Look for complete geospatial metadata in this layer's associated xml document available from the download link * Metric Name: Density – Large Trees * Tier: 2 * Data Vintage: 06/2020 * Unit Of Measure: Percent live trees per pixel * Metric Definition and Relevance: Large trees are important to forest managers for multiple reasons: they have a greater likelihood of survival from fire; they are an important source of seed stock; they provide vitally important wildlife habitat; and they contribute to other critical processes like carbon storage and nutrient cycling. Large trees are often the focus of management in order to protect existing ones and to foster recruitment of future ones. In consultation with National Forests, “large trees” have been designated in three categories, 24”-30”, 30”-40”, and >40”” dbh. The data provided are an estimate of density of trees (in each dbh class) within a pixel. * Creation Method: To determine the cutoff for large trees, we developed an allometric equation to predict tree diameter as a function of height. We selected data for plots located in the Northern California region from the USDA Forest Inventory and Analysis program (FIA) for California (FIA DataMart 2023; California 2022 database; ver. 9.0.1). We included trees that met the following criteria: alive; crown class code of open-grown, dominant, or co-dominant; diameter at breast height (DBH, breast height = 4.5 ft) at least 1 inch; and height (HT) at least 5 feet. To minimize the impact of outliers, we trimmed the maximum tree height to the 0.995th percentile. These selection criteria yielded 71,412 trees. We used an information theoretic approach to select the best allometric model (Burnham and Anderson 2002). We evaluated three alternative functions: : linear, power, and saturating. The criteria for model selection were based on the Akaike Information Criterion (AIC). For this set of 3 potential models, we calculated the difference in AIC between every model and the model with the lowest AIC (ΔAIC). The best allometric model was a power function ( ΔAIC = 58.7) where: DBH (in ) = 0.2071*HT(ft)1.0296 The root mean square error on the DBH prediction was 5.8 in and the pseudoR2 = 0.75. Predicted diameters from heights are summarized in Table 3. ~~~~ The Aggregate Tool (ArcGISPro) was run on California Forest Observatory (CFO) 10 metercanopy height pixels for the following ranges to create three rasters. Aggregate resamples a raster to a coarser resolution (10m to 30m in this case) based on the sum of cells per neighborhood. * 24in - 30in * 30in - 40in * greater than 40in If NoData values existed for any of the cells that fell within a larger cell on the output raster, the NoData values were ignored when determining the value for output cell locations. This method assigned the number of pixels per 30m (900m2) cell. Resultant values of 1 through 9 were converted to percent. All background values were calculated to equal 0, meaning 0% large tree existence. References Burnham, K.P., and D.R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd ed. New York, Springer-Verlag. FIA DataMart. 2023. USDA Forest Inventory and Analysis DataMart. * Credits: California Forest Observatory (Salo Sciences), 2020