Abstract:
Protracted drought in the southwest U.S. has had significant impacts on the
region’s keystone ecosystem, the pinyon-juniper (PJ) woodlands. Drought conditions
in 9 of the last 10 years, exacerbated by extreme drought in one year, stressed the
pinyon pine (Pinus edulis) trees, making them highly vulnerable to native ips beetles
(Ips confusus). Current global climate predictions include further increases in
temperatures and more extreme climatic events such as drought which will likely
cause even more large-scale disturbances such as shifts in ecotones and increased
insect outbreaks. This study investigated the utility of the Landsat satellite sensor for
predicting percent dead tree cover in the PJ ecosystem in Arizona and New Mexico
by relating field measurements of percent tree cover with Landsat spectral data.
Direct change modeling and state model differencing (Healey et al. 2006)were
compared in the two study areas in terms of RMSE for predicting percent dead cover.
Direct change modeling outperformed state model differencing in both AZ and NM,
with RMSEs of 4.2% and 7.3%, respectively. Landscape patterns were investigated
in maps of dead cover made from the models. Consistent with expectations of the
mechanism of the mortality, percent mortality on southwest aspects was higher than
on northeast aspects. Percent dead cover model predictions were also validated with
an independent data source (Forest Health Monitoring overflight data) in AZ and NM
with overall agreement of 80 and 76%, respectively.
