The ecological and economic impacts associated with invasive species are of critical
concern to land managers. The ability to map the extent and severity of invasions would be
a valuable contribution to management decisions relating to control and monitoring efforts. We
investigated the use of hyperspectral imagery for mapping invasive aquatic plant species in the
Sacramento-San Joaquin Delta in the Central Valley of California, at two spatial scales. Sixty-four
flightlines of HyMap hyperspectral imagery were acquired over the study region covering an area of
2,139 km2 and field work was conducted to acquire GPS locations of target invasive species.We used
spectral mixture analysis to classify two target invasive species; Brazilian waterweed (Egeria densa),
a submerged invasive, and water hyacinth (Eichhornia crassipes), a floating emergent invasive. At the
relatively fine spatial scale for five sites within the Delta (average size 51 ha) average classification
accuracies were 93% for Brazilian waterweed and 73% for water hyacinth. However, at the coarser,
Delta-wide scale (177,000 ha) these accuracy results were 29% for Brazilian waterweed and 65% for
water hyacinth. The difference in accuracy is likely accounted for by the broad range in water turbidity
and tide heights encountered across the Delta. These findings illustrate that hyperspectral imagery is
a promising tool for discriminating target invasive species within the Sacramento-San Joaquin Delta
waterways although more work is needed to develop classification tools that function under changing
environmental conditions.
| 
