Document: RIC-3-4-5

Lessons learned in applying a satellite-driven model (3-PGS) to map forest growth capacity across southwestern Oregon

WARING, R.H.* ¹, N.C.COOPS ² and J.J.LANDSBERG ³

Oregon State University, Corvallis, OR 97333 USA ¹
CSIRO Forestry and Forest Products, Victoria, Australia ²
22 Mirning Crescent,Canberra, A.C.T., Australia ³

Abstract:
Across a 54,000 km² area in southwestern Oregon we mapped forest growth capacity using a process model (3-PGS) at 1 km² resolution. Model predictions of growth capacity closely matched those measured (r² =0.8) where reliable data on soil fertility and soil water storage capacity were available. Across the region, however, we found broad scale soil maps were inadequate to provide good estimates of fertility. As a result, no benefits were gained in estimating light interception by vegetation with satellite sensors at finer resolutions than 1 km². On the other hand, finer resolution sensors (such as Landsat Thematic Mapper) permitted discrimination and mapping of 14 major forest types, for which model estimates of growth potential closely agreed with those measured (r²= 0.85) with similar standard errors as those recorded at more than 700 surveyed sites by Federal agencies. Mean monthly spatial coverages of temperature extremes and precipitation linked to digital elevation models proved adequate for extrapolation and conversion into solar radiation, vapor pressure deficits, and frost frequency at 200 m resolution. Remotely sensed estimates of canopy nitrogen concentrations (or chlorophyll) would be most beneficial in improving spatial estimates of forest growth capacity with this and related process models.

Keywords: process model, landscape ecology, G.I.S., remote sensing, forest growth capacity

This abstract is being presented at: 2:00 PM in session:
Oral Session #64: Remote Sensing.