Forest variable estimation from fusion of SAR and multispectral optical data (original) (raw)
Related papers
International Journal of Applied Earth Observation and Geoinformation
Spatially-explicit information on forest structure is paramount to estimating aboveground carbon stocks for designing sustainable forest management strategies and mitigating greenhouse gas emissions from deforestation and forest degradation. LiDAR measurements provide samples of forest structure that must be integrated with satellite imagery to predict and to map landscape scale variations of forest structure. Here we evaluate the capability of existing satellite synthetic aperture radar (SAR) with multispectral data to estimate forest canopy height over five study sites across two biomes in North America, namely temperate broadleaf and mixed forests and temperate coniferous forests. Pixel size affected the modelling results, with an improvement in model performance as pixel resolution coarsened from 25 m to 100 m. Likewise, the sample size was an important factor in the uncertainty of height prediction using the Support Vector Machine modelling approach. Larger sample size yielded better results but the improvement stabilised when the sample size reached approximately 10% of the study area. We also evaluated the impact of surface moisture (soil and vegetation moisture) on the modelling approach. Whereas the impact of surface moisture had a moderate effect on the proportion of the variance explained by the model (up to 14%), its impact was more evident in the bias of the models with bias reaching values up to 4 m. Averaging the incidence angle corrected radar backscatter coefficient (γ°) reduced the impact of surface moisture on the models and improved their performance at all study sites, with R 2 ranging between 0.61 and 0.82, RMSE between 2.02 and 5.64 and bias between 0.02 and −0.06, respectively, at 100 m spatial resolution. An evaluation of the relative importance of the variables in the model performance showed that for the study sites located within the temperate broadleaf and mixed forests biome ALOS-PALSAR HV polarised backscatter was the most important variable, with Landsat Tasselled Cap Transformation components barely contributing to the models for two of the study sites whereas it had a significant contribution at the third one. Over the temperate conifer forests, Landsat Tasselled Cap variables contributed more than the ALOS-PALSAR HV band to predict the landscape height variability. In all cases, incorporation of multispectral data improved the retrieval of forest canopy height and reduced the estimation uncertainty for tall forests. Finally, we concluded that models trained at one study site had higher uncertainty when applied to other sites, but a model developed from multiple sites performed equally to site-specific models to predict forest canopy height. This result suggest that a biome level model developed from several study sites can be used as a reliable estimator of biome-level forest structure from existing satellite imagery.
Combining remotely sensed optical and radar data in kNN-estimation of forest variables
Forest Science, 2003
The use of optical and radar data for estimation of forest variables has been investigated and evaluated by employing the k nearest neighbor (kNN) method. The investigation was performed at a test site located in the south of Sweden consisting mainly of Norway spruce and Scots pine forests with standwise stem volume in the range of 0–430 m3 ha–1. The kNN method imputes weighted reference plot variables to areas to be estimated (target areas), facilitating further use of data in forestry planning models. Remotely sensed multispectral optical data from the SPOT-4 XS satellite and radar data from the airborne CARABAS-II VHF SAR sensor were used, separately and combined, to define weights in the kNN algorithm. The weights were inversely proportional to the image feature distance between the reference plot and the target area. The distance metric was defined using regression models based on the image data sources. Positive impact on the accuracies of stem volume and age estimates was fou...
Tropical Dry Deciduous Forest Stand Variable Estimation Using SAR Data
Optical remote sensing data have been extensively used to derive biophysical properties that relate forest type and composition. However, stand density, stand height and stand volume cannot be estimated directly from optical remote sensing data owing to poor sensitivity between these parameters and reflectance values. The ability of microwave energy to penetrate within forest vegetation makes it possible to extract information on both the crown and trunk components from radar data. The type of polarization employed determines the radar response to the various shapes and orientations of the scattering mechanisms within the canopy or trunk. This study mainly presents experimental results obtained with airborne E-SAR using polarimetric C and L bands over the tropical dry deciduous forest of Chandrapur Forest Division, Maharashtra. A detailed documentation of the relationship between SAR C & L bands backscattering and forest stand variables has been established in the present study through linear correlation. Linear correlation of the single channel SAR derived estimates with the field measured means show a good correlation between L hv backscattering coefficient with stand volume (r 2 = 0.71) and L hh backscattering coefficient with stand density (r 2 = 0.75). The results imply that SAR data has significant potential for stand menstruation in operational forestry.
Forest Attributes from Radar Interferometric Structure and Its Fusion with Optical Remote Sensing
BioScience, 2004
I mportant ecological processes are expressed in and are affected by the vertical structure of forest canopies. Chief among these processes is the carbon cycle, the uncertainties in which limit understanding of climatic trends over the next century. Changes in vegetation structure, induced by climatic conditions, natural disturbance, and human activities, can have substantial impacts on carbon storage and the exchange of carbon dioxide (CO 2), water vapor, and heat with the atmosphere (Law et al. 2001), which can modify climate (Pielke and Avissar 1990). Trees store carbon in their biomass. Fossil fuel burning and deforestation, which release carbon as CO 2 into the atmosphere, are believed to be the two dominant contributing mechanisms to the rise in atmospheric CO 2 over the last 50 years. Forest biomass also has a potential role in reabsorbing some of the excess CO 2 , yet the dynamic responses of carbon fluxes to climate changes are still poorly understood. Knowledge of the level of carbon stored in forest biomass globally is highly uncertain (within about 40% of the true value; Waring and Running 1998), as is the spatiotemporal description of carbon flux associated with changes in above-ground biomass. Improved quantification of biomass will help reduce uncertainty in estimates of magnitudes, rates, and longevity of carbon sequestration by terrestrial ecosystems (Law et al. forthcoming), and therefore will enable better understanding of the global carbon cycle. In this article, we demonstrate quantification of, and magnitudes of uncertainties in, carbon sequestration; we also discuss scaling issues and the need for improving model estimates with data assimilation. Biomass may be most robustly determined from remotely sensed, three-dimensional (3-D) forest structure (Lefsky et al.
Forest Biomass Estimation using Multi-Polarization SAR Data Coupled with Optical Data
Current Science, 2020
This study was carried out to estimate biomass extraction from multi-frequency and multi-polarization of Synthetic Aperture Radar (SAR) data coupled with optical data. Further, the estimated biomass was validated with field-observed data. ALOS-2/PALSAR was utilized for retrieval of forest above-ground biomass (AGB) biophysical parameters. Subsequently, Sentinel-2 optical data and 90 m TanDEM were used to identify the bare ground area for calculating pseudo height. Ground-truth data were utilized for estimation and validation of the modelled biomass from radar data. In this study, five allometric models were used. Multivariate regression models were trained using backscatter from the same acquisition (date) on 10 randomly selected samples from 21 field plots. The validation was carried out on the remaining 11 field plots. Co-validation method was used to validate these models. Biomass was estimated from radar data using regression models. Since the objective of the study was to present generalized biomass estimation models using backscatter information and AGB, the AGB value range 100-400 tonne/ha was estimated/mapped. Combined backscatter and height inputs were better than backscatter models. In the estimation of AGB, polarimetric information content and backscatter information played a significant role.
Coupling SAR data with forest growth models
International Journal of Remote Sensing, 2000
Recent studies have shown the potential of remote sensing data at optical wavelengths to provide spatially referenced input data for process-based forest growth models. In comparison, radar remote sensing remains underexploited, despite having numerous advantages. This letter assesses the potential of radar remote sensing data to estimate tree biomass and various structural features. It is concluded that there exists a wide range of strategies for coupling radar remote sensing with forest growth models.
Forest Variable Estimation Using Radargrammetric Processing of TerraSAR-X Images in Boreal Forests
Remote Sensing, 2014
The last decade has seen launches of radar satellite missions operating in X-band with the sensors acquiring images with spatial resolutions on the order of 1 m. This study uses digital surface models (DSMs) extracted from stereo synthetic aperture radar images and a reference airborne laser scanning digital terrain model to calculate the above-ground biomass and tree height. The resulting values are compared to in situ data. Analyses were undertaken at the Swedish test sites Krycklan (64°N) and Remningstorp (58°N), which have different site conditions. The results showed that, for 459 forest stands in Remningstorp, biomass estimation at the stand level could be performed with 22.9% relative root mean square error, while the height estimation showed 9.4%. Many factors influenced the results and it was found that the topography has a significant effect on the generated DSMs and should therefore be taken into consideration when the stand level mean slope is above four degrees. Different tree species did not have a major effect on the models during leaf-on conditions. Moreover, correct estimation within young forest stands was problematic. The intersection angles resulting in the best results were in the range 8-16°. Based on the results in this study, radargrammetry appears to be a promising potential remote sensing technique for future forest applications.
Prediction of plot-level forest variables using TerraSAR-X stereo SAR data
Remote Sensing of Environment, 2012
Promising results have been obtained in recent years in the use of high-resolution X-band stereo SAR 13 satellite images (with the spatial resolution being in order of meters) in the extraction of elevation 14 data. In the case of forested areas, the extracted elevation values appear to be somewhere between 15 the ground surface and the top of the canopy, depending on the forest characteristics. If the ground 16 surface elevations are known by using a Digital Terrain Model derived from Airborne Laser Scanning 17 surveys, it is possible to obtain information related to forest resources. To the best of our 18 knowledge, this paper, presents the first attempt to obtain forest variables at plot level based on 19
Accuracy of High-Resolution Radar Images in the Estimation of Plot-Level Forest Variables
Lecture Notes in Geoinformation and Cartography, 2009
In the present study, we used the airborne E-SAR radar to simulate the satellite-borne high-resolution TerraSAR radar data and determined the accuracy of the plot-level forest variable estimates produced. Estimation was carried out using the nonparametric k-nearest neighbour (k-nn) method. Variables studied included mean volume, tree species-specific volumes and their proportions of total volume, basal area, mean height and mean diameter. E-SAR-based estimates were compared with those obtained using aerial photographs and medium-resolution satellite image (Landsat ETM+) recording optical wavelength energy. The study area was located in Kirkkonummi, southern Finland. The relative RMSEs for E-SAR were 45%, 29%, 28% and 38% for mean volume, mean diameter, mean height and basal area, respectively. For aerial photographs these were 51%, 26%, 27% and 42%, and for Landsat ETM+ images 58%, 40%, 35% and 49%. Combined datasets outperformed all single-source datasets, with relative RMSEs of 26%, 23%, 33% and 39%. Of the single-source datasets, the E-SAR images were well suited for estimating mean volume, while for mean diameter, mean height and basal area the E-SAR and aerial photographs performed similarly and far better than Landsat ETM+. The aerial photographs succeeded well in the estimation of species-specific volumes and their proportions, but the combined dataset was still significantly better in volume proportions. Due to its good temporal resolution, satellite-borne radar imaging is a promising data source for forest inventories, both in large-area forest inventories and operative forest management planning. Future high-resolution synthetic aperture radar (SAR) images could be combined with airborne laser scanner data when estimating forest or even tree characteristics.
Regression-Based Integrated Bi-sensor SAR Data Model to Estimate Forest Carbon Stock
Journal of the Indian Society of Remote Sensing, 2019
The objective of this study is to estimate the forest aboveground carbon (AGC) stock using integrated space-borne synthetic aperture radar (SAR) data from COSMO-Skymed (X band) and ALOS PALSAR (L band) with field inventory over a tropical deciduous mixed forest. Carbon acts as a vital constituent in the global decision making policy targeting the impact of reducing emissions from deforestation and forest degradation (REDD) and climate change. The study proposed an approach to develop regression models for assessing the forest AGC with synergistic use of SAR bi-sensor X and L band sigma nought data. The best-fit integrated aboveground biomass (AGB) model was validated with additional sample points that produced a model accuracy of 78.6%, adjusted R 2 = 0.88, RMSE = 16.6 Mg/ha, standard error of estimates of 16.03 and Willmott's index of agreement of 0.93. Resulting modeled AGB was converted to AGC using conversion factors. L band resulted in higher accuracy of estimates when compared to X band, while the estimation accuracy enhanced on integrating X-and L-band information. Hence, the study presents an approach using integrated SAR bi-sensor X and L bands that enhance the AGB and AGC estimation accuracy, which can contribute to the operational forestry and policy making related to forest conservation, REDD/REDD? climate change, etc.