Scott J. Goetz

High-resolution global maps of 21st-century forest cover change.

Forests in Flux Forests worldwide are in a state of flux, with accelerating losses in some regions and gains in others. Hansen et al. (p. 850) examined global Landsat data at a 30-meter spatial resolution to characterize forest extent, loss, and gain from 2000 to 2012. Globally, 2.3 million square kilometers of forest were lost during the 12-year study period and 0.8 million square kilometers of new forest were gained. The tropics exhibited both the greatest losses and the greatest gains (through regrowth and plantation), with losses outstripping gains. Landsat data reveals details of forest losses and gains across the globe on an annual basis from 2000 to 2012. Quantification of global forest change has been lacking despite the recognized importance of forest ecosystem services. In this study, Earth observation satellite data were used to map global forest loss (2.3 million square kilometers) and gain (0.8 million square kilometers) from 2000 to 2012 at a spatial resolution of 30 meters. The tropics were the only climate domain to exhibit a trend, with forest loss increasing by 2101 square kilometers per year. Brazil’s well-documented reduction in deforestation was offset by increasing forest loss in Indonesia, Malaysia, Paraguay, Bolivia, Zambia, Angola, and elsewhere. Intensive forestry practiced within subtropical forests resulted in the highest rates of forest change globally. Boreal forest loss due largely to fire and forestry was second to that in the tropics in absolute and proportional terms. These results depict a globally consistent and locally relevant record of forest change.

Radiometric rectification: Toward a common radiometric response among multidate, multisensor images☆

Abstract A common radiometric response is required for quantitative analysis of multiple satellite images of a scene acquired on different dates with different sensors. We describe a technique to “radiometrically rectify” multiple Landsat images of a scene to a reference image, and evaluate it using a pair of Landsat 5 images acquired 2 years apart. All rectified images should appear as if they were acquired with the same sensor, while observing through the atmospheric and illumination conditions of the reference image. If atmospheric optical depth and sensor calibration date are available for the reference image, then an atmospheric correction algorithm may be used to correct all the rectified images to absolute surface reflectance. The “radiometric rectification” algorithm identifies “radiometric control sets,” i.e., sets of scene landscape elements with a mean reflectance which is expected to change little with time. The average digital count values of these radiometric control sets are used to calculate linear transforms relating digital count values between images. We evaluate the technique empirically with a pair of Landsat 5 TM images of a scene for which surface reflectance and atmospheric optical depth data are available. We also examine its performance under a wide range of atmospheric conditions using simulations based on atmospheric models. We find that the radiometric rectification algorithm performed well for the visible and near infrared bands, adjusting surface reflectance for the effects of relative atmospheric differences to within 1%. The performance is not as good for the midinfrared bands on TM. There are several possible causes for this; we could not determine which was the most important. We conclude from these studies that for scenes containing reflectance stable elements, radiometric rectification should be a useful alternative to atmospheric radiative transfer codes and sensor calibration approaches when reliable atmospheric optical depth data or calibration coefficients are not available. When atmospheric optical data and sensor calibration information are available for one of a sequence of radiometrically rectified images, an atmospheric radiative transfer code may be used to correct each image in the sequence to absolute surface reflectance.

Shrub expansion in tundra ecosystems: dynamics, impacts and research priorities

Recent research using repeat photography, long-term ecological monitoring and dendrochronology has documented shrub expansion in arctic, high-latitude and alpine tundra

Satellite remote sensing of surface energy balance: Success, failures, and unresolved issues in FIFE

The FIFE staff science group, consisting of the authors, developed and evaluated process models relating surface energy and mass flux, that is, surface rates, to boundary layer and surface biophysical characteristics, that is, surface states. In addition, we developed and evaluated remote sensing algorithms for inferring surface state characteristics. In this paper we report the results of our efforts. We also look in detail at the sensor and satellite platform requirements (spatial resolution and orbital requirements) as driven by surface energy balance dynamics and spatial variability. We examine also the scale invariance of the process models and remote sensing algorithms, that is, to what degree do the remotely sensed parameters and energy balance relations translate from the patch level where they were developed to the mesoscale level where they are required? Finally, we examine the atmospheric correction and calibration issues involved in extending the remotely sensed observations within a season and between years. From these investigations we conclude that (1) existing formulations for the radiation balance and latent heat components of the surface energy balance equation are valid at the patch level. (2) Many of the surface physiological characteristics that parameterize these formulations can be estimated using satellite remote sensing at both local and regional scales; a few important ones cannot. (3) The mathematical structures relating radiation and surface energy flux to remote sensing parameters are, for the most part, scale invariant over the First International Satellite Land Surface Climatology Project (ISLSCP) Field Experiment (FIFE) study area. The conditions for scale invariance are derived. (4) The precision of satellite remote sensing estimates of surface reflectance, calibrated and corrected for atmospheric effects, is no worse than about 1% absolute. The errors may actually be smaller, but an upper bound of 1% results from sampling variance caused by differences among the satellite and ground sensors in spatial resolution, atmospheric effects, and calibration. (5) Afternoon cumulus in the study area required both the Landsat and the SPOT satellites for monitoring of the vegetation dynamics. This result implies the need for multiple polar orbiters, or geosynchronous satellites in an operational implementation. We found that canopy Fpar, the fraction of incident photosynthetically active radiation absorbed by a canopy, can be estimated with an error of about 10% using remote sensing, provided that regional variability in the reflectance of the canopy substrate is dealt with properly. We also found that spectral vegetation indices (VIs) respond primarily to the photosynthetically active radiation absorbed by the live or green component of the canopy as opposed to its necrotic or dead vegetation. This is of critical importance since radiation absorption by the live part of the canopy is the rate-limiting process for photosynthesis and other key process rates such as evaporation. We found for the FIFE study area the surface moisture content at O to 10 cm to be another key rate-limiting variable in photosynthesis and evaporation. At gravimetric soil moisture levels below 20%, photosynthesis and evaporation were strongly attenuated. Only microwave sensors have shown potential for satellite remote sensing of soil moisture and only in the top few centimeters. Hydrological models may also play a critical role in monitoring root zone soil moisture levels, but additional research is needed. From our review of the research of others in FIFE we conclude that downwelling shortwave radiation and surface albedo are also amenable to remote sensing. Unfortunately, from our research we also found that the remote estimation of surface temperature to useful accuracies is problematical; consequently, the use of thermal infrared measurements to infer sensible heat flux is probably not feasible to acceptable accuracies.

A first map of tropical Africa's above-ground biomass derived from satellite imagery

Observations from the moderate resolution imaging spectroradiometer (MODIS) were used in combination with a large data set of field measurements to map woody above-ground biomass (AGB) across tropical Africa. We generated a best-quality cloud-free mosaic of MODIS satellite reflectance observations for the period 2000‐2003 and used a regression tree model to predict AGB at 1 km resolution. Results based on a cross-validation approach show that the model explained 82% of the variance in AGB, with a root mean square error of 50.5 Mg ha −1 for a range of biomass between 0 and 454 Mg ha −1 . Analysis of lidar metrics from the Geoscience Laser Altimetry System (GLAS), which are sensitive to vegetation structure, indicate that the model successfully captured the regional distribution of AGB. The results showed a strong positive correlation (R 2 = 0.90) between the GLAS height metrics and predicted AGB.

Large-Scale Patterns of Forest Succession as Determined by Remote Sensing

The spatial pattern of and the transition rates between forest ecological states were inferred for °260 000 pixel—sized (3600 m2) landscape units using stallite remote sensing. Transition rates were estimated from 1973 to 1983 Landsat images of the study area, classified into ecological states associated with forest succession. The effects of classification error on transition rate estimates were modeled and error adjustments made. Classification of the 1973 and 1983 Landsat images of the 900 km2 study region required a relatively small set of ground—observed and photo—interpreted plots in 1983, with a total area of just 1.62 km2. An innovative technique for correcting multiyear Landsat images for between—image differences in atmospheric effects and sensor calibration, permitted classification of the 1973 Landsat image using 1983 ground observations. Given current Landsat data, and ground observations in one year, this technique would permit monitoring of forest succession and dynamics for nearly a 20—yr period. Results of applying these techniques to a forest ecosystem showed that during the 10—yr observation period it was patchy and dynamic. For both a wilderness and a nonwilderness area in the study region, sizeable values of transition rates were observed and over half of the landscape units were observed to change state: however, a Markov analysis, using the observed transition probabilities, suggests that at the regional level neither the wilderness nor the nonwilderness areal proportions of ecological states are undergoing rapid change.

Temperature and vegetation seasonality diminishment over northern lands

Pronounced increases in winter temperature result in lower seasonal temperature differences, with implications for vegetation seasonality and productivity. Research now indicates that temperature and vegetation seasonality in northern ecosystems have diminished to an extent equivalent to a southerly shift of 4°– 7° in latitude, and may reach the equivalent of up to 20° over the twenty-first century.

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.

Expansion of Industrial Logging in Central Africa

Industrial logging has become the most extensive land use in Central Africa, with more than 600,000 square kilometers (30%) of forest currently under concession. With use of a time series of satellite imagery for the period from 1976 to 2003, we measured 51,916 kilometers of new logging roads. The density of roads across the forested region was 0.03 kilometer per square kilometer, but areas of Gabon and Equatorial Guinea had values over 0.09 kilometer per square kilometer. A new frontier of logging expansion was identified within the Democratic Republic of Congo, which contains 63% of the remaining forest of the region. Tree felling and skid trails increased disturbance in selectively logged areas.

Using the SLEUTH urban growth model to simulate the impacts of future policy scenarios on urban land use in the Baltimore^Washington metropolitan area

Declining water quality in the Chesapeake Bay estuary is in part the result of disruptions in the hydrological system caused by urban and suburban development throughout its 167000 km2 watershed. A modeling system that could provide regional assessments of future development and explore the potential impacts of different regional management scenarios would be useful for a wide range of applications relevant to the future health of the Bay and its tributaries. We describe and test a regional predictive modeling system that could be used to meet these needs. An existing cellular automaton model, SLEUTH, was applied to a 23 700 km2 area centered on the Washington-Baltimore metropolitan region, which has experienced rapid land-use change in recent years. The model was calibrated using a historic time series of developed areas derived from remote sensing imagery, and future growth was projected out to 2030 assuming three different policy scenarios: (1) current trends, (2) managed growth, and (3) ecologically sustainable growth. The current trends scenario allowed areas on the urban fringe that are currently rural or forested to be developed, which would have implications for water quality in the Chesapeake Bay and its tributaries. The managed growth and ecologically sustainable scenarios produced growth patterns that were more constrained and which consumed less natural resource land. This application of the SLEUTH model demonstrates an ability to address a range of regional planning issues, but spatial accuracy and scale sensitivity are among the factors that must be further considered for practical application.