Countries may determine that other activities will be allowed in addition to those that result in forest clearing and regrowth. Among the suite of activities currently under consideration in this Special Report, few appear to be suitable for the use of remotely sensed data. An exception might be wood harvest. Clear-cutting would most likely be observed, yet distinguishing between clear-cutting and clearing would be difficult. Satellite data have also been used to identify selective logging (Stone and Lefebvre, 1998; Souza and Barreto, 1999), although the techniques are more complex. In this regard, it may be important to recognize that satellite data are not routinely used in national forest inventories of most Annex I countries. Satellite data are not a substitute for in situ measurements; even changes in area are largely obtained though other methods. On the other hand, the use of satellite data in forest inventories is being explored (Kilpel�inen and Tokola, 1999).
Satellite data might also be important for activities other than ARD if it were decided that the locations of additional activities were to be required. In this case, high or very high spatial resolution remote-sensing data or the use of GPS information could be useful to document the locations of additional activities.
For full carbon accounting or for estimation of baselines, determining the degree of natural variability in fluxes of carbon between terrestrial ecosystems and the atmosphere may be important. Remotely sensed data can contribute to this estimation in three ways: by providing maps of land cover through time (see Section 22.214.171.124); by providing parameters for models that calculate net exchanges of gases between terrestrial ecosystems and the atmosphere; and by estimating the area affected by particularly large or frequent disturbances, especially fire and extensive pest outbreaks.
In the last 10 years, there has been tremendous progress in the scientific community's ability to simulate exchanges of carbon (and other material) between terrestrial ecosystems and the atmosphere (Schimel et al., 1995). Many of these models use remotely sensed data to establish areas of land-cover types to which different attributes are assigned (e.g., production efficiency, surface roughness, albedo) or to determine parameters more directly (Sellers et al., 1997). All of these models are research tools at this time; they would require extensive validation and further development to be used in an operational context.
Fires are one of the largest and most frequent perturbations of many ecosystems; they constitute a major source of GHGs and aerosols. Satellite data have been useful in measuring the occurrence of active fires and areas burned in the recent past (through fire scars observable from space). Daily fire monitoring can be provided with existing hot-spot detection algorithms working on thermal channel data from the National Oceanographic and Atmospheric Administration (NOAA) AVHRR and European Space Agency (ESA) ATSR sensors. Fine-resolution mapping of selected burned areas can be accomplished with data from Landsat, SPOT, and IRS satellites. These satellites carry high-performance, well-calibrated sensors that include infrared, near-infrared, and shortwave infrared bands.
All of the satellite-based observational systems that are used to assess fire occurrence and burned area require in situ observations (or models) to assess fuel loads and thereby calculate GHG emissions. There currently is no direct way to measure GHG emissions from fire solely through the use of remotely sensed data. There also is no fully operational program for remote monitoring of fires, which would require observations throughout the day on a global basis, to characterize diurnal variability (Skole et al., 1997). Nevertheless, several efforts are underway to develop operational protocols.
Other reports in this collection