An Improved Method For Monitoring National And Global Deforestation

The researchers will discuss images and quantitative measures they have derived of the amount and distribution of forests in North America and some other areas of the world on Wednesday, May 30, at the American Geophysical Union's spring meeting in Boston. These forest measures were produced by analyzing data from Terra’s MODIS (Moderate Resolution Imaging Spectroradiometer) instrument using Maryland’s continuous fields method for tree cover mapping.

"It is essential that we know accurately how much forest cover there is on our planet to help us conserve what is left and to make the best use of forest resources," said John Townshend, a professor in the university’s department of geography and the UM Institute for Advanced Computer Studies. "The FAO [UN Food and Agriculture Organization] has recently made an enormous effort to derive statistics for forest cover for the world based on reports from the countries themselves. But these figures usually are simply not comparable from country to country, and it is abundantly clear that for many countries the FAO estimates are highly inaccurate."

The FAO released a controversial report earlier this year that said global rates of forest loss decreased during the 1990s. However, many scientists and environmentalists have criticized the methods the FAO used in assessing the extent of the world’s forestation and its change over time. Maryland researchers say the FAO analysis has many inherent scientific and political problems because of its use of a number of different definitions of what constitutes a forest. These definitions were based on both intended land use and land cover estimates. One result of this classification approach was that in some cases areas from which all the trees had been harvested were still counted by the FAO as forest because that was the land’s "intended use."

In addition the latest FAO report and its two earlier ones aren't comparable because all three use different methodologies. Thus they are of relatively little use in creating a long-term record of global changes in forest cover.

"Our work and that of many other scientists show that the way forward to more useful estimates of forest cover lies with the use of remote imaging data," said Ruth DeFries, an associate professor in the department of geography and the university's Earth System Science Interdisciplinary Center. "Remote sensing offers a globally repeatable and verifiable methodology which doesn't have the problems of bias that are inherent when forest estimates are provided by various agencies or institutions within each country."

DeFries said the advantages of this approach can be seen in the university's work with the MODIS instrument on NASA’s Terra satellite. Using data from this instrument and the continuous field method she co-developed, DeFries and other Maryland researchers have produced a series of new information products such as one that shows the percent tree cover over the conterminous United States at a spatial resolution of 250m. (For further information on these products see http://glcf.umiacs.umd.edu.)

DeFries noted that their latest work with the MODIS instrument builds on their previous work using 1 km resolution data from the AVHRR, a polar orbiting satellite of the National Oceanic and Atmospheric Administration that has been in orbit for two decades. Landsat 7 – whose science team is led by Maryland department of geography chair Samuel Goward – is another important satellite in the current fleet of spacecraft for remote sensing of land cover. According to DeFries these satellites, together with others planned as part of NASA's Earth Observing System, will enable scientists to identify land cover change and to more accurately determine forest extent in areas with fragmented forests.