Search

Empirical models are important tools for relating field-measured biophysical variables to remote sensing data. Regression analysis has been a popular empirical method of linking these two types of data to provide continuous estimates for variables such as biomass, percent woody canopy cover, and leaf area index (LAI). Traditional methods of regression are not sufficient when resulting biophysical surfaces derived from remote sensing are subsequently used to drive ecosystem process models. Most regression analyses in remote sensing rely on a single spectral vegetation index (SVI) based on red and near-infrared reflectance from a single date of imagery. There are compelling reasons for utilizing greater spectral dimensionality, and for including SVIs from multiple dates in a regression analysis. Moreover, when including multiple SVIs and/or dates, it is useful to integrate these into a single index for regression modeling. Selection of an appropriate regression model, use of multiple SVIs from multiple dates of imagery as predictor variables, and employment of canonical correlation analysis (CCA) to integrate these multiple indices into a single index represent a significant strategic improvement over existing uses of regression analysis in remote sensing. To demonstrate this improved strategy, we compared three different types of regression models to predict LAI for an agro-ecosystem and live tree canopy cover for a needleleaf evergreen boreal forest: traditional (Yon X) ordinary least squares (OLS) regression, inverse (X on Y) OLS regression, and an orthogonal regression method called reduced major axis (RMA). Each model incorporated multiple SVIs from multiple dates and CCA was used to integrate these. For a given dataset, the three regression-modeling approaches produced identical coefficients of determination and intercepts, but different slopes, giving rise to divergent predictive characteristics. The traditional approach yielded the lowest root mean square error (RMSE), but the variance in the predictions was lower than the variance in the observed dataset. The inverse method had the highest RMSE and the variance was inflated relative to the variance of the observed dataset. RMA provided an intermediate set of predictions in terms of the RMSE, and the variance in the observations was preserved in the predictions. These results are predictable from regression theory, but that theory has been essentially ignored within the discipline of remote sensing. D 2002 Elsevier Science Inc. All rights reserved.

Image-to-image registration requires identification of common points in both images (image tie-points; ITPs). Here, we describe software implementing an automated, area-based technique for identifying ITPs. The ITP software was designed to follow two strategies: (1) capitalize on human knowledge and pattern-recognition strengths, and (2) favour robustness in many situations over optimal performance in any one situation. We tested the software under several confounding conditions, representing image distortions, inaccurate user input, and changes between images. The software was robust to additive noise, moderate change between images, low levels of image geometric distortion, undocumented rotation, and inaccurate pixel size designation. At higher levels of image geometric distortion, the software was less robust, but such conditions are often visible to the eye. Methods are discussed for adjusting parameters in the ITP software to reduce error under such conditions. For more than 1600 tests, median time n...

As a signatory to the United Nations Framework Convention on Climate Change, the United States annually prepares an inventory of carbon that has been emitted and sequestered among sectors (e.g., energy, agriculture, and forests). For many years, the United States developed an inventory of forest carbon by comparing contemporary forest inventories to inventories that were collected using different techniques and definitions from more than 20 years ago. Recognizing the need to improve the U.S. forest carbon inventory budget, the United States is adopting the Forest Carbon Accounting Framework, a new approach that removes this older inventory information from the accounting procedures and enables the delineation of forest carbon accumulation by forest growth, land use change, and natural disturbances such as fire. By using the new accounting approach with consistent inventory information, it was found that net land use change is a substantial contributor to the United States forest carbon sink, with the entire forest sink offsetting approximately 15 percent of annual U.S. carbon dioxide emissions from the burning of fossil fuels. The new framework adheres to accounting guidelines set forth by the Intergovernmental Panel on Climate Change while charting a path forward for the incorporation of emerging research, data, and the needs of stakeholders (e.g., reporting at small scales and boreal forest carbon).

Background: Estimation of forest biomass on the regional and global scale is of great importance. Many studies have demonstrated that lidar is an accurate tool for estimating forest aboveground biomass. However, results vary with forest types, terrain conditions and the quality of the lidar data. Methods: In this study, we investigated the utility of low density lidar data (

Light detection and ranging (lidar) data provide accurate measurements of forest canopy structure in the vertical plane; however, current lidar sensors have limited coverage in the horizontal plane. Landsat data provide extensive coverage of generalized forest structural classes in the horizontal plane but are relatively insensitive to variation in forest canopy height. It would, therefore, be desirable to integrate lidar and Landsat data to improve the measurement, mapping, and monitoring of forest structural attributes. We tested five aspatial and spatial methods for predicting canopy height, using an airborne lidar system (Aeroscan) and Landsat Enhanced Thematic Mapper (ETM+) data: regression, kriging, cokriging, and kriging and cokriging of regression residuals. Our 200-km 2 study area in western Oregon encompassed Oregon State University’s McDonald–Dunn Research Forest, which is broadly representative of the age and structural classes common in the region. We sampled a spatially continuous lidar coverage in eight systematic patterns to determine which lidar sampling strategy would optimize lidar– Landsat integration in western Oregon forests: transects sampled at 2000, 1000, 500, and 250 m frequencies, and points sampled at these same spatial frequencies. The aspatial regression model results, regardless of sampling strategy, preserved actual vegetation pattern, but underestimated taller canopies and overestimated shorter canopies. The spatial models, kriging and cokriging, produced less biased results than regression but poorly reproduced vegetation pattern, especially at the sparser (2000 and 1000 m) sampling frequencies. The spatial model predictions were more accurate than the regression model predictions at locations

Estimation of the amount of carbon stored in forests is a key challenge for understanding the global carbon cycle, one which remote sensing is expected to help address. However, estimation of carbon storage in moderate to high biomass forests is difficult for conventional optical and radar sensors. Lidar (light detection and ranging) instruments measure the vertical structure of forests and thus hold great promise for remotely sensing the quantity and spatial organization of forest biomass. In this study, we compare the relationships between lidar-measured canopy structure and coincident field measurements of above-ground biomass at sites in the temperate deciduous, temperate coniferous, and boreal coniferous biomes. A single regression for all three sites is compared with equations derived for each site individually. The single equation explains 84% of variance in above-ground biomass (P < 0.0001) and shows no statistically significant bias in its predictions for any individual site.

Light use efficiency (LUE) algorithms are a potentially effective approach to monitoring global net primary production (NPP) using satellite-borne sensors such as the Moderate Resolution Imaging Spectroradiometer (MODIS). However, these algorithms are applied at relatively coarse spatial resolutions ( � 1 km), which may subsume significant heterogeneity in vegetation LUE (en, gM J � 1 ) and, hence, introduce error. To examine the effects of spatial heterogeneity on a LUE algorithm, imagery from the Advanced Very High Resolution Radiometer (AVHRR) at � 1-km resolution was used to implement a LUE approach for NPP estimation over a 25-km 2 area of corn (Zea mays L.) and soybean (Glycine max Merr.) in central Illinois, USA. Results from several en formulations were compared with a NPP reference surface based on measured NPPs and a high spatial resolution land cover surface derived from Landsat ETM+. Determination of en based on measurements of biomass production and monitoring of absorbed photosynthetically active radiation (APAR) revealed that en of soybean was 68% of that for corn. When a LUE algorithm for estimating NPP was implemented in the study area using the assumption of homogeneous cropland and the en for corn, the estimate for total biomass production was 126% of that from the NPP reference surface. Because of counteracting errors, total biomass production using the soybean en was closer (86%) to that from the NPP reference surface. Retention of high spatial resolution land cover to assign en resulted in a total NPP very similar to the reference NPP because differences in leaf phenology between the crop types were small except early in the growing season. These results suggest several alternative approaches to accounting for land cover heterogeneity in en when implementing LUE algorithms at coarse resolution. D 2002 Elsevier Science Inc. All rights reserved.

In western Oregon, forest ecosystem processes are greatly affected by patterns of stand replacement disturbance. A spatially explicit characterization of clear-cut logging and wildfire is a prerequisite to understanding the causes and consequences of disturbance in this region. We analyzed stand replacement disturbance over 4.6 million forested hectares of three major provinces in western Oregon between 1972 and 1995, contrasting the relative amounts of wildfire and harvest in each province and comparing harvest statistics among the dominant land ownership categories. Clear-cut harvest and wildfire occurred over 19.9% and 0.7% of the study area, respectively. The moist Coast Range Province (CRP) was dominated by private industrial (PI) ownerships and had the greatest concentration of cutting. The relatively dry Klamath Mountains Province (KMP) and the climatically moderate Western Cascades Province (WCP) were dominated by public landowners and had lower concentrations of cutting and larger amounts of wildfire than the CRP. Rates of harvest over time generally followed similar trends across landowners; it was lowest in the early 1970s, peaked in the late 1980s and early 1990s, and then decreased to near 1970s levels by the mid-1990s. PI landowners had harvest rates that were about two and one-half times as high as public owners throughout the study period. Public and private nonindustrial (PNI) owners tended to have relatively small cutting units that remained spatially dispersed throughout the study period, whereas the PI owners had larger individual cutting units that tended toward spatial aggregation over time. Comparing the managed disturbance regimes with historical wild disturbance regimes can help us to understand the relative impact of management regimes on ecosystems.

Nech Sar National Park (NSNP) is one of the most important biodiversity centers in Ethiopia. In recent years, a widespread decline of the terrestrial ecosystems has been reported, yet to date there is no comprehensive assessment on degradation across the park. In this study, changes in landcover were analyzed using 30 m spatial resolution Landsat imagery. Interannual variations of normalized difference vegetation index (NDVI) were examined and compared with climatic variables. The result presented seven landcover classes and five of the seven landcover classes (forest, bush/shrubland, wooded grassland, woodland and grassland) were related to natural vegetation and two landcover types (cultivated land and area under encroaching plants) were direct results of anthropogenic alterations of the landscape. The forest, grassland, and wooded grassland are the most threatened habitat types. A considerable area of the grassland has been replaced by encroaching plants, prominently by Dichrostachys cinerea, Acacia mellifera, A. nilotica, A. oerfota, and A. seyal and is greatly affected by expansion of herbaceous plants, most commonly the species of the family Malvaceae which include Abutilon anglosomaliae, A.bidentatum and A.figarianu. Thus, changes in vegetation of NSNP may be attributed to (i) degradation of existing vegetation through deforestation and (ii) replacement of existing vegetation by encroaching plants. While limited in local meteorological station, NDVI analysis indicated that climate related changes did not have major effects on park vegetation degradation, which suggests anthropogenic impacts as a major driver of observed disturbances.

A multiseasonal Landsat Thematic Mapper (TM) data set consisting of five image dates from a single year was used to characterize agricultural and related land cover in the Willamette River Basin (WRB) of western Oregon. Image registration was accomplished using an automated ground control point selection program. Radiometric normalization was performed using a semiautomated approach based on the identification of no-change pixels in forest, urban, and water classes. Reference data were developed using existing data sets, including low-level 35-mm color slide photographs, 1:24,000 color airphotos, and ancillary geographic information system (GIS) coverages. Preliminary examination of the data structure included plotting of training set temporal trajectories in spectral space with reference to existing crop calendars. A subsequent stratified, unsupervised classification algorithm, in combination with a geoclimatic rule set and regression analysis, was used to label mapped cells. A map of 20 land cover classes was developed. Classes included agricultural crops and orchards, forest and natural cover types, and urban building densities. An accuracy assessment indicated a final map error of only 26%. The map is now being used to model present and future landscapes for the basin.

This research evaluates the utility of several remote sensing data types for the purpose of mapping forest structure and related attributes at a regional scale. Several sensors were evaluated, including (i) single date Landsat Thematic Mapper (TM); (ii) multitemporal Landsat TM; (iii) Airborne Data Acquisition and Registration (ADAR), a sensor with high spatial resolution; (iv) Airborne Visible-Infrared Imaging Spectrometer (AVIRIS), a sensor with high spectral resolution; and (v) Scanning Lidar Imager Of Canopies By Echo Recovery (SLICER), a lidar sensor that directly measures the height and canopy structure of forest vegetation. To evaluate the ability of each of the sensors to predict stand structure attributes, we assembled a data set consisting of 92 field plots within the Willamette National Forest in the vicinity of the H.J. Andrews Experimental Forest. Stand structure attributes included age, basal area, aboveground biomass, mean diameter at breast height (DBH) of dominant and codominant stems, mean and standard deviation of the DBH of all stems, maximum height, and the density of stems with DBH greater than 100 cm. SLICER performed better than any other remote sensing system in its predictions of forest structural attributes. The performance of the imaging sensors (TM, multitemporal TM, ADAR, and AVIRIS) varied with respect to which forest structural variables were being examined. For one group of variables there was little difference in the ability of the these sensors to predict forest structural attributes. For the remaining variables, we found that multitemporal TM was as or more effective than either ADAR or AVIRIS. These results indicate that multitemporal TM should be investigated as an alternative to either hyperspectral or hyperspatial sensors, which are more expensive and more difficult to process than multitemporal Landsat TM.

Spatially-distributed estimates of biologically-driven CO 2 flux are of interest in relation to understanding the global carbon cycle. Global coverage by satellite sensors offers an opportunity to assess terrestrial carbon (C) flux using a variety of approaches and corresponding spatial resolutions. An important consideration in evaluating the approaches concerns the scale of the spatial heterogeneity in land cover over the domain being studied. In the Pacific Northwest region of the United States, forests are highly fragmented with respect to stand age class and hence C flux. In this study, the effects of spatial resolution on estimates of total annual net primary production (NPP) and net ecosystem production (NEP) for a 96 km 2 area in the central Cascades Mountains of western Oregon were examined. The scaling approach was a simple ‘measure and multiply’ algorithm. At the highest spatial resolution (25 m), a stand age map derived from Landsat Thematic Mapper imagery provided the area for each of six forest age classes. The products of area for each age class and its respective NPP or NEP were summed for the area wide estimates. In order to evaluate potential errors at coarser resolutions, the stand age map was resampled to grain sizes of 100, 250, 500 and 1000 m using a majority filter reclassification. Local variance in near-infrared (NIR) band digital number at successively coarser grain sizes was also examined to characterize the scale of the heterogeneity in the scene. For this managed forest landscape, proportional estimation error in land cover classification at the coarsest resolution varied from 1.0 toC0.6 depending on the initial representation and the spatial distribution of the age class. The overall accuracy of the 1000 m resolution map was 42% with respect to the 25 m map. Analysis of local variance in NIR digital number suggested a patch size on the order of 100‐500 m on a side. Total estimated NPP was 12% lower and total estimated NEP was 4% lower at 1000 m compared to 25 m. Carbon flux estimates based on quantifying differences in total biomass stored on the landscape at two points in time might be affected more strongly by a coarse resolution analysis because the differences among classes in biomass are more extreme than the differences in C flux and because the additional steps in the flux algorithm would contribute to error propagation. Scaling exercises involving reclassification of fine scale imagery over a range of grain sizes may be a useful screening tool for stratifying regions of the terrestrial surface relative to optimizing the spatial resolution for C flux estimation purposes.

When mapping land cover with satellite imagery in montane tropical regions, varying illumination angles and ecological zones can obscure the differences between spectral responses of old-growth forest, secondary forest and agricultural lands. We used multi-date, Landsat Thematic Mapper (TM) imagery to map secondary forests, agricultural lands and old-growth forests in the Talamanca Mountain Range in southern Costa Rica. With stratification by illumination and ecological zone, the overall accuracy for this classification was 87% with a Kappa coefficient of 0.83. We also examined spectral responses to forest successional stage, ecological zone and aspect illumination for the TM Tasselled Cap indices, TM (2 x 6)/7, TM 4/5 and TM difference bands, and whether using digital data from multiple decades improved classification accuracy. Digital maps of ecological zones should be useful for large-scale mapping of land use and forest successional stage in complex montane regions such as those in Central America.

21 of biomass and an LAI of 12, with 90% and like conventional microwave and optical sensors, lidar 75% of variance explained, respectively. Furthermore, we sensors directly measure the distribution of vegetation were able to make accurate estimates of other stand material along the vertical axis and can be used to pro- structure attributes, including the mean and standard devide three-dimensional, or volumetric, characterizations viation of diameter at breast height, the number of stems of vegetation structure. Ecological applications of scan- greater than 100 cm in diameter, and independent estining lidar have hitherto used one-dimensional indices to mates of the basal area of Douglas-fir and western hemcharacterize canopy height. A novel three-dimensional lock. These measurements can be directly related to indianalysis of lidar waveforms was developed to character- ces of forest stand structural complexity, such as those ize the total volume and spatial organization of vegeta- developed for old-growth forest characterization. Indices tion material and empty space within the forest canopy. of canopy structure developed using the novel, threeThese aspects of the physical structure of canopies have dimensional analysis accounted for most of the variables been infrequently measured, from either field or remote used in predictive equations generated by the stepwise methods. We applied this analysis to 22 plots in Douglas- multiple regression. Published by Elsevier Science Inc. fir/western hemlock stands on the west slope of the Cascades Range in Oregon. Each plot had coincident lidar data and field measurements of stand structure. We com- INTRODUCTION pared results from the novel analysis to two earlier methCharacterization of structure in moderate to high bioods of canopy description. Using the indices of canopy mass forests is a major challenge in remote sensing. structure from all three methods of description as inde

Validation, mapping, and modeling efforts require accurate methods to transform process rates and ecosystem attributes estimated from small field plots to the 250–1000-m-wide cells used by a new generation of land cover mapping sensors. We provide alternative scale transformations, each with attendant assumptions and limitations. The choice of method depends on spatial characteristics of the land cover variables in question and consequently may vary between biomes or with the intended application. We extend the fractal similarity dimension renormalization method, previously developed for binary maps, to continuous variables. The method can preserve both the mean and the multifractal properties of the image, thereby satisfying a major goal, namely, to provide accurate areal estimates without sacrificing information about within-site variation. The scale transformation enables the multifractal scaling exponents of landscapes or individual spectral bands to be brought in and out of register with each other, thereby opening another dimension upon which to detect the scales at which various land use or terrain processes operate. Alternatively, landscapes can be selectively rescaled to highlight patterns due to particular processes. We recommend geostatistical procedures with which to assess spatial characteristics both within a site and within individual image cells. We recommend that aggregation of fine-grain measurements during validation of the Moderate Resolution Imaging Spectrometer (MODIS) products be based on continuous variables to reduce errors that originate from uncertainties in binary maps.

Mapping and monitoring of leaf area index (LAI) is important for spatially distributed modeling of vegetation productivity, evapotranspiration, and surface energy balance. Global LAI surfaces will be an early product of the MODIS Land Science Team, and the requirements for LAI validation at selected sites have prompted interest in accurate LAI mapping at a more local scale. While spectral vegetation indices (SVIs) derived from satellite remote sensing have been used to map LAI, vegetation type, and related optical properties, and effects of Sun–surface–sensor geometry, background reflectance, and atmospheric quality can limit the strength and generality of empirical LAI–SVI relationships. In the interest of a preliminary assessment of the variability in LAI–SVI relationships across vegetation types, we compared Landsat 5 Thematic Mapper imagery from three temperate zone sites with on-site LAI measurements. The sites differed widely in location, vegetation physiognomy (grass, shrubs, hardwood forest, and conifer forest), and topographic complexity. Comparisons were made using three different red and near-infrared-based SVIs (NDVI, SR, SAVI). Several derivations of the SVIs were examined, including those based on raw digital numbers (DN), radiance, top of the atmosphere reflectance, and atmospherically corrected reflectance. For one of the sites, which had extreme topographic complexity, additional corrections were made for Sun–surface–sensor geometry. Across all sites, a strong general relationship was preserved, with SVIs increasing up to LAI values of 3 to 5. For all but the coniferous forest site, sensitivity of the SVIs was low at LAI values above 5. In coniferous forests, the SVIs decreased at the highest LAI values because of decreasing near-infrared reflectance associated with the complex canopy in these mature to old-growth stands. The cross-site LAI–SVI relationships based on atmospherically corrected imagery were stronger than those based on DN, radiance, or top of atmosphere reflectance. Topographic corrections at the conifer site altered the SVIs in some cases but had little effect on the LAI–SVI relationships. Significant effects of vegetation properties on SVIs, which were independent of LAI, were evident. The variability between and around the best fit LAI–SVI relationships for this dataset suggests that for local accuracy in development of LAI surfaces it will be desirable to stratify by land cover classes (e.g., physiognomic type and successional stage) and to vary the SVI.

The MODIS sensor to be launched on the EOS-AM platform will be the most important sensor for global vegetation mapping. Among the programmatic goals for the MODIS sensor are to assess and track changes in land use/landcover, leaf area index (LAI), and net primary productivity (NPP). For these products to be used in global models, they must be rigorously validated with site-specific data products. This article presents a review of some of the problems facing a regional- to global-scale validation effort and presents strategies for coordinating the land-cover classification process across multiple sites. We suggest the Enhanced Thematic Mapper (ETM+) as the source of remotely sensed data for validation, and that the IGBP 17-class land-cover classification system be used to provide a link between more complex site-specific systems and global-scale data products. We further recommend that the best site-specific land-cover classifications be obtained, using whatever ancillary data are found to be useful, as a basis for validation. In addition, we propose ways in which ambiguities in translation of classes, from specific to general systems, may be identified. Finally, we stress that even though standardization of methodology among sites may not be appropriate to the goal of obtaining the best possible land-cover products, there should be standardization of error analysis and metadata reporting.

A method of predicting two forest stand structure attributes, basal area and aboveground biomass, from measurements of forest vertical structure was developed and tested using field and remotely sensed canopy structure measurements. Coincident estimates of the vertical distribution of canopy surface area (the canopy height profile), and field-measured stand structure attributes were acquired for two data sets. The chronosequence data set consists of 48 plots in stands distributed within 25 miles of Annapolis, MD, with canopy height profiles measured in the field using the optical-quadrat method. The stem-map data set consists of 75 plots subsetted from a single 32 ha stem-mapped stand, with measurements of their canopy height profiles made using the SLICER (Scanning Lidar Imager of Canopies by Echo Recovery) instrument, an airborne surface lidar system. Four height indices, maximum, median, mean, and quadratic mean canopy height (QMCH) were calculated from the canopy height profiles. Regressions between the indices and stand basal area and biomass were developed using the chronosequence data set. The regression equations developed from the chronosequence data set were then applied to height indices calculated from the remotely sensed canopy height profiles from the stem map data set, and the ability of the regression equations to predict the stem map plot’s stand structure attributes was then evaluated. The QMCH was found to explain the most variance in the chronosequence data set’s stand structure attributes, and to most accurately predict the values of the same attributes in the stem map data set. For the chronosequence data set, the QMCH predicted 70% of variance in stand basal area, and 80% of variance in aboveground biomass, and remained nonasymptotic with basal areas up to 50 m2 ha−1, and aboveground biomass values up to 450 Mg ha−1. When applied to the stem-map data set, the regression equations resulted in basal areas that were, on average, underestimated by 2.1 m2 ha−1, and biomass values were underestimated by 16 Mg ha−1, and explained 37% and 33% of variance, respectively. Differences in the magnitude of the coefficients of determination were due to the wider range of stand conditions found in the chronosequence data set; the standard deviation of residual values were lower in the stem map data set than on the chronosequence data sets. Stepwise multiple regression was performed to predict the two stand structure attributes using the canopy height profile data directly as independent variables, but they did not improve the accuracy of the estimates over the height index approach.

To consider the regional scale effects of forest management requires complete and consistent data over large areas. We used Landsat Thematic Mapper and Multispectral Scanner (TM and MSS) imagery to map forest cover and detect major disturbances between 1975 and 1992 for a 4.2 x 106 ha area of interior British Columbia. Forested pixels were mapped into closed conifer, semiopen conifer, deciduous, and mixed forest classes, with further subdivision of the closed conifer type into three age-classes. The image-based estimate of harvested area was similar to an independent estimate from forest inventory data. Changes in landscape pattern from 1975 to 1992 were examined by calculating indices that describe overall landscape pattern and that of conifer and harvested patches in each biogeoclimatic zone. Harvesting affected 8.4% of the forest area outside provincial parks during the 17-year period. Harvested areas were consistently much smaller than conifer patches in all biogeoclimatic zones and had a lower percentage of interior area and perimeter/area ratio. Conifer patch-shape complexity varied between zones; harvested patches had simpler shapes and were similar in all zones. Results indicate that this landscape is only in the early stages of fragmentation, but a similar harvest pattern has been imposed on differing ecological zones.

A view-angle-dependent brightness gradient was observed in an AVIRIS image of a forested region in Oregon's Cascade Mountains. A method of removing the view-angle effect was sought that would not alter the radiometric integrity of the image, and which would require minimal ancillary information. Four methods were tested and evaluated in terms of remaining brightness gradient and in terms of retention of spectral characteristics. All methods used a quadratic fitting equation to model the changes in brightness across view angles. Other descriptive coefficients were calculated to aid in interpretation. The observed view-angle effect varied with wavelength in a manner consistent with predictions of bidirectional reflectance distribution function characteristics for vegetation. View-angle effects were determined to contain both additive and multiplicative components, with multiplicative components being strong in the chlorophyll absorption region. The view-angle effect in a given pixel was a function of both an underlying viewangle response determined by surface structure and the inherent brightness of that pixel. The most successful compensation method was the one that best accounted for broad differences between pixels in these two components. Despite the simplifying assumptions necessary for empirical view-atigle correction techniques, they can still be useful for hyperspectral remote-sensing data in situations where the view-angle brightness variations would mask variance useful for extracting scene information.

Earth`s climate is greatly influenced by carbon dioxide concentrations in the atmosphere. Besides combustion of fossil fuels, another important anthropogenic source of atmospheric carbon is associated with deforestation, especially in primary forests, which can store large amounts of carbon in the form of organic biomass. There are a number of problems with models of carbon flux previously developed. The authors are developing a strategy to estimate regional carbon fluxes designed to overcome many of them. This article describes the modeling strategy and the results of the carbon flux modeling in forest of the Pacific Northwest region of the USA using the following topic areas: C flux modeling strategy, demonstration of modeling strategy, significance to regional and global carbon cycle, and resource management implications. 53 refs., 7 figs., 1 tab.

RESUMELa detection et la classification des patrons de defoliation par la tordeuse occidentale de l'epinette (Choristoneura occidentalis, Freeman) dans une foret subalpine de l'ouest de l'Oregon ont ete realisees a l'aide d'images video acquises a deux altitudes a partir d'un aeronef ultra leger et a l'aide d'images numeriques multibandes acquises par le capteur thematique de Landsat.L'analyse de texture des images video a donne lieu a une precision de classification de l'ordre de 78 % en reference avec un releve de terrain de la defoliation et de l'interception de lumiere du couvert forestier effectue sur 21 parcelles en septembre 1994. Les donnees acquises par le capteur thematique en 1993, a elles seules, ont permis d'obtenir des resultats de classification satisfaisants dans une proportion de 75 % pour quatre classes de dommages (aucun, leger, modere et severe) dans ces memes parcelles. La precision cartographique etait de l'ordre de 67 % d'apres les resultats d'une classification (K-moyennes) non dir...

Forests of the Pacific Northwest region of the U.S.A. are part of an ongoing political debate that focuses on the trade-offs between commodity and non-commodity values. A key issue in this debate is the location and extent of closed canopy mature and old-growth forest remaining in the region. Remote sensing can play a major part in locating mature and old-growth forests, but. several challenges must be overcome to do so with acceptable accuracy. Conifer forests of the region have high leaf area indices. Thus, most incident solar energy is absorbed, making these forests difficult targets for discrimination of classes. Additionally, spectral characteristics can be affected more by the effects of steep topography than condition of the closed canopy forest. Experimenting with a number of techniques, we estimated and mapped forest age and structure in 1988 over a 1 237 482 ha area on the west side of the Oregon Cascade Range with an overall accuracy of 82 per cent. Unsupervised classification enabled ...

Incorporating ecological disturbance into biogeochemical models is critical for estimating current and future carbon stocks and fluxes. In particular, anthropogenic disturbances, such as forest conversion and wood harvest, strongly affect forest carbon dynamics within North America. This paper summarizes recent (2000.2008) rates of extraction, including both conversion and harvest, derived from national forest inventories for North America (the United States, Canada, and Mexico). During the 2000s, 6.1 million ha/yr were affected by harvest, another 1.0 million ha/yr were converted to other land uses through gross deforestation, and 0.4 million ha/yr were degraded. Thus about 1.0% of North America fs forests experienced some form of anthropogenic disturbance each year. However, due to harvest recovery, afforestation, and reforestation, the total forest area on the continent has been roughly stable during the decade. On average, about 110 m3 of roundwood volume was extracted per hectare harvested across the continent. Patterns of extraction vary among the three countries, with U.S. and Canadian activity dominated by partial and clear ]cut harvest, respectively, and activity in Mexico dominated by conversion (deforestation) for agriculture. Temporal trends in harvest and clearing may be affected by economic variables, technology, and forest policy decisions. While overall rates of extraction appear fairly stable in all three countries since the 1980s, harvest within the United States has shifted toward the southern United States and away from the Pacific Northwest.

All Long-Term Ecological Research (LTER) asks for spatially explicit information. Remote sensing-based data and related analysis products are major sources of such information. This chapter expands on general concepts of remote sensing and most important methodologies in relation to LTER. Examples from US-LTER sites exemplify opportunities related to remote sensing.

To help determine the utility of satellite data for analysis and inventory of Douglas-fir/western hemlock forests west of the Cascade Mountains crest in Oregon and Washington, USA, we evaluate relationships between spectral and texture variables derived from SPOT HRV 10 m panchromatic and LANDSAT TM 30 m multispectral data and 16 forest stand structural attributes. Texture of the HRV data was strongly related to many of the stand attributes evaluated, whereas TM texture was weakly related to all attributes. Wetness, a feature of the TM Tasseled Cap, was the spectral variable most highly correlated to all stand attributes. Wetness appears to respond to the degree of maturity in a forest stand. One of the primary reasons HRV texture and TM wetness exhibited strong relationships with stand attributes is their relative insensitivity to topographically induced illumination angle. Although TM texture also was insensitive to topography, the spatial resolution of TM data is too coarse to detect the spatial variability within the forest stands evaluated. Regression models used to estimate values for the stand attributes from the satellite data indicate that both TM and HRV imagery should yield equally accurate estimates of forest age class and stand structure. Of all stand attributes evaluated, the standard deviation of tree sizes, mean size and density of trees in the upper canopy layers, a structural complexity index, and stand age can be most reliably estimated using the satellite data.

Leaf reflectance changes associated with changes in water stress were analyzed in two separate experiments. Results indicate that the variation in reflectance among collections of leaves of a given species all at the same level of water stress is at least as great as the variation in reflectance associated with changes in water stress for a given leaf collection of that species. The implications is that results from leaf reflectance-water stress studies have only limited applicability to the remote sensing of plant canopy water stress.

Effects of drought-induced water stress on heating-related water transport in three woody plants is examined. By comparing residual water in intact and detached branch tips after heating in a specially designed tree branch calorimeter, and also the energy absorbed during heating, we illustrate that water is transported to heated branch material from unheated material. Similarly, we demonstrate that the amount of water transported is at least partly a function of plant water stress. The implication is that crown fire models do not adequately account for all water vaporized during heating of woody plant crowns.

MODIS (the Moderate Resolution Imaging Spectrometer) is the principal high temporal frequency global mapping sensor on-board NASA's Earth Observation System (EOS) (http://modarch.gsfc.nasa.gov/MODIS/). The MODIS instrument views the entire Earth's surface every 1-2 days, acquiring data in 36 spectral bands at spatial resolutions from 250 m to 1 Ian (Running et al., 1994). The specifications for the MODIS instrument push the limits of engineering (Barnes et al., 1998), and MODIS data volumes will be several times that of the NOAAAVHRR (Masuoka et al., 1998). MODIS data are processed to provide well-quantified and calibrated data sets of the Earth's surface, corrected for instrument radiometry, geometric distortions, atmospheric attenuation, and cloud effects (Justice et aI., 1998a). As these data are used, they will improve our understanding of global dynamics and processes occurring on the land surface and in the oceans and atmosphere. This unprecedented data volume has led the MODIS instrument team to develop a number of derived data products, with the intent of reducing the burden of data processing on the user. A series of land product algorithms were selected by open competition and have been peer-reviewed twice during their development. The MODIS Land Discipline Group (MODLAND) has been charged with development of the MODIS-based algo

The goal is to quantify and explain the carbon (C) budget for Oregon and N. California. The research compares "bottom -up" and "top-down" methods, and develops prototype analytical systems for regional analysis of the carbon balance that are potentially applicable to other continental regions, and that can be used to explore climate, disturbance and land-use effects on the carbon cycle. Objectives are: 1) Improve, test and apply a bottom up approach that synthesizes a spatially nested hierarchy of observations (multispectral remote sensing, inventories, flux and extensive sites), and the Biome-BGC model to quantify the C balance across the region; 2) Improve, test and apply a top down approach for regional and global C flux modeling that uses a model-data fusion scheme (MODIS products, AmeriFlux, atmospheric CO2 concentration network), and a boundary layer model to estimate net ecosystem production (NEP) across the region and partition it among GPP, R(a) and R(h). 3) Provide critical understanding of the controls on regional C balance (how NEP and carbon stocks are influenced by disturbance from fire and management, land use, and interannual climate variation). The key science questions are, "What are the magnitudes and distributions of C sources and sinks on seasonal to decadalmore » time scales, and what processes are controlling their dynamics? What are regional spatial and temporal variations of C sources and sinks? What are the errors and uncertainties in the data products and results (i.e., in situ observations, remote sensing, models)?« less

Initiated in 1972, the Landsat program has provided a continuous record of earth observation for 35 years. The assemblage of Landsat spatial, spectral, and temporal resolutions, over a reasonably sized image extent, results in imagery that can be processed to represent land cover over large areas with an amount of spatial detail that is absolutely unique and indispensable for monitoring, management, and scientific activities. Recent technical problems with the two existing Landsat satellites, and delays in the development and launch of a successor, increase the likelihood that a gap in Landsat continuity may occur. In this communication, we identify the key features of the Landsat program that have resulted in the extensive use of Landsat data for large area land cover mapping and monitoring. We then augment this list of key features by examining the data needs of existing large area land cover monitoring programs. Subsequently, we use this list as a basis for reviewing the current constellation of earth observation satellites to identify potential alternative data sources for large area land cover applications. Notions of a virtual constellation of satellites to meet large area land cover mapping and monitoring needs are also presented. Finally, research priorities that would facilitate the integration of these alternative data sources into existing large area land cover monitoring programs are identified. Continuity of the Landsat program and the measurements provided are critical for scientific, environmental, economic, and social purposes. It is difficult to overstate the importance of Landsat; there are no other systems in orbit, or planned for launch in the short-term, that can duplicate or approach replication, of the measurements and information conferred by Landsat. While technical and political options are being pursued, there is no satellite image data stream poised to enter the National Satellite Land Remote Sensing Data Archive should system failures occur to Landsat-5 and -7.

Semivariograms were used to exploit the spatial information inherent in digital imagery of a variety of Douglas-fir [Pseudotsuga menziesii (Mirbel) Franco] forest stands in the Pacific Northwest region of the United Stataes. Calculations involved digital numbers from single transects of pixels representative of each stand and from a full spatial matrix of pixels from each stand. Digitized aerial video images having pixel sizes of 1 m, 10 m, and 30 m were used. At 1 m spatial resolution the ranges of the matrix semivariograms related to the mean tree canopy sizes of the stands. The sills responded to the presence of vertical layering in the canopies and to percent canopy cover. The transect semivariograms were less representative of overall stand structure, but exhibited periodicity that was suggestive of patterns in stand structure. Semivariograms based on 10 m and 30 m pixels contained significantly less useful information. The ranges of the 10 m matrix semivariograms revealed only whether the tree canopy sizes were less than 10 m or were between 10 m and 20 m. The sills were greatly reduced, but still related well to canopy layering and percent cover. Periodicity in the transect semivariograms was greatly reduced, and in some cases, eliminated. At 30 m resolution, only the sills of the matrix semivariograms contained useful information. However, actual differences between the sills were small.

A 34 year time series (1972–2006) of Landsat imagery for a portion of Snohomish and King Counties, Washington (the Snohomish Water Resource Inventory Area (WRIA)) was analyzed to estimate the amount of land that was converted into impervious surface as a result of urban and residential development. Spectral unmixing was used to determine the fractional composition of vegetation, open, and shadow for each pixel. Unsupervised and supervised classification techniques were then used to derive preliminary land cover maps for each time period. Digital orthophotos were used to create agricultural, forest management, high elevation, and riparian masks. In conjunction with established Urban Growth Areas (UGAs), these masks were utilized for the application of spatial rules that identified impervious surface as a surrogate for urban and residential development. Temporal rules, that minimized classification error, were developed based on each pixel's classified trajectory over the time series of imagery. Overall cross-date classification accuracy for impervious v. non-impervious surface was 95%. The results of the analysis indicate that the area of impervious surface in the Snohomish WRIA increased by 255% over 34 years, from 3285 ha in 1972 to 11,652 ha in 2006. This approach demonstrates the unique value of the 35 year Landsat archive for monitoring impervious surface trends in rapidly urbanizing areas.

Forest disturbances are thought to play a major role in controlling land-atmosphere fluxes of carbon. Under the auspices of the North American Carbon Program, the LEDAPS (Landsat Ecosystem Disturbance Adaptive Processing System) and NACP-FIA projects have been analyzing the Landsat satellite record to assess rates of forest disturbance across North America. In the LEDAPS project, wall-to-wall Landsat imagery for the period 1975-2000 has been converted to surface reflectance and analyzed for decadal losses (disturbance) or gains (regrowth) in biomass using a spectral "disturbance index". The NACP-FIA project relies on a geographic sample of dense Landsat image time series, allowing both disturbance rates and recovery trends to be characterized. Preliminary results for the 1990's indicate high rates of harvest within the southeastern US, Eastern Canada, and the Pacific Northwest, with spatially averaged (~50times50 km) turnover periods as low as 25-40 years. Lower rates of disturbance are found in the Rockies and Northeastern US.

Bottom-up scaling of net ecosystem production (NEP) and net biome production (NBP) was used to generate a carbon budget for a large heterogeneous region (the state of Oregon, 2.5×105 km2) in the western United States. Landsat resolution (30 m) remote sensing provided the basis for mapping land cover and disturbance history, thus allowing us to account for all major fire and logging events over the last 30 years. For NEP, a 23-year record (1980–2002) of distributed meteorology (1 km resolution) at the daily time step was used to drive a process-based carbon cycle model (Biome-BGC). For NBP, fire emissions were computed from remote sensing based estimates of area burned and our mapped biomass estimates. Our estimates for the contribution of logging and crop harvest removals to NBP were from the model simulations and were checked against public records of forest and crop harvesting. The predominately forested ecoregions within our study region had the highest NEP sinks, with ecoregion averages up to 197 gC m−2 yr−1. Agricultural ecoregions were also NEP sinks, reflecting the imbalance of NPP and decomposition of crop residues. For the period 1996–2000, mean NEP for the study area was 17.0 TgC yr−1, with strong interannual variation (SD of 10.6). The sum of forest harvest removals, crop removals, and direct fire emissions amounted to 63% of NEP, leaving a mean NBP of 6.1 TgC yr−1. Carbon sequestration was predominantly on public forestland, where the harvest rate has fallen dramatically in the recent years. Comparison of simulation results with estimates of carbon stocks, and changes in carbon stocks, based on forest inventory data showed generally good agreement. The carbon sequestered as NBP, plus accumulation of forest products in slow turnover pools, offset 51% of the annual emissions of fossil fuel CO2 for the state. State-level NBP dropped below zero in 2002 because of the combination of a dry climate year and a large (200 000 ha) fire. These results highlight the strong influence of land management and interannual variation in climate on the terrestrial carbon flux in the temperate zone.

[1] Knowledge of fuel load and composition is critical in fighting, preventing, and understanding wildfires. Commonly, the generation of fuel maps from remotely sensed imagery has made use of medium-resolution sensors such as Landsat. This paper presents a methodology to generate fuel type maps from high spatial resolution satellite data through object-oriented classification. Fuel maps were derived from QuickBird imagery, which offers a panchromatic and four multispectral bands ranging from 0.61 to 2.44 m resolution. The image used for this paper dated from July 2002 and is located in the NW region of Madrid, Spain. The Prometheus system, a fuel type classification adapted to the ecological characteristics of the European Mediterranean basin, was adopted for this study. Viewed with high-resolution imagery, fuel-related features are often aggregations of pixels exhibiting a variety of spectral properties. Correct identification and classification of these objects requires an explicit consideration of spatial context. We used an object-oriented approach, which allowed context consideration during the classification process, as a complement to traditional pixel-based methods. The map created with this approach was assessed to have greater than 80% accuracy for the prediction of six fuel classes. Results suggested that object-oriented classification of high-resolution imagery has the potential to create accurate and spatially precise fuel maps.

[1] This paper discusses the quality and the accuracy of the Joint Research Center (JRC) fraction of absorbed photosynthetically active radiation (FAPAR) products generated from an analysis of Sea-viewing Wide Field-of-view Sensor (SeaWiFS) data. The FAPAR value acts as an indicator of the presence and state of the vegetation and it can be estimated from remote sensing measurements using a physically based approach. The quality of the SeaWiFS FAPAR products assessed in this paper capitalizes on the availability of a 6-year FAPAR time series over the full globe. This evaluation exercise is performed in two phases involving, first, an analysis of the verisimilitude of the FAPAR products under documented environmental conditions and, second, a direct comparison of the FAPAR values with ground-based estimations where and when the latter are available. This second phase is conducted following a careful analysis of problems arising for performing such a comparison. This results in the grouping of available field information into broad categories representing different radiative transfer regimes. This strategy greatly helps the interpretation of the results since it recognizes the various levels of difficulty and sources of uncertainty associated with the radiative sampling of different types of vegetation canopies.

Exchange of carbon between forests and the atmosphere is a vital component of the global carbon cycle. Satellite laser altimetry has a unique capability for estimating forest canopy height, which has a direct and increasingly well understood relationship to aboveground carbon storage. While the Geoscience Laser Altimeter System (GLAS) onboard the Ice, Cloud and land Elevation Satellite (ICESat) has collected an unparalleled dataset of lidar waveforms over terrestrial targets, processing of ICESat data to estimate forest height is complicated by the pulse broadening associated with large-footprint, waveform-sampling lidar. We combined ICESat waveforms and ancillary topography from the Shuttle Radar Topography Mission to estimate maximum forest height in three ecosystems; tropical broadleaf forests in Brazil, temperate broadleaf forests in Tennessee, and temperate needleleaf forests in Oregon. Final models for each site explained between 59% and 68% of variance in field-measured forest canopy height (RMSE between 4.85 and 12.66 m). In addition, ICESat-derived heights for the Brazilian plots were correlated with field-estimates of aboveground biomass (r(2) = 73%, RMSE = 58.3 Mgha(-1)).

There is concern as to how the balance of carbon in the terrestrial ecosystem will change in response to a variety of land use practices. A study is described in which a methodology is being developed to help narrow this uncertainty for the temperate forets of the Pacific Northwest region of the US. A carbon storage model is being developed to respond to forest harvesting, the dominant use of land in the region. By linking the carbon model to satellite imagery and a climate simulation model, the current amount of carbon stored in the forests of the Pacific northwest is estimated. The archive of Landsat multispectral scanner (MSS) images permits a 20-year historical perspective of land use changes in the region. With these data, the recent impact of regional land use in forest carbon stores is assessed.

We monitored the status and trend of late-successional and old-growth forest (older forest) on 24 million ac of land managed by the Forest Service, Bureau of Land Management, and National Park Service in the Northwest Forest Plan (the Plan) area between 1994 and 2003. We developed baseline maps from satellite imagery of older forest conditions at the start of the Plan. We used remotely sensed change detection to track losses of older forests on federally managed lands to standreplacing harvest and wildfire, and we analyzed the amounts and spatial distribution of older forests by using the mapped data. We also performed statistical analysis on inventory plot information collected on Forest Service and Bureau of Land Management lands. These analyses provided statistically rigorous estimates of older forest acres bracketed by confidence intervals. We analyzed remeasured inventory plots to estimate net change in the amount of older forests on federally managed lands. We estimated the amount of older forest at the start of the Plan corresponding to three different older forest definitions based on average tree size, canopy layering, canopy closure, and life form. The results ranged from 7.87 million ac (± 1.96 million ac) of federally managed lands with average tree size at least 20 in (medium and large older forest), to 7.04 million ac (± 1.93 million ac) using a definition that recognizes variation in regional forest vegetation (older forest with size indexed to potential natural vegetation zone). We found 2.72 million ac (± 0.35 million ac) were in stands with average tree size 30 in and greater, with multistoried canopies (large, multistoried older forest). At least 1.7 million ac of existing “medium and large” older forest were in fire-adapted vegetation types characterized by high fire frequency and low severity in the Eastern Cascades and Klamath provinces. Up to 1 million additional older forest ac occurred in dry mixed-conifer types in the Western Cascades. Our data from remeasured inventory plots indicated that the annual net rate of increase of “medium and large” older forest was about 1.9 percent, outpacing losses from all sources. The extrapolated gain in older forest 20 in was between 1.25 million ac and 1.5 million ac in the first decade after the Plan. The gain came primarily from increases in the area of forest at the lower end of the diameter range for older forest. The net increase took into account the older forest removed by stand-replacing harvest, 0.2 percent of the total (about 16,900 ac on all federally managed lands), and the amount burned by stand-replacing wildfire, about 1.3 percent (about 102,500 ac on all federally managed lands). The area mapped as logged or burned had an error estimate of between 7 and 12 percent. The initial amount, distribution, and arrangement of older forest on federally managed land appears to have met or exceeded Northwest Forest Plan expectations. But the large amount of older forest susceptible to catastrophic wildfire may be a concern for managers. Losses to wildfire in the first decade were in line with assumptions for the Plan area, but rates of loss were highly variable among provinces, with the highest rates of loss occurring in the dry provinces. Loss of older forest to harvest was a fraction of the approximately 230,000 ac of older forest expected to have been harvested. Overall gain was about twice the 600,000 ac expected during the first decade of the Plan.

An increasing number of sensors are available for forest ecologists and managers seeking to map attributes of forest canopy cover, forest structure and composition, and their dynamics. This Chapter seeks to put these advances within the context of the needs of forest managers and scientists. To do so, we review the basic physics behind a variety of imagery types, discuss fundamental limitations and trade-offs that apply to all remotely sensed data, review sensor options for several established and emerging technologies, and present our approach for matching imagery and attributes of interest.