Kathleen M. Bergen

Estimation of forest biophysical characteristics in Northern Michigan with SIR-C/X-SAR

A three-step process is presented for estimation of forest biophysical properties from orbital polarimetric SAR data. Simple direct retrieval of total aboveground biomass is shown to be ill-posed unless the effects of forest structure are explicitly taken into account. The process first involves classification by (1) using SAR data to classify terrain on the basis of structural categories or (2) a priori classification of vegetation type on some other basis. Next, polarimetric SAR data at L- and C-bands are used to estimate basal area, height and dry crown biomass for forested areas. The estimation algorithms are empirically determined and are specific to each structural class. The last step uses a simple biophysical model to combine the estimates of basal area and height with ancillary information on trunk taper factor and wood density to estimate trunk biomass. Total biomass is estimated as the sum of crown and trunk biomass. The methodology is tested using SIR-C data obtained from the Raco Supersite in Northern Michigan on Apr. 15, 1994. This site is located at the ecotone between the boreal forest and northern temperate forests, and includes forest communities common to both. The results show that for the forest communities examined, biophysical attributes can be estimated with relatively small rms errors: (1) height (0-23 m) with rms error of 2.4 m, (2) basal area (0-72 m/sup 2//ha) with rms error of 3.5 m/sup 2//ha, (3) dry trunk biomass (0-19 kg/m/sup 2/) with rms error of 1.1 kg/m/sup 2/, (4) dry crown biomass (0-6 kg/m/sup 2/) with rms error of 0.5 kg/m/sup 2/, and (5) total aboveground biomass (0-25 kg/m/sup 2/) with rms error of 1.4 kg/m/sup 2/. The addition of X-SAR data to SIR-C was found to yield substantial further improvement in estimates of crown biomass in particular. However, due to a small sample size resulting from antenna misalignment between SIR-C and X-SAR, the statistical significance of this improvement cannot be reliably established until further data are analyzed. Finally, the results reported are for a small subset of the data acquired by SIR-C/X-SAR. >

Multitemporal Land-Cover Classification Using SIR-C/X-SAR Imagery

Abstract The dual-flight program (April and October) for the SIR-C/X-SAR instrument aboard the shuttle Endeavor was designed expressly to acquire Synthetic Aperture Radar (SAR) imagery at two significantly different seasons. At the Michigan Forests Test Site (MFTS), the April mission occurred at the beginning of the spring thaw and the October mission occurred just prior to and during the fall color change. Four scenes are evaluated at a constant incidence angle. Seven features are extracted from the SAR data for potential use in classification using powers at different frequencies and polarizations. Given multiseason SIR-C/X-SAR imagery, there are three possible approaches in the classifier development: 1) Under the assumption that the scene does not change significantly as a function of time, develop one classification for a set of x scenes using n features, with x times the number of samples per feature; 2) ignore the multiseason availability and develop independent classifications for each scene using n features; 3) develop a true multitemporal classification where N of features equals n (number of features) times x (number of scenes). Each of these is applied using a combination knowledge-based and Bayesian classifier. Level II (roughly forest community) results show that the true multitemporal April/October classification works very well (97%), as do those for the individual scenes (>90%). A pooled classifier works poorly (April=90%, October=77%) and shows that temporal changes in phenology and moisture conditions contribute significant noise in terrain classification.

Inundation extent and flood frequency mapping using LANDSAT imagery and digital elevation models.

We modeled the extent of inundation around Poyang Lake, China using 13 Landsat images and two digital elevation models (DEMs). Boundaries of the observed inundation extents were (a) labeled with lake-level measurements taken at a representative hydrological station and (b) interpolated to create a Water Line DEM (WL-DEM) that was used to map inundation frequency. A 30 m contour-based DEM produced slightly better results than the Shuttle Radar Topography Mission DEM, but neither DEM was accurate for medium and low lake levels. The WL-DEM exhibited improved accuracy at medium lake levels, but had relatively high errors at low lake levels.

Satellite imagery gives clear picture of Russia's boreal forest fires

Boreal forest fires in Russia in 1998 may have covered a much larger area than originally reported and may have been mostly crown fires, releasing more greenhouse gases into the atmosphere than surface fires. These conclusions are based on analysis of Advanced Very High Resolution Radiometer (AVHRR) data and discussions with Russian fire officials. A more complete analysis of AVHRR data collected since 1980 is underway to determine overall fire patterns in Russias boreal forests. The year 1998 was an extreme fire year throughout the boreal forest regions of both North America and Eurasia. Over 11 million hectares burned. In Alaska and Canada alone, 4.7 million hectares of boreal forests burned in 1998, according to estimates by national fire monitoring systems in those places.

Integration of remotely sensed radar imagery in modeling and mapping of forest biomass and net primary production

Abstract New remote sensing programs provide the opportunity to optimize the connection of remotely sensed data with key parameters in measuring and modeling net primary production (NPP). Synthetic aperture radars (SARs) are discussed in terms of their ability to measure more directly certain parameters related to NPP. The purpose of this paper is to introduce SAR-based methodologies and results for (1) deriving parameters which may be considered input datasets for NPP models and (2) the subsequent application of an aboveground annual NPP (ANNP) model for these datasets. Derivations are land cover and biophysical parameters including forest height, aboveground forest tree biomass (and carbon fraction), and belowground coarse root biomass (and carbon fraction). An allometric ANPP model is applied to demonstrate the applicability of these SAR-derived datasets to NPP modeling. Results are regional quantifications and mapped distributions of forest height, above and belowground tree biomass (and carbon fraction), aboveground ANPP, and the relationship of forest stage to production.

Increasing Gross Primary Production (GPP) in the Urbanizing Landscapes of Southeastern Michigan

In order to understand the impact of urbanizing landscapes on regional gross primary production (GPP), we analyzed changes in land-cover and annual GPP over an urban-rural gradient in ten Southeastern Michigan counties between 1991 and 1999. Landsat and AVHRR remote sensing data and biophysical parameters corresponding to three major landcover types (i.e., built-up, tree, and crop/grass) were used to estimate the annual GPP synthesized during the growing season of 1991 and 1999. According to the numbers of households reported by the U.S. Census in 1990 and 2000, the area settled at urban (� 1 housing unit acre � 1 ), suburban (0.1 to 1 housing units acre � 1 ), and exurban (0.025 to 0.1 housing units acre � 1 ) densities expanded, while the area settled at rural (� 0.025 housing units acre � 1 ) densities reduced. GPP in this urbanizing area, however, was found to increase from 1991 to 1999. Increasing annual GPP was attributed mainly to a region-wide increase in tree cover in 1999. In addition, the estimated annual GPP and its changes between 1991 and 1999 were found to be spatially heterogeneous. The exurban category (including constantly exurban and exurban converted from rural) was associated with the highest annual GPP as well as an intensified increase in GPP. Our study indicates that lowdensity exurban development, characterized by large proportions of vegetation, can be more productive in the form of GPP than the agricultural land it replaces. Therefore, low-density development of agricultural areas in U.S. Midwest, comprising significant fractions of highly productive tree and grass species, may not degrade, but enhance, the regional CO2 uptake from the atmosphere.

Multi-dimensional vegetation structure in modeling avian habitat

Abstract The goal of this study was to evaluate the contributions of forest and landscape structure derived from remote sensing instruments to habitat mapping. Our empirical data focused at the landscape scale on a test site in northern Michigan, using radar and Landsat imagery and bird-presence data by species. We tested the contributions of multi-dimensional forest and landscape structure variables using GARP (Genetic Algorithm for Rule-Set Production), a representative modeling methodology used in biodiversity informatics. For our multi-dimensional variables, radar data were processed to derive forest biomass maps and these data were used with a Landsat-derived vegetation type classification and spatial neighborhood analyses. We collected field data on bird species presence and habitat for northern forest birds known to have a range of vegetation habitat requirements. We modeled and tested the relationships between bird presence and 1) vegetation type, 2) vegetation type and spatial neighborhood descriptions, 3) vegetation type and biomass, and 4) all variables together, using GARP, for three bird species. Modeled results showed that inclusion of biomass or neighborhoods improved the accuracy of bird habitat prediction over vegetation type alone, and that the inclusion of neighborhoods and biomass together generally produced the greatest improvement. The maps and model rules resulting from the multiple factor models were interpreted to be more precise depictions of a particular species habitat when compared with the models that used vegetation type only. We suggest that for bird species whose niche requirements include forest and landscape structure, inclusion of multi-dimensional information may be advantageous in habitat modeling at the landscape level. Further research should focus on testing additional variables and species, on further integration of newer radar and lidar remote sensing capabilities with multi-spectral sensors for quantifying forest and landscape multi-dimensional structure, and incorporating these in biodiversity informatics modeling.

Land-cover change and vulnerability to flooding near Poyang Lake, Jiangxi Province, China

Inhabitants near Poyang Lake, in the Central Yangtze River Basin, China, are vulnerable to loss of life and livelihood because of the interactions of flooding and land-use policies and decisions. We analyzed implications of land-cover patterns for vulnerability to flooding in the Poyang Lake Region. Land-cover and change were mapped using multi

Uncertainty of Forest Biomass Estimates in North Temperate Forests Due to Allometry: Implications for Remote Sensing

Estimates of above ground biomass density in forests are crucial for refining global climate models and understanding climate change. Although data from field studies can be aggregated to estimate carbon stocks on global scales, the sparsity of such field data, temporal heterogeneity and methodological variations introduce large errors. Remote sensing measurements from spaceborne sensors are a realistic alternative for global carbon accounting; however, the uncertainty of such measurements is not well known and remains an active area of research. This article describes an effort to collect field data at the Harvard and Howland Forest sites, set in the temperate forests of the Northeastern United States in an attempt to establish ground truth forest biomass for calibration of remote sensing measurements. We present an assessment of the quality of ground truth biomass estimates derived from three different sets of diameter-based allometric equations over the Harvard and Howland Forests to establish the contribution of errors in ground truth data to the error in biomass estimates from remote sensing measurements.

Characterizing Carbon in a Northern Forest by Using SIR-C/X-SAR Imagery

Abstract A significant large-scale question in ecology and earth systems science pertains to the amount of carbon (C) stored in terrestrial vegetation. In this paper, a synthetic aperture radar (SAR)–based methodology is developed and evaluated for quantification of several key vegetative C components—both natural and human induced—of the northern forest. Specifically, the methodology provides estimates of C stored in living forest vegetation, above-ground C gain from annual growth (aboveground net primary productivity, or ANPP), and C removal due to managed forest disturbance in the form of clear-cutting. The inputs are shuttle imaging radar (SIR)-C/X-SAR- derived terrain classifications, SIR-C/X-SAR-derived biomass estimation, and allometric relations and equations developed for the northern hardwood and conifer forest in general and from local test stand data. Results are mapped quantitatively in the image domain for above-ground C storage, below-ground C storage, above- to below-ground ratios, total C in living woody vegetation, and forest absolute and relative ANPP rates. Numeric estimates also are extracted from each of these in tabular form; for example, results show that the forested parts of the sampled area contain 2.73×10 9 kg of aboveground C and 4.86×10 8 kg of below-ground C in 51,448 ha. When combined with a SAR-derived classification, similar quantifications can be extracted for each of the several forest communities present in the region. Estimates of forest ANPP show that it ranges from 0.5 kg/m 2 /yr of biomass to 2.7 kg/m 2 /yr in the test site, with an average of 1.09 kg/m 2 /yr. Estimation of C removal due to clear-cutting is done by using multidate classifications of SAR imagery and a procedure including image differencing and decision rules. Clear-cuts that were made between SIR-C/X-SAR Shuttle Radar Laboratory (SRL)-1 (April) and SRL-2 (October) are identified by differencing the classified imagery. When combined with the SAR-derived biomass image, results show that an estimated 300 ha of forest with 6.02×10 6 kg of C were removed in this 6-month period.