Assessing biomass and forest area classifications from modis satellite data while incrementing the number of FIA data panels
Our objective was to determine at what level biomass and forest area obtained from 2, 3, 4, or 5 panels of forest inventory data compares well with forested area and biomass estimates from the national inventory data. A subset of 2605 inventory plots (100% forested, 100% non-forested) was used to classify the land cover and model the biomass in South Carolina. Mixed plots containing both forest and non-forest conditions have been excluded. Forest inventory data have been further subdivided into four datasets containing the most recent 5, 4, 3, and 2 panels of data. Separately, each of these four datasets was used in a decision tree classification process applied to MODIS satellite data (250-meter resolution) and ancillary data to classify the land cover and model the forest biomass. The satellite, ancillary, and plot data have been subdivided into three mapping zones 54, 58, and 59 for processing in See5 and Cubist software. Classification results for trials with 2, 3, 4, and the entire cycle of 5 panels show that overall classification accuracy for the percent of pixels correctly classified (%PCC) increased from 72.3% to 79.2%. Comparison between classified forest area with 4 and 5 panels and the inventory forest area shows less than 2% difference. The forest/non-forest single layer classification from each trial was used to mask out non-forested areas for the forest biomass classification. Accuracy of modeled forest biomass was compared with plot data estimates of forest biomass. Biomass obtained from Cubist models with 3, 4, and 5 inventory panels when compared to the biomass from the published plot data estimates show a difference of 29.2%, 11.0% and respectively 1.6%.