Potentials of Public and Commercial Satellite Data for Total Tree Counts

Public and commercial satellites’ imaging-non-imaging data have been used for global climate variation studies, mainly measuring impacts on vegetation; the continuous advent of new sensors and platforms makes it possible and affordable to develop a methodology for total tree counts as indicators to macroclimatic variations in the local settings. This paper is the product of the ongoing USDA-funded research; earlier research presentation by the corresponding author estimated total tree counts on the Earth as 1.5 trillion and 3.37 trillion (assuming one tree per 9 and 4 square meters, respectively) for the presentation at the SAF National Convention 2013). Recent publications claim that there are 3+ trillion trees on the Earth (FAO, Yale University, and other reports). Precise information on the number and distribution of tree species is essential for managing plantations, urban planning, parks and recreation, and precision farming in the surrounding areas. During our earlier research, we discovered treetop counts or crown area measurements, using image classification techniques, were complex, even ambiguous in dense deciduous two-storied stands, e.g., Bernheim Research Forest, using high-precision and resolution airborne 4-band imageries (the commercial software packages were not able to discern types of vegetation: spectral translation and derived indices). We plan to overcome the above-discussed problem by optimizing ground sampling data at grids first (1 km x 1 km) and then to plats (15-m x 15-m) using GNSS, UAS multispectral/hyperspectral/lidar ground truthing data for enhanced classifications of commercial high-resolution (30 to 50-cm) multispectral imageries. The study area of the research is 7,138 square kilometers, covering seven contiguous counties in the Appalachian region of Eastern Kentucky with a high record of surface mining, approximately 7,727 (active and inactive) mines. Our earlier research at this site utilized public domain satellite imageries and other existing geospatial data (USGS, NASA, USDA, and Kentucky Geoportal) from 1986 to 2016 to evaluate the patterns of surface mining and reclamation in the study region, and the research is ongoing. This paper will present outcomes of the research performed to accomplish the following objectives: one (1) evaluate the suitability (accuracy and cost-effectiveness) of commercial high-resolution multispectral imageries for total tree counts at county levels, two (2) analyze possibilities of integrating various bands of Landsat 9 and Sentinel 2 imageries with the multispectral imageries of WorldView-3 and Planet satellites to increase accuracy and repeat cycles.
Key Words: Total Tree Count, Treetop, Tree Canopy, Commercial Satellites, Image Classifications, Appalachian