Detailed maps of tropical forests are within reach: forest tree communities for Trinidad mapped with multiseason Landsat and Google Earth

Eileen H Helmer, USDA Forest Service, ehelmer@fs.fed.us (Presenter)
Thomas S Ruzycki, Colorado State University, tom.ruzycki@colostate.edu
Jay Benner, Colorado State University, jay.benner@colostate.edu
Shannon M Voggesser, Colorado State University, shannon.voggesser@colostate.edu
Barbara P Scobie, Trinidad and Tobago Forestry Division, pearldraw@yahoo.com
Courtenay Park, Trinidad and Tobago Forestry Division, cpark24@yahoo.com
David W Fanning, Colorado State University, david@idlcoyote.com
Seepersad Ramnarine, Trinidad and Tobago Forestry Division, ramnarine@hotmail.com

Detailed maps of forest types are needed for REDD+ carbon accounting and biodiversity conservation, but spectral similarity among forest types; image cloud and scan-line gaps; and scarce vegetation plots complicate producing detailed maps from moderate resolution satellite imagery. We tested if : 1) floristic classes of tropical forest trees can be mapped with multiseason, multidecade, gap-filled Landsat by judicious combination of field and remote sensing work; and 2) synthetic multiseason Landsat imagery improves results.

We produced multidecade, multiseason cloud- and scan-line-gap-filled Landsat imagery with regression-tree normalization. Reference data came from field data, multiseason high-resolution imagery including that on Google Earth, and Landsat archive imagery from two epochs including phenologically unique dates. We used supervised decision tree classification to map floristic tropical forest classes and land cover.

We discovered that given a set of floristic tropical tree communities and general knowledge of their spatial distribution, features in multiseason and fine resolution reference imagery can allow discrimination among adjacent classes based on inundation or canopy attributes including deciduousness, flushing, flowering, or structure, allowing the extensive training data collection needed to map floristic classes with noisy, gap-filled multiseason Landsat imagery.



Synthetic multiseason Landsat imagery significantly improved floristic class discrimination (by 14-21% for deciduous, 7-36% for semi-evergreen, and 3-11% for seasonal evergreen associations, and by 5-8% for secondary forest and woody agriculture). Imagery from climate extremes like severe drought increased accuracy the most. Geology is an important determinant of tropical forest species composition and structure. Here we learned that the xerophytic rain forest of Tobago is closely associated with ultramafic geology, helping to explain its unique physiognomy. At the same time, seasonal spectral patterns in multiseason Landsat imagery can have more spatial detail than maps of environmental variables and can be more useful when mapping tropical forest tree communities with Landsat.

Presentation Type:  Poster

Session:  Poster Session 1-B   (Tue 4:30 PM)

Associated Project(s): 

• Related Activity or Previously Funded TE Activity

Poster Location ID: 43