Michael L. Benson

Forest structure estimation using SAR, LiDAR, and optical data in the Canadian Boreal forest

One of the most fundamental new technical challenges of a DESDynI space-borne mission is the fusion of the several sensor modalities - LiDAR, SAR, InSAR, and Optical - in order to accurately estimate desired 3D vegetation structures and biomass parameters in areas where the sensors overlap, and to extrapolate them over continuous areas where lidar data is absent. The objective of this paper is to use measured datasets in conjunction with our sensor forward models to develop and validate an estimation algorithm that fuses various remote sensing technologies with a minimal amount of ground information and yields an accurate estimate of forest structure, including biomass, canopy height, and tree species.

Extrapolation of LiDAR for forest structure estimation using SAR, InSAR, and optical data

One of the most fundamental new technical challenges of a DES-DynI spaceborne mission is the fusion of the several sensor modalities - LiDAR, SAR, InSAR, and Optical - in order to accurately estimate desired 3D Vegetation structures and biomass parameters at their point of intersection and to extrapolate them over continuous areas.

Variable wind influence on InSAR imagery of forests

The horizontal and vertical (3D) structure of Earths forested ecosystems are of great significance to their ecological functioning and societal uses. An InSAR approach is one methodology whereby a forests structure and height in particular can be successfully estimated. Critical to the successful estimation is a high correlation between multiple SAR images. Regardless of a forests location on the Earth, wind can significantly alter a forests appearance to an L-band SAR system and so decrease this necessary correlation. In order to investigate the wind-induced decorrelation, we have developed a model for the repeat-pass interferometric SAR response of a forested area taking into account wind effects. The simulation consists of multiple interconnected parts including static tree geometrys generation, a wind simulator to apply to a static tree, and an electromagnetic model to allow us to calculate the interferometric SAR response. The static tree geometry generation process generates a pseudo-random tree based on a given DNA file which specifies a species specific structure. This geometry is then modified by the wind simulator producing snapshots of tree-geometry as a function of time. Each snapshot is then used in the interferometric SAR simulator to synthesize the wind-blown geometrys InSAR response. Results present coherence as a function of wind speed and forest structure. An important feature of this research is the usage of a physically based realistic wind model that is based on measurements of wind effects on trees as well as realistic models of fluid flow and simple harmonic branch resonators. Allowing branches to bend and move out of the plane of the incident wind field enables our model to capture numerous features of a physical tree blowing in the wind. This realistic model is necessary for a realistic simulation of the effects that wind has on a given InSAR imaging system.

Quantifying the results of wind and rain on ifsar tree height estimation

The horizontal and vertical (3D) structure of Earths forested ecosystems are of great significance to their ecological functioning and societal uses. An IfSAR approach is one methodology whereby a forests structure and height in particular can be successfully estimated. Critical to the successful estimation is a high correlation between multiple SAR images. Regardless of a forests location on the Earth, wind and precipitation can significantly alter a forests appearance to a SAR system operating in either the L or C bands and so too decrease this necessary correlation. In order to investigate and quantize the decorrelation induced by factors such as wind and rain, we have developed a model for the repeat-pass interferometric SAR response of a forest including the application of a wind field and / or a rain storm. The simulation consists of multiple interconnected parts including the generation of fractal tree geometries, a wind simulator to apply apply variable wind forces to the generated trees, an electromagnetic model to allow us to calculate a Single Look Complex value for the SAR return of the combined target, an image forming technique based on antenna array theory, and an image processing algorithm. Results present polarmetric coherence as a function of platform look angle, wind speed, and moisture content. An important feature of this research is the usage of a physically based realistic wind model that is based on measurements of wind effects on trees as well as realistic models of fluid flow and simple harmonic branch segment resonators. Allowing branches to bend and move out of the plane of the incident wind field enables our model to capture numerous features of a physical tree blowing in the wind. This realistic model is necessary for a realistic simulation of the effects that wind has on a given InSAR imaging system as expressed in this study by the interferometric coherence.

Estimating the three dimensional structure of the harvard forest using a database driven multi-modal remote sensing technique

The global forest covers over 30% of the Earths landmass and plays a critical role in a number of global systems including the carbon cycle. Developing methods to track the carbon flow into and out of forests is necessary to gain a complete understanding of the global carbon cycle and, in turn, its effect on the climate. Remote sensing technologies such as satellite based passive optical remote sensing and synthetic aperture radar are uniquely capable of interrogating forests. Using data fusion or synergy, we present a novel approach to estimating forest aboveground biomass and mean canopy height in heterogeneous forest regions with minimal required ancillary ground measurements. We present a dynamic database driven approach wherein a region of study is divided into stands and each stand is compared to a set of simulated forest stands. The simulated stands each contain a set of simulated fractal trees with distributions based on observed ranges within the area of study. Each stand within the region of study is examined and compared to the simulated forest stands using a measure of similarity. If a simulated stand is not found to be similar to the measured stand, an iterative process is employed wherein the most similar simulated stand is dynamically modified including its species composition and the mean canopy height and above-ground biomass within each species until a similar simulated stand is constructed. The simulated stand, regardless of if it existed previously in the set of simulated stands or if it needed to be dynamically generated, is considered a reasonable representation for the measured stand under test and its height and biomass are reported. This approach relies heavily on our sensor simulators, including our fractal-based tree geometry generator, as well as SAR, IfSAR, LiDAR, and Optical simulators. We propose to validate our approach in the Harvard Forest, a heavily studied region in central Massachusetts.

Model-based estimation of large area forest canopy height and biomass using radar and optical remote sensing with limited lidar data

Data synergy or fusion is a mechanism whereby discrete types of data are used together to achieve a better understanding than was possible with each individually. Spanning over 30% of the Earths landmass, the global forest plays a role in numerous planetary systems including the carbon cycle. The objective of this study is to couple simulated forest stands with measured datasets from various instruments to estimate a forests mean canopy height and aboveground dry-biomass in large regions spanning many square kilometers. We present a method to combine measured datasets with our sensor models to develop a feature estimation algorithm that fuses multi-modal remote sensing technologies with a minimal amount of ground information and yields an accurate estimate of forest structure including dry biomass and canopy height in a region spanning over 60 km2.

Estimating boreal forest canopy height and above ground biomass using multi-modal remote sensing; a database driven approach

Data synergy or fusion is a mechanism whereby discrete types of data are used together to achieve a better understanding than was possible with each individually. While the DESDynI missions have been reduced and renamed, their original goal to fuse several sensor modalities to achieve an understanding of the global carbon cycle is still valid. Spanning over 30% of the Earths landmass, the global forest plays a significant role in numerous planetary systems; the carbon cycle included. The objective of this paper is to couple simulated forest stands with measured datasets from various instruments to estimate a forests mean canopy height and aboveground dry-biomass. We use existing datasets to develop and validate our fusion and extrapolation approach, which involves using our four sensor simulators, including our fractal-based tree geometry generator, in tandem with our in-house parameter estimation software which performs fusion and retrieval functions. We then use existing field and radarlidar-VNIR data for the Boreas southern study area to validate our simulators in this region and construct a large set of boreal trees for use in our fusion and extrapolation processes.

Exploring the Canadian boreal forest using airsar, LandSAT5, and virtual lidar

Data synergy or fusion is a mechanism whereby discrete types of data are used together to achieve a better understanding than was possible with each individually. Spanning over 30% of the Earths landmass, the global forest plays a significant role in numerous planetary systems; the carbon cycle included. The objective of this paper is to couple simulated forest stands with measured datasets from various instruments to estimate a forests mean canopy height and aboveground dry-biomass in regions where lidar measurements are sparse; we propose to create a virtual lidar instrument based on other readily available sensor measurements. We present a method to combine measured datasets with our sensor models to develop a classification algorithm that fuses multi-modal remote sensing technologies with a minimal amount of ground information and yields an accurate estimate of forest structure including dry biomass and canopy height. We show the performance of our proposed method in regions lidar measurements as well as in regions lacking these measurements. Finally, we present our method using virtual lidar and show that there is minimal degradation in our estimation.

Estimating the three dimensional structure of heterogeneous forests using multi-modal remote sensing and sensor extrapolation techniques

Data synergy or fusion is a mechanism whereby discrete types of data are used together to achieve a better understanding than was possible with each individually. Spanning over 30% of the Earths landmass, the global forest plays a significant role in numerous planetary systems including the carbon cycle. This paper presents a novel approach to estimating forest aboveground biomass and mean canopy height with minimal required ancillary ground measurements. We present a dynamic database driven model wherein a simulated forest is generated on the fly and is iteratively modified to find not only the canopy height and biomass, but also to approximate the stands species composition. This approach relies heavily on our sensor simulators, including our fractal-based tree geometry generator, as well as SAR, IfSAR, LiDAR, and Optical simulators. We propose to validate our approach in the Harvard Forest, a heavily studied region in central Massachusetts.

Estimating the ground heightwith L-band IfSAR in a wind-blown forest environment

The horizontal and vertical (3D) structure of Earths forested ecosystems are of great significance to their ecological functioning and societal uses. An IfSAR approach is one methodology whereby a forests structure and height in particular can be successfully estimated. Critical to the successful estimation is a high correlation between multiple SAR images. Regardless of a forests location on the Earth, motion due to wind can significantly alter a forests appearance to a radar system operating at L-band and so too decrease this necessary correlation. In order to investigate and quantize the decorrelation induced by the wind, we have developed a model that is capable of generating both a single-pass and repeat-pass interferometric SAR response of a forest including the application of a randomly oriented wind field. The simulation consists of multiple interconnected parts including the generation of fractal tree geometries, a wind simulator to apply variable wind forces to the generated trees, an electromagnetic model to allow us to calculate a fully polarimetric Single Look Complex value for the SAR return of the combined target, and an IfSAR processing algorithm capable of calculating a scattering phase center in the presence of wind. Results present polarimetric scattering phase centers and interferogram coherence as a function of wind speed. We further deconstruct our SAR processor to yield individual scatting mechanisms and their resultant scattering phase centers. This deconstruction allows for an accurate estimate of the underlying ground to be generated, even in the presence of a strong wind field.