Josef Kellndorfer

Mapping and monitoring carbon stocks with satellite observations: a comparison of methods.

Mapping and monitoring carbon stocks in forested regions of the world, particularly the tropics, has attracted a great deal of attention in recent years as deforestation and forest degradation account for up to 30% of anthropogenic carbon emissions, and are now included in climate change negotiations. We review the potential for satellites to measure carbon stocks, specifically aboveground biomass (AGB), and provide an overview of a range of approaches that have been developed and used to map AGB across a diverse set of conditions and geographic areas. We provide a summary of types of remote sensing measurements relevant to mapping AGB, and assess the relative merits and limitations of each. We then provide an overview of traditional techniques of mapping AGB based on ascribing field measurements to vegetation or land cover type classes, and describe the merits and limitations of those relative to recent data mining algorithms used in the context of an approach based on direct utilization of remote sensing measurements, whether optical or lidar reflectance, or radar backscatter. We conclude that while satellite remote sensing has often been discounted as inadequate for the task, attempts to map AGB without satellite imagery are insufficient. Moreover, the direct remote sensing approach provided more coherent maps of AGB relative to traditional approaches. We demonstrate this with a case study focused on continental Africa and discuss the work in the context of reducing uncertainty for carbon monitoring and markets.

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. >

Large-Area Classification and Mapping of Forest and Land Cover in the Brazilian Amazon: A Comparative Analysis of ALOS/PALSAR and Landsat Data Sources

Information on the distribution of tropical forests is critical to decision-making on a host of globally significant issues ranging from climate stabilization and biodiversity conservation to poverty reduction and human health. The majority of tropical nations need high-resolution, satellite-based maps of their forests as the international community now works to craft an incentive-based mechanism to compensate tropical nations for maintaining their forests intact. The effectiveness of such a mechanism will depend in large part on the capacity of current and near-future Earth observation satellites to provide information that meets the requirements of international monitoring protocols now being discussed. Here we assess the ability of a state-of-the-art satellite radar sensor, the ALOS/PALSAR, to support large-area land cover classification as well as high-resolution baseline mapping of tropical forest cover. Through a comprehensive comparative analysis involving twenty separate PALSAR- and Landsat-based classifications, we confirm the potential of PALSAR as an accurate (>90%) source for spatially explicit estimates of forest cover based on data and analyses from a large and diverse region encompassing the Xingu River headwaters in southeastern Amazonia. Pair-wise spatial comparisons among maps derived from PALSAR, Landsat, and PRODES, the Brazilian Amazon deforestation monitoring program, revealed a high degree of spatial similarity. Given that a long-term data record consisting of current and future spaceborne radar sensors is now expected, our results point to the important role that spaceborne imaging radar can play in complementing optical remote sensing to enable the design of robust forest monitoring systems.

Mapping vegetated wetlands of Alaska using L-band radar satellite imagery.

Wetlands act as major sinks and sources of important atmospheric greenhouse gases and can switch between atmospheric sink and source in response to climatic and anthropogenic forces in ways that are poorly understood. Despite their importance in the carbon cycle, the locations, types, and extents of northern wetlands are not accurately known. We have used two seasons of L-band synthetic aperture radar (SAR) imagery to produce a thematic map of wetlands throughout Alaska. The classification is developed using the Random Forests decision tree algorithm with training and testing data compiled from the National Wetlands Inventory (NWI) and the Alaska Geospatial Data Clearinghouse (AGDC). Mosaics of summer and winter Japanese Earth Resources Satellite 1 (JERS-1) SAR imagery were employed together with other inputs and ancillary datasets, including the SAR backscatter texture map, slope and elevation maps from a digital elevation model (DEM), an open-water map, a map of proximity to water, data collection dates, and geographic latitude. The accuracy of the resulting thematic map was quantified using extensive ground reference data. This approach distinguished as many as nine different wetlands classes, which were aggregated into four vegetated wetland classes. The per-class average error rate for aggregate wetlands classes ranged between 5.0% and 30.5%, and the total aggregate accuracy calculated based on all classified pixels was 89.5%. As the first high-resolution large-scale synoptic wetlands map of Alaska, this product provides an initial basis for improved characterization of land-atmosphere CH4 and CO2 fluxes and climate change impacts associated with thawing soils and changes in extent and drying of wetland ecosystems.

Toward consistent regional-to-global-scale vegetation characterization using orbital SAR systems

A study was conducted to assess the potential of combined imagery from the existing European and Japanese orbitar synthetic aperture radar (SAR) systems, ERS-1 (C-hand, VV-polarization) and JERS-1 (L-band, HH-palarization), for regional-to-global-scale vegetation classification. For seven test sites from various ecoregions in North and South America, ERS-1/JERS-1 composites were generated using high-resolution digital elevation model (DEM) data for terrain correction of geometric and radiometric distortions. An edge-preserving speckle reduction process was applied to reduce the fading variance and prepare the data for an unsupervised clustering of the two-dimensional (2D) SAR feature space. Signature-based classification of the clusters was performed for all test sites with the same set of radar backscatter signatures, which were measured from well-defined polygons throughout all test sites. While trained on one-half of the polygons, the classification result was tested against the other half of the total sample population. The multisite study was followed by a multitemporal study in one test site, clearly showing the necessity of including multitemporal data beyond a level 1 (woody, herbaceous, mixed) vegetation characterization. Finally, classifications with simulation of backscatter variations shows the dependence of the classification results on calibration accuracy and on naturally occurring backscatter changes of natural surfaces. Overall, it is demonstrated that the combination of existing orbital L- and C-band SAR data is quite powerful for structural vegetation characterization.

A National, Detailed Map of Forest Aboveground Carbon Stocks in Mexico

A spatially explicit map of aboveground carbon stored in Mexico’s forests was generated from empirical modeling on forest inventory and spaceborne optical and radar data. Between 2004 and 2007, the Mexican National Forestry Commission (CONAFOR) established a network of ~26,000 permanent inventory plots in the frame of their national inventory program, the Inventario Nacional Forestal y de Suelos (INFyS). INFyS data served as model response for spatially extending the field-based estimates of carbon stored in the aboveground live dry biomass to a wall-to-wall map, with 30 × 30 m2 pixel posting using canopy density estimates derived from Landsat, L-Band radar data from ALOS PALSAR, as well as elevation information derived from the Shuttle Radar Topography Mission (SRTM) data set. Validation against an independent set of INFyS plots resulted in a coefficient of determination (R2) of 0.5 with a root mean square error (RMSE) of 14 t∙C/ha in the case of flat terrain. The validation for different forest types showed a consistently low estimation bias ( 15°) with an R2 of 0.34. A comparison of the average carbon stocks computed from: (a) the map; and (b) statistical estimates from INFyS, at the scale of ~650 km2 large hexagons (R2 of 0.78, RMSE of 5 t∙C/ha) and Mexican states (R2 of 0.98, RMSE of 1.4 t∙C/ha), showed strong agreement.

Toward precision forestry: plot-level parameter retrieval for slash pine plantations with JPL AIRSAR

During an EOCAP-SAR project, Airsar P-, L-, and C-band data were used to test the capability of synthetic aperture radar (SAR) to predict biometric parameters which are frequently used by timber managers as inputs to growth, harvest, and yield models. Test site was a commercially managed area in Jesup, southeastern Georgia, with stands owned by Plum Creek Timber Company. A biometric survey data set from 118 cluster plots in two thinned and two unthinned slash pine stands was used to determine the correlation with the Airsar data at the plot level. A statistical model was used to test all possible scenarios of polarimetric and frequency combinations. Using a weighted least squares linear regression model, P/sub HV/ proved to be the single most significant band/polarization combination to correlate with dominant height, basal area, and volume at adjusted squared correlation coefficients of 0.70, 0.80, and 0.84, respectively.

Mapping Canopy Height and Growing Stock Volume Using Airborne Lidar, ALOS PALSAR and Landsat ETM+

Abstract: We have investigated for forest plantations in Chile the stand-level retrieval of canopy height (CH) and growing stock volume (GSV) using Airborne Laser Scanner (ALS), ALOS PALSAR and Landsat. In a two-stage up-scaling approach, ensemble regression tree models (randomForest) were used to relate a suite of ALS canopy structure indices to stand-level in situ measurements of CH and GSV for 319 stands. The retrieval of CH and GSV with ALS yielded high accuracies with R 2 s of 0.93 and 0.81, respectively. A second set of randomForest models was developed using multi-temporal ALOS PALSAR intensities and repeat-pass coherences in two polarizations as well as Landsat data as predictor and stand-level ALS based estimates of CH and GSV as response variables. At three test sites, the retrieval of CH and GSV with PALSAR/Landsat reached promising accuracies with R 2 s in the range of 0.7 to 0.85. We show that the combined use of multi-temporal PALSAR intensity, coherence and Landsat yields higher retrieval accuracies than the retrieval with any of the datasets alone. Potential limitations for the large-area application of the fusion approach included (1) the low sensitivity of ALS first/last return data to forest horizontal structure, affecting the retrieval of GSV in less managed types of forest, and (2) the dense ALS sampling required to achieve high retrieval accuracies at larger scale.

MAD-MEX: Automatic Wall-to-Wall Land Cover Monitoring for the Mexican REDD-MRV Program Using All Landsat Data

Estimating forest area at a national scale within the United Nations program of Reducing Emissions from Deforestation and Forest Degradation (REDD) is primarily based on land cover information using remote sensing technologies. Timely delivery for a country of a size like Mexico can only be achieved in a standardized and cost-effective manner by automatic image classification. This paper describes the operational land cover monitoring system for Mexico. It utilizes national-scale cartographic reference data, all available Landsat satellite imagery, and field inventory data for validation. Seven annual national land cover maps between 1993 and 2008 were produced. The classification scheme defined 9 and 12 classes at two hierarchical levels. Overall accuracies achieved were up to 76%. Tropical and temperate forest was classified with accuracy up to 78% and 82%, respectively. Although specifically designed for the needs of Mexico, the general process is suitable for other participating countries in the REDD+ program to comply with guidelines on standardization and transparency of methods and to assure comparability. However, reporting of change is ill-advised based on the annual land cover products and a combination of annual land cover and change detection algorithms is suggested.

The role of science in Reducing Emissions from Deforestation and Forest Degradation (REDD)

Emissions of carbon from tropical deforestation and degradation currently account for 12–15% of total anthropogenic carbon emissions each year, and Reducing Emissions from Deforestation and Forest Degradation (REDD; including REDD+) is poised to be the primary international mechanism with the potential to reduce these emissions. This article provides a brief summary of the scientific research that led to REDD, and that continues to help refine and resolve issues of effectiveness, efficiency and equitability for a REDD mechanism. However, REDD deals only with tropical forests and there are other regions, ecosystems and processes that govern the sources and sinks of carbon in terrestrial ecosystems. Ongoing research will reveal which of these other flows of carbon are most important, and which of them might present further opportunities to reduce emissions (or enhance sinks) through environmental policy mechanisms, as well as how they might do this.