Satellite images taken during certain key periods -- like stretches of drought or the flowering time for a tree species -- can be key to learning more about forest distribution, a Forest Service-led study finds.

The study, an observation of the tropical forests of the islands of Trinidad and Tobago, indicates that widely available and often free satellite imagery could help map forests in much more detail than was previously possible. This is especially important for countries required to create detailed forest maps to participate in international Reducing Emissions From Deforestation and Forest Degradation (REDD+) programs, which provide financial incentives for countries not to cut or burn down forests.

Led by Eileen Helmer, a research scientist at the Forest Service's International Institute of Tropical Forestry, the research team found that old Landsat satellite images from drought years could help fill gaps in information. Images with "noise" -- or variability that lies outside a normal forest pattern -- can give an inaccurate idea of how a particular forest is composed. Variability makes it more difficult to distinguish a forest from a plantation, field or another land use.

"There's so much more data now than there used to be," Helmer said. "We can look through imagery, and we can see things that can help make these maps much more detailed."

Use of Landsat satellites was once prohibitively expensive, said Helmer, but programs like Google Earth allow access to free images from years ago. Landsat, a program managed by the U.S. Geological Survey, takes satellite images of the Earth's landmass and maintains an image archive.

The researchers were able to identify various types of forests, including former agricultural land that had been abandoned and overtaken by natural vegetation. By observing images from the 1987 drought in Trinidad, the scientists could better identify mountainous seasonal forests, which lost more leaves than in typically dry months and could be better distinguished from the surrounding non-forested areas.

"Even in a normal dry season, it looks the same," Helmer said. "The difference is too subtle to detect with Landsat imagery."

The study was published in Forest Ecology and Management.