The effects of rectification and Global Positioning System errors on satellite image-based estimates of forest area

Satellite image-based maps of forest attributes are of considerable interest and are used for multiple purposes such as international reporting by countries that have no national forest inventory and small area estimation for all countries. Construction of the maps typically entails, in part, rectifying the satellite images to a geographic coordinate system, observing ground plots whose coordinates are obtained from Global Positioning System (GPS) receivers that are calibrated to the same geographic coordinate system, and then matching ground plots to image pixels containing the centers of the ground plots. Errors in rectification and GPS coordinates cause observations of ground attributes to be associated with spectral values of incorrect pixels which, in turn, introduces classification errors into the resulting maps. The most important finding of the study is that for common magnitudes of rectification and GPS errors, as many as half the ground plots may be assigned to incorrect pixels. The effects on areal estimates obtained by aggregating class predictions for individual pixels are deviation of the estimates from their true values, erroneous confidence intervals, and incorrect inferences. Results are reported in detail for both probability-based (design-based) and model-based approaches to inference for proportion forest area using maps constructed from Landsat imagery, forest inventory plot observations and a logistic regression model.     

Performance of different spectral and textural aerial photograph features in multi-source forest inventory

Most multi-source forest inventory (MSFI) applications have thus far been based on the use of medium resolution satellite imagery, such as Landsat TM. The high plot and stand level estimation errors of these applications have, however, restricted their use in forest management planning. One reason suggested for the high estimation errors has been the coarse spatial resolution of the imagery employed. Therefore, very high spatial resolution (VHR) imagery sources provide interesting data for stand-level inventory applications. However, digital interpretation of VHR imagery, such as aerial photographs, is more complicated than the use of traditional satellite imagery. Pixel-by-pixel analysis is not applicable to VHR imagery because a single pixel is small in relation to the object of interest, i.e. a forest stand, and therefore it does not adequately represent the spectral properties of a stand. Additionally in aerial photographs, the spectral properties of the objects are dependent on their location in the image. Therefore, MSFI applications based on aerial imagery must employ features that are less sensitive to their location in the image and that have been derived using the spatial neighborhood of each pixel, e.g. a square-shaped window of pixels. In this experiment several spectral and textural features were extracted from color-infrared aerial photographs and employed in estimation of forest attributes. The features were extracted from original, normalized difference vegetation index and channel ratio images. The correlations between the extracted image features and forest attributes measured from sample plots were examined. Additionally, the spectral and textural features were used for estimating the forest attributes of sample plots, applying the k nearest neighbor estimation method. The results show that several spectral and textural image features that are moderately or well correlated with the forest attributes. Furthermore, the accuracy of forest attribute estimation can be significantly improved by a careful selection of image features.     

Histogram matching for the calibration of kNN stem volume estimates

The k-nearest neighbours (kNN) methods have been used successfully in many countries for the production of spatially comprehensive raster databases of forest attributes, made from the combination of National Forest Inventory (NFI) and satellite data. In Sweden, country-wide kNN estimates of forest variables have been produced to represent the forest condition in the years 2000 and 2005 by using a combination of Système Pour l'Observation de la Terre 5 (SPOT 5) satellite data and field data from the Swedish NFI. The resulting products are wall-to-wall raster maps with estimates of total stem volume, stem volume per tree species, tree height and stand age and a 25?×?25 m² pixel resolution. However, probability-based kNN stem volume estimates tend to have a suppressed variation range as large values are usually underestimated and small values are overestimated. One way to handle this problem is to calibrate the kNN stem volume estimates to the reference distribution of stem volume observations by histogram matching (HM) for a defined geographic area.

In this study, we have tested HM for the calibration of kNN total stem volume raster maps to the reference distribution captured by a forest inventory (FI) from 106 stands in Strömsjöliden, in the north of Sweden. The available field FI data set comprises 1084 circular plots, divided into a reference data set and an evaluation data set of total stem volume observations. The reference data set was used for the creation of a cumulative frequency histogram of total stem volume and the evaluation data set was used to assess the accuracy of volume estimates, before and after HM. The HM adjusted the cumulative distribution of the kNN data set to the distribution of the reference observations and resulted in a distribution of kNN estimates of total stem volume, which corresponded closely to that of the evaluation data set. The results show that the variation range of the kNN stem volume estimates can be extended by HM both on the pixel and stand levels. The extension of the range of estimates towards the range provided by the field observations allows improvement of kNN volume estimation for use in forest management planning based on stand-level analysis, given that the reference stem volume distribution can be estimated accurately, for example, using field data from NFI.     

Using satellite imagery and the k-nearest neighbors technique as a bridge between strategic and management forest inventories

Strategic forest inventory programs produce forest resource estimates for large areas such as states and provinces using data collected for a large number of variables on a relatively sparse array of field plots. Management inventories produce stand-level estimates to guide management decisions using data obtained with sampling intensities much greater than for strategic inventories. The costs associated with these greater sampling intensities have motivated investigations of alternatives to traditional sample-based management inventories. This study focused on a relatively inexpensive alternative to management inventories that uses strategic forest inventory plot data, Landsat Thematic Mapper (TM) satellite imagery, and the k-Nearest Neighbors (k-NN) technique. The approach entailed constructing stem density and basal area per unit area maps from which stand-level means were estimated as averages of k-NN pixel predictions. The study included investigations of the benefits of selecting optimal combinations of k-NN feature space variables derived from the TM imagery and the benefits of modifying the k-NN technique to eliminate spurious nearest neighbors. For both the stem density and basal area per unit area training data, the selection of optimal feature space covariates produced less than 1.5% improvement in root mean square error relative to using all covariates. The k-NN modification improved the sum of mean squared deviations for stand-level stem density and basal area per unit area estimates by 7–20% depending on the k-NN feature space covariates. For the best combination of feature space covariates, estimates of stand-level means were within confidence intervals for validation estimates for 11 of 12 stands for stem density and for 10 of 12 stands for basal area per unit area.     

Application of two regression-based methods to estimate the effects of partial harvest on forest structure using Landsat data

Although partial harvests are common in many forest types globally, there has been little assessment of the potential to map the intensity of these harvests using Landsat data. We modeled basal area removal and percent cover change in a study area in central Washington (northwestern USA) using biennial Landsat imagery and reference data from historical aerial photos and a system of inventory plots. First, we assessed the correlation of Landsat spectral bands and associated indices with measured levels of forest removal. The variables most closely associated with forest removal were the shortwave infrared (SWIR) bands (5 and 7) and those strongly influenced by SWIR reflectance (particularly Tasseled Cap Wetness, and the Disturbance Index). The band and indices associated with near-infrared reflectance (band 4, Tasseled Cap Greenness, and the Normalized Difference Vegetation Index) were only weakly correlated with degree of forest removal. Two regression-based methods of estimating forest loss were tested. The first, termed “state model differencing” (SMD), involves creating a model representing the relationship between inventory data from any date and corresponding, cross-normalized spectral data. This “state model” is then applied to imagery from two dates, with the difference between the two estimates taken as estimated change. The second approach, which we called “direct change modeling” (DCM), involves modeling forest structure changes as a single term using re-measured inventory data and spectral differences from corresponding image pairs. In a leave-one-out cross-validation process, DCM-derived estimates of harvest intensity had lower root mean square errors than SMD for both relative basal area change and relative cover change. The higher measured accuracy of DCM in this project must be weighed against several operational advantages of SMD relating to less restrictive reference data requirements and more specific resultant estimates of change.     

A method for calibrated maximum likelihood classification of forest types

A modification to the maximum likelihood algorithm was developed for classification of forest types in Sweden's part of the CORINE land cover mapping project. The new method, called the “calibrated maximum likelihood classification” involves an automated and iterative adjustment of prior weights until class frequency in the output corresponds to class frequency as calculated from objective (field-inventoried) estimates. This modification compensates for the maximum likelihood algorithm's tendency to over-represent dominant classes and under-represent less frequent ones. National forest inventory plot data measured from a five-year period are used to estimate relative frequency of class occurrence and to derive spectral signatures for each forest class. The classification method was implemented operationally within an automated production system which allowed rapid production of a country-wide forest type map from Landsat TM/ETM+ satellite data. The production system automated the retrieval and updating of forest inventory plots, a plot-to-image matching routine, illumination and haze correction of satellite imagery, and classification into forest classes using the calibrated maximum likelihood classification. This paper describes the details of the method and demonstrates the result of using an iterative adjustment of prior weights versus unadjusted prior weights. It shows that the calibrated maximum likelihood algorithm adjusts for the overclassification of classes that are well represented in the training data as well as for other classes, resulting in an output where class proportions are close to those as expected based on forest inventory data.     

Combining national forest inventory field plots and remote sensing data for forest databases

Information about forest cover is needed by all of the nine societal benefit areas identified by the Group of Earth Observation (GEO). In particular, the biodiversity and ecosystem areas need information on landscape composition, structure of forests, species richness, as well as their changes. Field sample plots from National Forest Inventories (NFI) are, in combination with satellite data, a tremendous resource for fulfilling these information needs. NFIs have a history of almost 100 years and have developed in parallel in several countries. For example, the NFIs in Finland and Sweden measure annually more than 10,000 field plots with approximately 200 variables per plot. The inventories are designed for five-year rotations. In Finland nationwide forest cover maps have been produced operationally since 1990 by using the k-NN algorithm to combine satellite data, field sample plot information, and other georeferenced digital data. A similar k-NN database has also been created for Sweden. The potentials of NFIs to fulfil diverse information needs are currently analyzed also in the COST Action E43 project of the European Union. In this article, we provide a review of how NFI field plot information has been used for parameterization of image data in Sweden and Finland, including pre-processing steps like haze correction, slope correction, and the optimization of the estimation variables. Furthermore, we review how the produced small-area statistics and forest cover data have been used in forestry, including forest biodiversity monitoring and habitat modelling. We also show how remote sensing data can be used for post-stratification to derive the sample plot based estimates, which cannot be directly estimated from the spectral data.     

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.     

A comparative assessment of multi-temporal Landsat 8 and machine learning algorithms for estimating aboveground carbon stock in coppice oak forests

Remote sensing of low biomass forests has challenges related to the contribution of soil and understory reflectance recorded by sensors, hampering accurate forest aboveground carbon (AGC) quantification. To improve Landsat-based AGC estimates in forests with low biomass, this study explored the use of multi-temporal Landsat 8 Operational Land Imager (OLI) derived spectral information in Zagros forests by testing four machine learning algorithms: support vector machine (SVM), boosted regression trees (BRT), random forest (RF) and multivariate adaptive regression splines (MARS). We selected two forest areas with different levels of human activity for AGC reference plots: un-degraded forest (UD) and highly-degraded forest (HD). The results of the study showed that the Landsat image acquired in the peak of the growing season (10 August) provided the best AGC estimates for the UD site, but that for the HD site, AGC estimates were not affected by the timing of the imagery. The comparison of different modelling methods demonstrated lower accuracies from BRT, considerably biased estimates from SVM, and generally robust results from the RF algorithm. Overall, the study demonstrated the utility of applying the free Landsat 8 OLI dataset to AGC estimation, in particular non-commercial forests in developing countries where little budget is allocated for management.     

A model-based approach to estimating forest area

A logistic regression model based on forest inventory plot data and transformations of Landsat Thematic Mapper satellite imagery was used to predict the probability of forest for 15 study areas in Indiana, USA, and 15 in Minnesota, USA. Within each study area, model-based estimates of forest area were obtained for circular areas with radii of 5 km, 10 km, and 15 km and were compared to design-based estimates based on inventory plot data. Precision estimates for the circular areas were also obtained using variance formulae developed for this application that incorporated spatial correlation among model predictions for individual pixels. The model-based estimates were generally comparable to the design-based estimates. The advantages of the model-based approach are that maps and small areas estimates may be obtained and the necessity of releasing exact plot locations for user-specific applications is alleviated.     

Point accuracy of a non-parametric method in estimation of forest characteristics with different satellite materials

The reference sample plot (RSP) method is a distance-weighted k nearest neighbour estimation method, which allows simultaneous interpretation of several variables. In the RSP method, the k spectrally nearest field plots are looked at separately for each unknown pixel, and the area weight of the unknown pixel is divided as a function of the spectral distances to the nearest plots. The RSP method was examined in a forest inventory for estimating stem volumes by tree species groups using different satellite materials. Two methods were tested both in searching for and weighting the nearest field plots. Euclidean distance functions worked steadily with all the volume variables studied. The other distance measure tested was based on regression modelling. With more than 15 plots, both covariance weighting and inverse distance weighting gave similar results. Considering the field data of this study, the suitable number of the nearest plots in plotwise estimation appeared to be between 10 and 15 plots. With Landsat TM, SPOT XS and SPOT P, the differences in standard errors were minor. When combined TM and SPOT P were used, the plotwise standard error of total volume was still over 60 per cent.     

A two-step nearest neighbors algorithm using satellite imagery for predicting forest structure within species composition classes

Nearest neighbors techniques have been shown to be useful for predicting multiple forest attributes from forest inventory and Landsat satellite image data. However, in regions lacking good digital land cover information, nearest neighbors selected to predict continuous variables such as tree volume must be selected without regard to relevant categorical variables such as forest/non-forest. The result is that non-zero volume predictions may be obtained for pixels predicted to be non-forest, and volume predictions for pixels predicted to be forest may be erroneously small due to non-forest nearest neighbors. For users who wish to circumvent this discrepancy, a two-step algorithm is proposed in which the class of a relevant categorical variable such as land cover is predicted in the first step, and continuous variables such as volume are predicted in the second step subject to the constraint that all nearest neighbors must come from the predicted class of the categorical variable. Nearest neighbors, multinomial logistic regression, and discriminant analysis techniques were investigated for use in the first step. The results were generally similar for the three techniques, although the multinomial logistic regression technique was slightly superior. The k-Nearest Neighbors technique was used in the second step because many continuous forest inventory variables do not satisfy the distributional assumptions necessary for parametric multivariate techniques. The results for six 15-km × 15-km areas of interest in northern Minnesota, USA, indicate that areal estimates of tree volume, basal area, and density obtained from pixel predictions are comparable to plot-based estimates and estimates by conifer and deciduous classes are also comparable to plot-based estimates. When a mixed conifer/deciduous class was included, predictions for the mixed and deciduous class were confused.     

Estimation of forest Leaf Area Index using remote sensing and GIS data for modelling net primary production

Ecosystem models can be used to estimate potential net primary production (pNPP) using GIS data, and remote sensing input of actual forest leaf area to such models can provide estimates of current actual net primary production (aNPP) . Comparisons of pNPP and aNPP for a given site or regional landscape can be used to identify forest stands for different management treatments, and may provide new information on wildlife habitat, forest diversity and growth characteristics. Leaf area estimates may be obtained from satellite imagery through correlation with physiologically-based vegetation indices such as the Normalized Difference Vegetation Index (NDVI). However, in areas with high Leaf Area Index (LAI), vegetation indices usually saturate at leaf areas greater than about 4. In predominantly deciduous (hardwood) and mixedwood stands remote sensing estimates may be influenced by understory and other factors. We examined digital Landsat TM imagery and GIS data in the Fundy Model Forest of southeastern New Brunswick to determine relations to forest leaf area index within different stand structures or covertypes. The image data were stratified using GIS covertype information prior to development of LAI predictive equations using spectral reflectance, and the prediction of LAI from Landsat TM imagery was improved with reference to estimates of stem density which are standard forest inventory information contained in GIS databases. Actual stand LAI was compared to assumed maximum LAI values for several species and sites using an ecosystem process model (BIOME-BGC) which relies on climate, soils and topographic information also obtained from the GIS. Subsequent comparison of pNPP and aNPP revealed that even disturbed sites in this environment can reach close to maximum site potential. Specific sites with suboptimal species composition were identified. A future refinement of this approach is to classify the imagery independently of the GIS, which assumes a homogeneous covertype for each polygon in the system, and thus improve still further the aNPP estimates through higher covertype and LAI estimation accuracy.     

A technique for extrapolating and validating forest cover across large regions Calibrating AVHRR data with TM data

Abstract

An approach to extending high-resolution forest cover information across large regions is presented and validated. Landsat Thematic Mapper (TM) data were classified into forest and nonforest for a portion of Jackson County, Illinois. The classified TM image was then used to determine the relationship between forest cover and the spectral signature of Advanced Very High Resolution Radiometer (AVHRR) pixels covering the same location. Regression analysis was used to develop an empirical relationship between AVHRR spectral signatures and forest cover. The regression equation developed from data from the single county calibration area in southern Illinois was then applied to the entire AVHRR scene, which covered all or parts of ten states, to produce a regional map of forest cover. This map was used to derive estimates of forest cover, within a geographical information system (GIS), for each of the 428 counties located within the boundaries of the original AVHRR scene. The validity of the overall regional map was tested by comparing the AVHRR/TM-derived estimates of county forest cover with independent estimates of county forest cover developed by the U.S. Forest Service (USFS). The overall correlation coefficient of the AVHRR/TM and USFS county forest cover estimates was r=0-89 (n=428 counties). Not surpris0ingly, some individual states and the areas nearer to the southern Illinois calibration centre had higher correlation coefficients. Absolute estimates of forest cover percentages were also significantly well predicted. With the future inclusion of multiple calibration centres representing a number of physiographic regions, the method shows promise for predicting continental and global estimates of forest cover.     

Assessment of forest cover in Russia by combining a wall-to-wall coarse-resolution land-cover map with a sample of 30 m resolution forest maps

The process of gathering land-cover information has evolved significantly over the last decade (2000–2010). In addition to this, current technical infrastructure allows for more rapid and efficient processing of large multi-temporal image databases at continental scale. But whereas the data availability and processing capabilities have increased, the production of dedicated land-cover products with adequate accuracy is still a prerequisite for most users. Indeed, spatially explicit land-cover information is important and does not exist for many regions. Our study focuses on the boreal Eurasia region for which limited land-cover information is available at regional level.

The main aim of this paper is to demonstrate that a coarse-resolution land-cover map of the Russian Federation, the ‘TerraNorte’ map at 230 m × 230 m resolution for the year 2010, can be used in combination with a sample of reference forest maps at 30 m resolution to correctly assess forest cover in the Russian federation.

First, an accuracy assessment of the TerraNorte map is carried out through the use of reference forest maps derived from finer-resolution satellite imagery (Landsat Thematic Mapper (TM) sensor). A sample of 32 sites was selected for the detailed identification of forest cover from Landsat TM imagery. A methodological approach is developed to process and analyse the Landsat imagery based on unsupervised classification and cluster-based visual labelling. The resulting forest maps over the 32 sites are then used to evaluate the accuracy of the forest classes of the TerraNorte land-cover map. A regression analysis shows that the TerraNorte map produces satisfactory results for areas south of 65° N, whereas several forest classes in more northern areas have lower accuracy. This might be explained by the strong reflectance of background (i.e. non-tree) cover.

A forest area estimate is then derived by calibration of the TerraNorte Russian map using a sample of Landsat-derived reference maps (using a regression estimator approach). This estimate compares very well with the FAO FRA exercise for 2010 (1% difference for total forested area). We conclude that the TerraNorte map combined with finer-resolution reference maps can be used as a reliable spatial information layer for forest resources assessment over the Russian Federation at national scale.     

Experiments with an adaptive Bayesian restoration method

This paper describes a Bayesian restoration method applied to two-dimensional measured images, whose detector response function is not completely known. The response function is assumed Gaussian with standard deviation depending on the estimate of the local density of the image. The convex hull of the K-nearest neighbours (KNN) of each ‘on’ pixel is used to compute the local density. The method has been tested on ‘sparse’ images, with and without noise background.     

Modelling deforestation using GIS and artificial neural networks

This study aims to predict the spatial distribution of tropical deforestation. Landsat images dated 1974, 1986 and 1991 were classified in order to generate digital deforestation maps which locate deforestation and forest persistence areas. The deforestation maps were overlaid with various spatial variables such as the proximity to roads and to settlements, forest fragmentation, elevation, slope and soil type to determine the relationship between deforestation and these explanatory variables. A multi-layer perceptron was trained in order to estimate the propensity to deforestation as a function of the explanatory variables and was used to develop deforestation risk assessment maps. The comparison of risk assessment map and actual deforestation indicates that the model was able to classify correctly 69% of the grid cells, for two categories: forest persistence versus deforestation. Artificial neural networks approach was found to have a great potential to predict land cover changes because it permits to develop complex, non-linear models.     

Improving local forest volume estimates by fusion of multi-temporal forest type maps

An alternative method for estimating standing wood volume based on the fusion of multi-temporal forest type maps and single reduced associated ground-based inventories is proposed. With the integration of photo-interpreted forest map realizations from different years into a single “Fused Map” resulting in an improved local estimate of the forest type, the proposed method offers more accurate estimates than the approach traditionally used by the Quebec Ministry of Natural Resources, even with reduced ground-based inventory effort. A Fused Map is basically an adapted mean of local forest type that accounts for differences among the classification systems used for each map, temporal differences between maps, and subjectivity associated with photo-interpreted data.     

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.     

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.