Mutli-temporal Imaging Spectroscopy Analysis for the Identification of Coniferous Forest Mortality Related to Drought Stress in the Central Sierra Nevada, California

Abstract:

The ongoing drought in California has had a significant impact on the vegetation communities of California. As a result of the drought, there has been a notable increase in forest mortality throughout the state. In this presentation Airborne Visible / Infrared Imaging Spectrometer (AVIRIS) imagery acquired for the HyspIRI Preparatory Mission over the western Sierra Nevada Range in 2013 and 2014 was used to quantify the mortality of conifers in 2014. Data products provided by NASA’s Jet Propulsion Lab (NASA-JPL) were re-sampled to a common 15meter pixel resolution, co-registered, and geo-referenced. Ecological cover type was first assessed using the random forest machine learning technique with training data produced from AVIRIS summer 2013 imagery and comparison with high-spatial resolution World View-2 imagery. Then, in areas identified as being primarily composed of needle-leaf tree cover, the change in fractional change in green vegetation cover was assessed using Multiple Endmember Spectral Mixture Analysis (MESMA) in fall 2013 and fall 2014 AVIRIS images. The source spectral library for the MESMA endmembers was created from AVIRIS-Next Generation (AVIRIS-NG) images taken over Sierra National Forest in 2014. False positives were further reduced using a spatio-temporal filtering approach. Final accuracy of the modeled areas of conifer mortality were assessed by comparison with 2015 WorldView-2 and WorldView-3 imagery over the study area, as well as with recently acquired field data within the southern Sierra Nevada. Early results support the need for increased fidelity data for providing timely information on ecosystem dynamics to land management agencies.