Mapping Bull Kelp Canopy in Northern California Using Landsat to Enable Long-term Monitoring
Abstract
While field surveys and aerial imagery can provide detailed biological data, satellite remote sensing enables consistent, long-term imagery over a large spatial scale. Such imagery enables us to track changes in vegetation over time, monitoring how environmental conditions have affected ecosystems. Long-term time series from satellite imagery have been developed for giant kelp in Southern California, but there have been no such surveys for bull kelp. Bull kelp is a Northern California brown algae species that supports fisheries and provides a habitat for keystone species. Using Multiple Endmember Spectral Mixture Analysis (MESMA) of Landsat imagery, we test bull kelp classification methods using California Department of Fish and Wildlife (CDFW) aerial bull kelp surveys for validation. We find that a 7-endmember MESMA model records the lowest Root-Mean-Square Error (RMSE: 0.2352; 2.30%) during CDFW validation, but it shows no statistical difference (p=0.5071) from a Python scripted method (0.2652; 2.59%), which we use here to decrease user and computer processing requirements. Compared to CDFW aerial surveys, Landsat methods record less overall kelp coverage (roughly 20%) and are less accurate for estimating subsurface kelp coverage (NRMSE: 0.6437). Nonetheless, Landsat enables broad and cost-efficient coverage of bull kelp to supplement existing methods. We construct a 36-year Landsat time series of Northern California bull kelp coverage and compare canopy reductions and recovery to ENSO and other basin-scale oscillations. We find that bull kelp decreased sharply during historically strong ENSO events (correlation of -0.3522 with 1-year lag; p=0.0444) but also that it recovered quickly following these events (correlation of 0.4075 with 3-year lag; p=0.0186). Since 2014, though, bull kelp has experienced a prolonged decline, with canopy coverage suppressed for four years below one standard deviation of the time series mean, which calls into question its resilience in a changing world.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFMOS31B..13F
- Keywords:
-
- 4813 Ecological prediction;
- OCEANOGRAPHY: BIOLOGICAL;
- 4815 Ecosystems;
- structure;
- dynamics;
- and modeling;
- OCEANOGRAPHY: BIOLOGICAL;
- 4264 Ocean optics;
- OCEANOGRAPHY: GENERAL;
- 4556 Sea level: variations and mean;
- OCEANOGRAPHY: PHYSICAL