Background
For DNR to succeed in developing a forest assessment and treatment framework, a regularly updated set of current forest condition layers is needed. Further, to monitor progress, it is critical to produce a new current forest conditions layer on a regular biannual interval.
The purpose of this agreement is to implement a project, which involves production of forest condition spatial datasets from remotely sensed imagery, and coordination of project partners to ensure successful integration of products and research. The methodology to produce the datasets will build off demonstrated successful approaches used by the DNR and other project partners and incorporate newly available analytical techniques as they evolve.
Our Work
Ground Model
All metrics developed from the digital aerial photogrammetry (DAP) point clouds require a ground model so that elevations of the points can be normalized to height above ground. The accuracy of the ground model has an impact on the accuracy of DAP metrics, so it is very important to use the highest accuracy ground model possible.
For this project, a custom statewide ground model was developed. The DNR provided a LIDAR-derived ground model for much of the state. It combines vendor produced LIDAR ground models into a single raster that crosses LIDAR acquisition boundaries. In areas where the DNR ground model did not have data, alternate data sources were used. The DNR LIDAR portal has additional LIDAR ground models that were not part of the DNR’s dataset. These were added where possible, and any remaining areas without ground data were filled in using the USGS 3DEP 1/3rd Arc-Second Seamless DEM.
Digital Aerial Photogrammetry (DAP)
The DAP point clouds were derived from NAIP imagery, and were available statewide for the 2017 imagery and in select study areas for the 2015 and 2019 imagery. DAP data for the three years was processed for individual plots using Fusion cloudmetrics and statewide using Fusion gridmetrics to produce plot-level metrics and statewide rasters.
Plot Database
A significant portion of this project was the development of a relational forest inventory plot database containing standardized field inventory attributes, DAP metrics, and phsyical and environmental attributes, that can be used for modeling. The database is a combination of forest inventory plots provided by the USDA Forest Service (USFS), the Washington Department of Natural Resources (DNR), and the University of Washington (UW). The plot database is the source of all inventory model parameters.
All plots were measured with the intention of using them with DAP and/or LIDAR to develop remote sensing based forest inventory. As such, they were located with survey-grade GPS systems. The plot sizes vary from organization to organization and primarily range from 1/10th acre to 1/4 acre plots with various sizes of nested subplots for small tree measurements. Measurement protocols and specific attributes measured also varied. A standardized set of field attributes that were commonly measured for most plots was developed.
The flexible database structure allows for the addition of new plots, re-measurement of existing plots, the addition of further attributes, the ability to provide or restrict access to particular sets of plots, selection of particular sets of plots or plot attributes for specific models, and the testing of varied modeling approaches.
It is hoped that this database can be made publically available through an online service, with appropriate controls in place, so that any researchers interested in developing remote sensing based forest inventory models and modeling approaches will be able to use it.
Linear Regression Models
Development of linear regression models for selected forest inventory attributes.
Data Products
Development of statewide model prediction and forest structure class rasters.
Products
Reports
Project Report
ArcGIS Image Services
Forest Structure Model Image Services
Metadata
Image Service Metadata
Database Metadata
Model Diagnostic Plots
