What New Research Is There On Monitoring Land-Cover Change?


New Maps Of Global Land Cover

Although new conventional maps of vegetation and land cover are being generated at the local and national levels, the distribution of global land cover and vegetation types at any one time is poorly known. For large areas of the world, maps of land cover are often out of date, inaccurate or at a scale inappropriate for global change studies. Ecosystem process models currently require a classification of land cover types. Coarse resolution multi-temporal satellites provide a means for generating up-to-date global and regional data sets. Improved global data sets from the Advanced Very High Resolution Radiometer (AVHRR) are being made available to the science community. Daily global orbit data are being processed and compiled into time series data sets at scales ranging from one degree to one kilometer. Phenological differences represented by vegetation indices from the satellite data are being used to discriminate different vegetation types. Multi-spectral data are also being used to discriminate different land cover types. Validation of these new global maps is being undertaken by the use of sample high resolution data. The U.S. is also currently contributing to an international global 1 km land cover data set production and validation effort coordinated by the IGBP.
Reference: NDVI-Derived Land Cover Classifications at a Global Scale, Defries, R. S., and J. R. G. Townshend, International Journal of Remote Sensing, Vol. 15, pp. 3567-3586, 1994.

Overview Of The Areal Extent And Geographic Pattern Of Extreme Anthropogenic Perturbations Of The Terrestrial Biosphere

Since the development of agriculture some 9000 years ago, an ever-increasing proportion of the Earth's surface has been expropriated for the production of food crops. Conversion to cultivated systems has resulted in wholesale restructuring of biotic communities and the pathways of energy and nutrient flow that they control. The table below summarizes the areal extent and geographical pattern of extreme anthropogenic perturbations of the terrestrial biosphere. Sisk et al. (pp. 592-604) argue that species extinction's are most likely to occur where such areas coincide with high biological diversity. Attempts to identify areas of broad and intense perturbation of native habitats should be an integral part of any effort to identify priority areas for conservation.

Breakdown By Vegetation Class

Class Pre-agricultural
area
(square kilometers)
Post-agricultrual
area
(square kilometers)
Percentage
converted to
agriculture
Forest 49,992,600 42,100,100 15.79
Woodland 12,334,000 10,440,000 15.36
Shrubland 13,163,700 12,143,000 7.75
Tundra 7,294,000 7,294,000 0.00
Grassland 34,150,500 27,197,650 20.36
Desert 15,910,000 15,580,000 2.07
Cultivation
18,090,050
Total ice-free
land area
132,844,800 132,844,800 13.62
Reference: Mapping Human Impacts on the Global Biosphere, Imhoff, M.L., BioScience, Vol. 44 p. 598, 1994.

Global Land Cover Analysis By Satellite

A simple new logic for classifying vegetation using satellite data is a crucial step towards characterizing the Earth's biosphere for input into general circulation models. This classification is based on fundamental, morphological aspects of the vegetation, such as leaf shape and longevity. It is a classification that lends itself easily to biophysical measurements, such as leaf area index and total biomass. The data source is the 1 km Advanced Very High Resolution Radiometer (AVHRR).
Reference: A Remote Sensing Based Vegetation Classification Logic for Global Land Cover Analysis, Running, S. W., T. R. Loveland, L. L. Pierce, R. R. Nemani, and E. R. Hunt, Jr., Remote Sensing of the Environment, Vol. 51, pp. 39-48, 1995.

Biomass Density Inferred Over Numerous Test Sites Using A New Satellite Remote Sensing Algorithm

A new satellite remote sensing algorithm is used to infer a group of forest biophysical characteristics key to studies of land-use change, energy, water and carbon cycling in communities of black spruce (Picea mariana), the most common boreal forest dominant. The algorithm is based on geometric radiative transfer models and spectral linear mixing models. The algorithm infers values for the sunlit canopy fraction, sunlit background fraction and shadow fraction in forest stands, based on their visible and near infrared reflectance in Thematic Mapper bands 3 and 4. Using relations between these fractions and the stand biophysical characteristics, biomass density is inferred over 31 test sites, with very good results.
Reference: Remote Sensing of Forest Biophysical Structure in Boreal Stands of Picea Mariana Using Mixture Decomposition and Geometric Reflectance Models, Hall, F. G., Y. E. Shimabukuro, and K. F. Huemmrich, Ecological Applications, in press, 1995.

The Change In Closed Canopy Coniferous Forest Due Primarily To Clear Cutting Was Greatest In The Private Land And Least In Public Reserves

The process of fragmentation was examined in a managed forest landscape by comparing rates and patterns of disturbance (primarily clear-cutting) and regrowth between 1972 and 1988 using Landsat imagery. A 2589 square kilometer managed forest landscape in western Oregon was classified into two major forest types, closed-canopy conifer forest (CF - typically greater than 60% conifer cover) and other forest and nonforest types (OT - typically less than 40-year-old or deciduous forest).

The percentage of CF declined from 71% to 58% between 1972 and 1988. Declines were greatest on private land, least in public reserves, and intermediate in public non reserves. High elevations (greater than 914 meters) maintained a greater percentage of CF than lower elevations (less than 914 meters). The percentage of the area at the edge of the two cover types increased on all ownership's and in both elevational zones, whereas the amount of interior habitat (defined as CF at least 100 meters from OT) decreased on all ownerships and elevational zones. By 1988 public lands contained roughly 45% interior habitat while private lands had 12% interior habitat. Mean interior patch area declined from 160 to 62 hectares. The annual rate of disturbance (primarily clear-cutting) for the entire area was 1.19%, which corresponds to a cutting rotation of 84 years. The forest landscape was not in a steady state or regulated condition, a situation that is not projected to occur for at least 40 years under current forest plans. Variability in cutting rates within ownerships was higher on private land than on non reserve public land. However, despite the use of dispersed cutting patterns on public land, spatial patterns of cutting and remnant forest patches were non uniform across the entire public ownership. Large remaining patches (less than 5000 hectares) of contiguous interior forest were restricted to public lands designated for uses other than timber production such as wilderness areas and research natural areas.

Reference: Dynamics and Pattern of Managed Coniferous Forest Landscape in Oregon, Spies, T. A., W. J. Ripple, and G. A. Bradshaw, Ecological Applications, Vol. 4, pp. 555-568, 1994.

A New Approach To Monitoring Deforestation And Urbanization

Thermal infrared surface temperature measurements, in conjunction with remote measurements of vegetation index, are being used for the first time to study changes in surface hydroclimate brought about by urbanization or deforestation. Two land surface parameters, the surface moisture availability and the fractional vegetation cover, are obtained from satellite measurements. These parameters, varying within numerical limits of zero to one, are useful in two ways. First, they reflect most of the variation in the turbulent heat flux between the land surface and atmosphere; as such, they constitute robust measures as key input quantities in climate models. Second, the two parameters, being sensitive to changes in the land surface cover, serve as joint indices of urbanization and deforestation. Thus, baseline images of these parameters can be created and their evolution charted over periods such as a decade or more. Consequently, changes in land surface character can be monitored. Studies over target areas in Pennsylvania (urbanization) and Costa Rica (deforestation) have been initiated using these parameters to monitor changes in land use.
Reference: Thermal Remote Sensing of Surface Soil Water Content with Partial Vegetation Cover for Incorporation into Climate Models, Gillies, R. R., and T. N. Carlson, Journal of Applied Meteorology, 34, 745-756, 1995.

Forest Fragmentation Studies Suggest Carbon Fluxes From The Brazilian Amazon May Have Been Overestimated

Landsat satellite imagery was used to measure deforestation and forest fragmentation for 1978 and 1988 in the Brazilian Amazon Basin. The deforestation was concentrated in a crescent along the southern and eastern fringe of the Amazon forest. The area deforested increased from 78,000 square km to 230,000 square km from 1978 to 1988. This result agrees closely with independently derived estimates made by the Brazilian Space Agency (INPE) and is lower than a number of recent estimates in the published literature. It also demonstrated the previous estimates based on coarse resolution (AVHRR) satellite data overestimated Brazilian deforestation by about 50 percent. These estimates will be used to refine assessments of net flux of carbon from land clearing and biomass burning. Model estimates based on higher deforestation values are probably too high. In addition, many deforested areas are in stages of regrowth following abandonment, and if this regrowth is widespread, estimates of net carbon flux may need to be further reduced because of carbon accumulation in this regrowing biomass. Work to produce similar estimates for other parts of the humid tropics continues under the multi-agency Landsat Pathfinder Program.

In this same study, fragmented forest was also estimated. Fragmented forest was defined as areas of less than 100 square km surrounded by deforestation, and edge effects of 1 km into forest from adjacent areas of deforestation. The area of fragmented forest increased from 162,000 square km to 588,000 square km. This much larger area that is impacted by deforestation has serious implications with regard to other land-atmosphere interactions as well as biological diversity.

Reference: Tropical Deforestation and Habitat Fragmentation in the Amazon: Satellite Data from 1978 to 1988, Skole, D., and Compton Tucker, Science, Vol. 260, pp. 1905-1910, 1993.

For comments, please contact the GCRIO Web Team at: help@gcrio.org
Last updated 04/10/96