Abstract
AbstractMaps of the number, size, and species of trees in forests across the western United States are desirable for many applications such as estimating terrestrial carbon resources, predicting tree mortality following wildfires, and for forest inventory. However, detailed mapping of trees for large areas is not feasible with current technologies, but statistical methods for matching the forest plot data with biophysical characteristics of the landscape offer a practical means to populate landscapes with a limited set of forest plot inventory data. We used a modified random forests approach with Landscape Fire and Resource Management Planning Tools (LANDFIRE) vegetation and biophysical predictors to impute plot data collected by the US Forest Service's Forest Inventory Analysis (FIA) to the landscape at 30‐m grid resolution. This method imputes the plot with the best statistical match, according to a “forest” of decision trees, to each pixel of gridded landscape data. In this work, we used the LANDFIRE data set for gridded input because it is publicly available, offers seamless coverage of variables needed for fire models, and is consistent with other data sets, including burn probabilities and flame length probabilities generated for the continental United States. The main output of this project is a map of imputed plot identifiers at 30 × 30 m spatial resolution for the western United States that can be linked to the FIA databases to produce tree‐level maps or to map other plot attributes. In addition, we used the imputed inventory data to generate maps of forest cover, forest height, and vegetation group at 30 × 30 m resolution for all forested pixels in the western United States, as a means of assessing the accuracy of our methodology. The results showed good correspondence between the target LANDFIRE data and the imputed plot data, with an overall within‐class agreement of 79% for forest cover, 96% for forest height, and 92% for vegetation group.
Highlights
We found that plots tend to impute to a cluster of pixels, due to similarities in the topographic and biophysical predictor variables, as clustering is not imposed by the random forests method (Fig. 5)
We compared the values of the response variables in the imputed plot data to the LANDFIRE target raster grids in order to obtain the estimated levels of agreement for the tree list
The imputation reproduced the patterns in the gridded LANDFIRE data (Fig. 6)
Summary
Geospatial data describing tree species or forest structure are required for many analyses and models of forest landscape dynamics, including estimating stocks of terrestrial carbon (Jenkins et al 2001), forest biomass (Blackard et al 2008), forest growth and mortality (Brown and Schroeder 1999, Falkowski et al 2010), national-level fire planning and risk assessment (Schmidt et al 2002), simulating continental-scale burn probabilities (Finney et al 2011), simulating wildfire intensity patterns and fuel treatment strategies (Finney et al 2007), tree species abundance and distribution (Wilson et al 2012), basal area (Wilson et al 2012), estimating timber volume (Franco-Lopez et al 2001, Muinonen et al 2001), mapping wildland fuels for simulating fire growth (Keane et al 2001), and v www.esajournals.orgOctober 2016 v Volume 7(10) v Article e01472 RILEy ET AL.simulating wildfire risk transmission from federal lands to the wildland–urban interface (Haas et al 2014). Forest data must have resolution and continuity sufficient to reflect site gradients in mountainous terrain and stand boundaries imposed by historical events, such as wildland fire and timber harvest Such detailed forest structure data are not available for large areas of public and private lands in the United States, which rely on forest inventory at fixed plot locations at sparse densities of one plot per 6000 acres (Burkman 2005). Models of geospatial forest structure have utilized various statistical methods to assign the measured plots from a sparse sample to unmeasured locations using a set of predictor variables These methods have a common goal: to take more detailed observations at relatively few locations (e.g., field plots or stand inventories) and assign their characteristics to the unmeasured locations on the landscape in order to provide seamless information about all locations. A distinguishing factor among all of these methods is whether they allow for the use of categorical predictor variables as well as continuous variables
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
