Get reliable rangeland science

Tree density has been increasing due to the exclusion of fires over years in the American Southwest and ecological restoration treatments are aimed to reduce such increasing fuel for extreme fire events and to increase understory biodiversity. This study provides an overview of remote sensing techniques to estimate tree density in the Collaborative Forest Landscape Restoration Project (CFLRP) area of Jemez Mountains, New Mexico. We used reflectance values from individual Landsat 8 bands (bands 4, 5, 6, and 7) and vegetation indices (NDVI- Normalized Difference Vegetation Index, DVI- Difference Vegetation Index, SR- Simple Ratio, and ND57- Normalized Difference bands 5 and 7) to estimate the density of trees. Models including multiple predictor variables derived from remote sensing data performed best (R� = 0.46 - 0.95) for each vegetation types rather than using a single model for the entire project area. The area dominated by non-conifer tree species (aspen) showed the highest value of R�. The combined tree density map shows most of the grasslands without trees and recent wildfire areas with less (1-100 trees/hectare) to no trees. About 50% of the study area is distributed with medium density of trees (101-1000 trees/hectare). A small fraction of the areas covered by spruce-fir, pinyon-juniper, and ponderosa pine is distributed with high density of trees (>1000 trees/hectare). Test of accuracy of the models shows 77% of estimated density values are in the same density classes as the observed values and demonstrates the effective use of reflectance values from satellite images while predicting tree density in the areas dominated by coniferous forest and complex topography. Project managers could use this baseline info to identify the areas in need of restoration treatments.