Robert G. Knox

Estimation of tropical forest structural characteristics using large-footprint lidar

Quantification of forest structure is important for developing a better understanding of how forest ecosystems function. Additionally, estimation of forest structural attributes, such as aboveground biomass (AGBM), is an important step in identifying the amount of carbon in terrestrial vegetation pools and is central to global carbon cycle studies. Although current remote sensing techniques recover such tropical forest structure poorly, new large-footprint lidar instruments show great promise. As part of a prelaunch validation plan for the Vegetation Canopy Lidar (VCL) mission, the Laser Vegetation Imaging Sensor (LVIS), a large-footprint airborne scanning lidar, was flown over the La Selva Biological Station, a tropical wet forest site in Costa Rica. The primary objective of this study was to test the ability of large-footprint lidar instruments to recover forest structural characteristics across a spectrum of land cover types from pasture to secondary and primary tropical forests. LVIS metrics were able to predict field-derived quadratic mean stem diameter (QMSD), basal area, and AGBM with R 2 values of up to .93, .72, and .93, respectively. These relationships were significant and nonasymptotic through the entire range of conditions sampled at the La Selva. Our results confirm the ability of large-footprint lidar instruments to estimate important structural attributes, including biomass in dense tropical forests, and when taken along with similar results from studies in temperate forests, strongly validate the VCL mission framework. D 2002 Elsevier Science Inc. All rights reserved.

Sensitivity of large-footprint lidar to canopy structure and biomass in a neotropical rainforest

Accurate estimates of the total biomass in terrestrial vegetation are important for carbon dynamics studies at a variety of scales. Although aboveground biomass is difficult to quantify over large areas using traditional techniques, lidar remote sensing holds great promise for biomass estimation because it directly measures components of canopy structure such as canopy height and the vertical distribution of intercepted canopy surfaces. In this study, our primary goal was to explore the sensitivity of lidar to differences in canopy structure and aboveground biomass in a dense, neotropical rainforest. We first examined the relationship between simple vertical canopy profiles derived from field measurements and the estimated aboveground biomass (EAGB) across a range of field plots located in primary and secondary tropical rainforest and in agroforesty areas. We found that metrics from field-derived vertical canopy profiles are highly correlated (R 2 up to .94) with EAGB across the entire range of conditions sampled. Next, we found that vertical canopy profiles from a large-footprint lidar instrument were closely related with coincident field profiles, and that metrics from both field and lidar profiles are highly correlated. As a result, metrics from lidar profiles are also highly correlated (R 2 up to .94) with EAGB across this neotropical landscape. These results help to explain the nature of the relationship between lidar data and EAGB, and also lay the foundation to explore the generality of the relationship between vertical canopy profiles and biomass in other tropical regions. D 2002 Elsevier Science Inc. All rights reserved.

Volumetric lidar return patterns from an old-growth tropical rainforest canopy

Rainforests represent the epitome of structural complexity in terrestrial ecosystems. However, measures of three-dimensional canopy structure are limited to a few areas typically < 1 ha with construction crane or walkway/platform access. An innovative laser profiling system, the Laser Vegetation Imaging Sensor (LVIS), was used to map canopy structure (i.e. based on height and vertical distribution of laser returns) of a tropical rainforest in Costa Rica. Within a 1km2 area of mature rainforest, canopy top height ranged from 8.4 to 51.6m based on the altimeter measures. The laser return density was most concentrated in the horizontal layer located 20-30m above the ground. Spatial patterns of the return were found to be isotropic based on north-south versus east-west vertical return profiles and exhibited properties of self-similarity.

Forest biomass from combined ecosystem and radar backscatter modeling

Abstract Above-ground woody biomass is an important parameter for describing the function and productivity of forested ecosystems. Recent studies have demonstrated that synthetic aperture radar (SAR) can be used to estimate above-ground standing biomass. To date, these studies have relied on extensive ground-truth measurements to construct relationships between biomass and SAR backscatter. In this article we discuss the use of models to help develop a relationship between biomass and radar backscatter and compare the predictions with measurements. A gap-type forest succession model was used to simulate growth and development of a northern hardwood-boreal transitional forest typical of central Maine, USA. Model results of species, and bole diameter at breast height (dbh) of individual trees in a 900 m 2 stand were used to run discontinuous canopy backscatter models to determine radar backscatter coefficients for a wide range of simulated forest stands. Using model results, relationships of copolarized backscatter to forest biomass were developed and applied to airborne SAR (AIRSAR) image over a forested area in Maine. A relationship derived totally from model results was found to underestimate biomass. Calibrating the modeled backscatter with limited AIRSAR backscatter measurements improved the biomass estimation when compared to field measurements. The approach of using a combination of forest succession and remote sensing models to develop algorithms for inferring forest attributes produced comparable results with techniques using only measurements. Applying the model derived algorithm to SAR imagery produced reasonable results when mapped biomass was limited to 15 kg/m 2 or less.

Relationships between soil properties and vegetation at the Northern Experimental Forest, Howland, Maine

Abstract This research relates the results of a survey and detailed analysis of soils in a northern mixed conifer forest to vegetation characteristics as represented by remotely sensed data. The work was conducted at International Papers Northern Experimental Forest (NEF) at Howland, Maine as part of NASAs Forest Ecosystem Dynamics (FED) project. An intensive soil survey was performed and relationships between soil properties (i.e., drainage class, depth of active zone, water holding capacity, carbon / nitrogen ratio, pH, and sum of bases), species composition, and normalized difference vegetation index (NDVI) from the Advanced Visible and Infrared Imaging Spectrometer (AVIRIS) were derived. Results showed that there was great variability in soil properties across the landscape due to complex regional glacial activity and recent alluvial events. Significant statistical differences were observed in species composition and NDVI between soil mapping units and with soil drainage class. However, other specific soil properties could not be used to explain these differences given the number of soil samples characterized, or without taking disturbance and management history into account. Simulation modeling, which would include soil data and stand history information as inputs, would provide an additional means of interpreting the relationship between remotely sensed imagery, inferred ecosystem properties, and complex, landscape-level patterns of soil characteristics.

NASA Mission to Measure Global Plant Physiology and Functional Types

A NASA Earth mission concept has been developed that focuses on a set of science objectives related to the measurement of plant physiology and functional type for terrestrial and aquatic ecosystems. The NRC Decadal Survey specifically calls for the HyspIRI mission to measure terrestrial and aquatic ecosystems. A review of the literature in conjunction with analysis of ongoing ecosystem research established imaging spectroscopy in the solar reflected portion of the spectrum as the appropriate approach to address these objectives. For these topics a detailed requirement analysis was performed that specified the measurement objectives, measurement requirements, instrument requirement and other requirements. These were distilled into a single set of spectral, radiometric, spatial, uniformity and temporal requirement. Key among these are: spectral coverage from 380 to 2500 nm at 10 nm sampling, radiometric resolution and precision giving high signal-to-noise ratios for dark aquatic targets, spatial sampling of 60 m, spectral and spectral IFOV uniformity > 95%, and temporal coverage with a 19 day repeat at the equator.

Northern Forest Ecosystem Dynamics Using Coupled Models and Remote Sensing

Abstract Forest ecosystem dynamics modeling, remote sensing data analysis, and a geographical information system (GIS) were used together to determine the possible growth and development of a northern forest in Maine, USA. Field measurements and airborne synthetic aperture radar (SAR) data were used to produce maps of forest cover type and above ground biomass. These forest attribute maps derived with remote sensing data, along with a conventional soils map, were used to identify the initial conditions for forest ecosystem model simulations. Using this information along with ecosystem model results enabled the development of predictive maps of forest development. The results obtained were consistent with observed forest conditions and expected successional trajectories. The study demonstrated that ecosystem models might be used in a spatial context when parametrized and used with georeferenced data sets.

Tornado and fire effects on tree species composition in a savanna in the Big Thicket National Preserve, southeast Texas, USA

Abstract Ordination showed that species composition in a savanna shifted toward mixed pine-hardwood types after the tornado damage in 1983. Of the twenty 250 m2 study plots, one baygall plot remained unchanged in its position in ordination space, as did five mixed pine-hardwood plots and six savanna plots. However, eight savanna plots moved from the savanna space to the mixed pine-hardwood space after the tornado. Prescribed burns in 1986 and 1991 had a modest effect in reversing this trend: only four of the eight plots returned to the savanna ordination space by 1991. The changes in species composition, summarized by movement of plots in ordination space, reflected the conflicting affects of the tornado and fire: the tornado tended to change savanna to mixed pine-hardwood by differentially removing pines and stimulating hardwood growth. Fires tended to reduce hardwood density. However, stand opening and increased fuel loads following the tornado did not result in fires intense enough to dramatically enhance savanna recovery.

Spectral ratio biospheric lidar

A new active vegetation index measurement technique has been developed and demonstrated using low power laser diodes to make horizontal-path lidar measurements of nearby deciduous foliage. The two wavelength laser transmitter operates within and adjacent to the 680 nm absorption feature exhibited by all chlorophyll containing vegetation. Measurements from early October through late November 2003 are presented and the results are discussed

Effects of prescribed fire on the composition of woody plant communities in southeastern Texas

. The response to prescribed burning of plant communities ranging from dry to wet habitats was monitored using permanent plots sampled from 1989 to 1993. Temporal controls for fire effects were provided by matched sets of plots protected from fire by newly constructed fire breaks. Changes in species composition were studied by ordination of strata of trees (> 5 cm DBH), small trees (2–5 cm DBH), large saplings (1–2 cm DBH), and small saplings and seedlings (50–140 cm tall). Results show that changes occurred largely in the small tree stratum, in which xeric species increased in importance. Although there were changes in sapling and seedling strata, no clear direction of change was recognized. Fire had little effect on the tree stratum. Of the seven community types under study, three types, sandhill, upland pine, and upperslope pine-oak, were most strongly affected, as indicated by post-fire change in positions of samples representing these communities in ordination space and reduction in understory species abundance. Samples representing the other four mesic and wet communities showed little or no change in their positions. These short-term results indicate that changes in vegetation resulting from fire were small and were mostly restricted to the dry types in which possible compositional change is expected to occur. This differential effect of fire suggests that the influence of fire is secondary to that of topographical and soil gradients in determining vegetation pattern under current fire regimes. Fire seems to reinforce an overall vegetation gradient controlled by soil in southeastern Texas.