to determine the reflectances of the spectral scene components, which are regarded as prior knowledge in the retrieval of forest canopy cover and effective plant area index (PAI_e) using a simplified Li–Strahler geometric-optical model based on a Satellites Pour l'Observation de la Terre 5 (SPOT-5) high-resolution geometry (HRG) image. The airborne LiDAR data are first used to retrieve the forest structure parameters and then the proportion of the SPOT pixel not covered by crown or shadow K_g of each pixel in the sample was calculated, which was used to extract the reflectances of the spectral scene components by a linear spectrum decomposition model. Finally, the forest canopy cover and PAI_e are retrieved by the geometric-optical model. As the acquired time of SPOT-5 image and measured data has a discrepancy of about 2 months, the retrieved result of forest canopy cover needs a further validation. The relatively high value of R ² between the retrieval result of PAI_e and the measurements indicates the efficiency of our methods.     

Estimating areal means and variances of forest attributes using the k-Nearest Neighbors technique and satellite imagery

The k-Nearest Neighbor (k-NN) technique has become extremely popular for a variety of forest inventory mapping and estimation applications. Much of this popularity may be attributed to the non-parametric, multivariate features of the technique, its intuitiveness, and its ease of use. When used with satellite imagery and forest inventory plot data, the technique has been shown to produce useful estimates of many forest attributes including forest/non-forest, volume, and basal area. However, variance estimators for quantifying the uncertainty of means or sums of k-NN pixel-level predictions for areas of interest (AOI) consisting of multiple pixels have not been reported. The primary objectives of the study were to derive variance estimators for AOI estimates obtained from k-NN predictions and to compare precision estimates resulting from different approaches to k-NN prediction and different interpretations of those predictions. The approaches were illustrated by estimating proportion forest area, tree volume per unit area, tree basal area per unit area, and tree density per unit area for 10-km AOIs. Estimates obtained using k-NN approaches and traditional inventory approaches were compared and found to be similar. Further, variance estimates based on different interpretations of k-NN predictions were similar. The results facilitate small area estimation and simultaneous and consistent mapping and estimation of multiple forest attributes.     

Modeling gross primary production of temperate deciduous broadleaf forest using satellite images and climate data

Net ecosystem exchange (NEE) of CO₂ between the atmosphere and forest ecosystems is determined by gross primary production (GPP) of vegetation and ecosystem respiration. CO₂ flux measurements at individual CO₂ eddy flux sites provide valuable information on the seasonal dynamics of GPP. In this paper, we developed and validated the satellite-based Vegetation Photosynthesis Model (VPM), using site-specific CO₂ flux and climate data from a temperate deciduous broadleaf forest at Harvard Forest, Massachusetts, USA. The VPM model is built upon the conceptual partitioning of photosynthetically active vegetation and non-photosynthetic vegetation (NPV) within the leaf and canopy. It estimates GPP, using satellite-derived Enhanced Vegetation Index (EVI), Land Surface Water Index (LSWI), air temperature and photosynthetically active radiation (PAR). Multi-year (1998-2001) data analyses have shown that EVI had a stronger linear relationship with GPP than did the Normalized Difference Vegetation Index (NDVI). Two simulations of the VPM model were conducted, using vegetation indices from the VEGETATION (VGT) sensor onboard the SPOT-4 satellite and the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor onboard the Terra satellite. The predicted GPP values agreed reasonably well with observed GPP of the deciduous broadleaf forest at Harvard Forest, Massachusetts. This study highlighted the biophysical performance of improved vegetation indices in relation to GPP and demonstrated the potential of the VPM model for scaling-up of GPP of deciduous broadleaf forests.     

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.