Remote Sensing Lab, Ecosystem Planning

Home | Projects | Vegetation Inventory | Forest Data Links Text or Forest Data Links Map | Inventory Data Download | Socal Mort

Vegetation Inventory - Southern California Mortality

Table of Contents

• Reports
• Regionalization Method Applied
• Renumbering of Plots
• Stratification Notes

Reports

	Inventory plot level data, Strata map, Strata acres, Inventory Tree reports
	Walker, R.; Rosenberg, M.; Warbington, R.; Schwind, B.; Beardsley, D.; Ramirez, C.; Fischer, L.; Frerichs, B. June 2006. Inventory of Tree Mortality in Southern California Mountains (2001-2004) due to Bark Beetle Impacts California Dept. of Forestry and USDA Forest Service. (pdf 15.1 Mb)
	Assessing 2001-2003 Forest Mortality in Southern California USDA-FS and CDF-FRAP. April 2006. (pdf 6.8 Mb)
	CDF-FRAP, Southern California Insect Related Tree Mortality
	Southern California Mortality 2003 Recent dead trees to a 4 inch top
	Southern California Vegetation Mortality Related Canopy Loss 1997-2003
	Southern California Vegetation Life Form Conversion Due To Mortality 1997-2003

Top of Page

Regionalization Method Applied

Regionalization Method Applied to Southern CA Mortality Data

Data Sources: 03 Aerial Sketch Maps, 97-02 change detection + 02-03 mortality detection

Polygons developed for the 02-03 vegetation layer update were used to regionalize the two sources of geospatial mortality data. Polygons were developed from pan sharpened imagery created from 2003 SPOT 5 five meter panchromatic data and 2003 Landsat 5 TM data. The Landsat data were initially acquired and used for the mortality detection. Image segmentation via eCognition software was the method used to generate the actual delineations. These polygons were used to provide larger spatial definition to the mortality pixels and to create a relationship to the existing vegetation layer. The 03 sketch mapping polygons defined the extent of the analysis area.

The mortality pixels from the combined change and mortality detection products were originally classified into low, medium and high classes. Uncertainty about the thematic precision and spatial application of these classes resulted in assigning each class a specific value. Each class represented a range of canopy cover loss and the mid-point was the assumed to be the average canopy lost for a specific mortality class. The following shows the canopy cover loss values assigned to each mortality class:

All areas within the sketch mapping mortality polygons, but not classified by the change and mortality detection pixels, were assumed to be little or no change and assigned a mid-point value of 8. Sketch mapping polygons were rasterized (5m) and combined with resampled change-mortality pixels. One acre MMU polygons from the vegetation update layer were used to regionalize these pixels using a zonal mean function in GRID. The output depicted a continuous range of canopy loss values between 8 and 86 percent. The data were vectorized to assign a unique polygon id to each group of canopy loss pixels.

The final regionalization step added information from the vegetation layer for subsequent use in developing mortality strata. Vegetation type and regional type labels were assigned to mortality polygons in conifer, hardwood and mixed stand conditions. The updated vegetation layer labels were used unless a polygon had been assigned a mortality flag code during the update process. In these cases the 1997 eveg layer label was used to ensure that pre-mortality forested conditions were included. A small portion of the analysis area did not include mortality flag codes in the updated veg data. Mortality polygons in these areas were assigned 1997 eveg labels in cases where pre-mortality labels were equal to conifer, hardwood, or mix and updated labels were equal to shrub or hardwood(mix only).

Top of Page

Renumbering of Plots

The plots had to be renumbered so that the plot numbers would be unique and the data could be used with the software parameters. The table below shows how the plots numbers were temporarily renamed. The table also shows what Forest the plots are on and the original plot name.

Top of Page

Stratification Notes

We assigned all mortality trees to salvageable dead code in the inventory, just to get the recent dead trees reported all together. We did not have the information on what was truly salvageable. It was not a value judgement, it was just an easy way for us to get it reported using our existing software systems.

We have 152 cc plots within the perimeter of the assessed area that were reviewed or visited on the ground. Of those 152, we removed from our sample 4 cc plots that fell in the 2003 FIRE area and assigned these to the FX stratum. We also removed from our sample 24 cc plots in the non-forest area (either R5_type U or X). We assigned these to the UX stratum.

That gave us 124 plots in the forested non-fire area as our forestland sample to assess the affects of the mortality event.

Of those 124 plots we have 86 in LOW <=8% loss. We tried playing around with different cutoffs for MED and HIGH, and thought the following was best for the distribution of # of plots by type, the trends in mortality basal area, and standard errors of estimates. We considerd MED to be >=9-19% loss and we considered HIGH to be >=20% loss.

We decided to lump the Hardwood types together (into stratum-HX) because there were only 2 productive hardwood plots. Also we decided to lump all hardwood together regardless of mortality class. Because when we looked at the L-M-H for the hardwood plots there was no trend and there were only 2 High plots. Perhaps we could create two hardwood class LOW and MEDIUM.

Similar thinking went into the Pinyon juniper and Subalpine. There just weren't enough plots to make a L-M-H split, and we thought these types were different enough to keep them separate from the other types.

We put the following FIR types together:

• Westside Mixed Conifer
• Eastside Mixed Conifer
• White fir
• Bigcone Douglas fir

We put the following PINE types together:

• Coulter pine
• Eastside pine
• Jeffrey pine
• Ponderosa pine

When we looked at the trends in Mortality Basal Area per Acre by stratum, the FIR classes looked good, and the MED was pretty close to the LOW. However, the trend makes sense so we kept the LOW-MED-HIGH strata.

The PINE classes were a bit questionable. The MED was pretty close to the LOW. However, the trend makes sense so we kept the LOW-MED-HIGH strata.

Top of Page

USDA Forest Service - Pacific Southwest Region
Last Modified: Thursday, 31 August 2006 at 14:05:42 EDT