Giant kelp forests are amongst the most productive ecosystems on Earth. They provide food and shelter for a highly diverse community of fish, invertebrates and under story algae. Kelp forests are also highly dynamic ecosystems. Maximum growth rates for giant kelp (Macrocystis pyrifera) fronds can exceed a foot a day while entire kelp forests can be wiped out by a single winter storm.
Remote sensing data have long provided a method for integrating field point and transect data sets from local, to regional, to global scales. Currently, we use high spatial resolution, multi-spectral imagery from the SPOT-5 satellite to integrate the SBC-LTER diver observations to regional scales. SPOT is an acronym for Système Pour l’Observation de la Terre, a French series of high resolution Earth remote sensing satellites. Three SPOT satellites are currently in operation above the earth. Giant kelp, like most terrestrial plants, has a spectral signature characterized by high near-infrared reflectance and much lower reflectance in the green and red spectral regions. SPOT-5 provides multispectral imagery with ability to distinguish these spectral bands at a 10 meter spatial resolution appropriate for assessing changes in the giant kelp forests.
The process of assessing kelp canopy coverage begins with an atmospheric correction to the green, red and near-infrared bands of SPOT-5 multi-spectral imagery using the dark-pixel method. The resulting corrected multi-spectral imagery is transformed using a principal component analysis, which results in two dominant spectral modes of variability. The first mode shows positive contribution from all three spectral bands and accounts for issues of sediment loads and variations in the atmospheric correction procedure.
The second mode shows spectral patterns consistent with the reflectance signals expected for giant kelp and is used to assess kelp covered pixels. The resulting cover observations compare well with annual aerial survey results published by the California Department of Fish and Game (CDFG; from the fall of 2004) on an administrative bed scale (r2 = 0.92 with a slope between the two cover measures that is insignificantly different from 1). This comparison provides a good first order validation of our SPOT-5-derived determinations of giant kelp cover.
Examples of SPOT-5 derived kelp cover distribution are shown in Figure 1. The bright green pixels are classified kelp cover. The upper panel of Figure 1 shows a regional perspective of kelp forest cover for November 11, 2006 while the lower four panels shows the time evolution of kelp cover at the local scale at the SBC-LTER site at Arroyo Quemado (AQ). These maps provide an integrated assessment of kelp canopy area and changes in kelp canopies from the scale of diver transects (Figure 1, red dots) to local (Figure 1, lower panels) and regional (Figure 1, upper panels) scales. Sequential images illustrate how the morphology of kelp forests change in time (lower panel). Here, the kelp forest growing season at AQ is observed from April to June (increase of 5.7 hectares), followed by the gradual deterioration of surface canopies from June to November (decrease of 7.9 hectares) due to senescence and disturbance by surface waves. These images clearly show the dynamic nature of giant kelp forests.
Areal cover is one measure for a kelp forest; however, assessments of kelp biomass are more useful for understanding and modeling ecosystem interactions. The SBC-LTER monthly diver transect surveys are made over a 160 m2 region (~16 SPOT-5 multi-spectral pixels) and provide estimates of kelp plant biomass per unit area among many other parameters. The location of the AQ diver survey transect is indicated by the red pixels in the lower panels of Figure 1. Comparison of diver estimates of kelp biomass and SPOT-5 determinations of the Normalized Difference Vegetation Index (NDVI; calculated using the green and nearinfrared bands) from all three SBC-LTER sites shows an excellent correspondence (r2 = 0.71; n = 37). This provides a robust empirical algorithm for estimating kelp biomass from SPOT-5 multi-spectral imagery on local to regional scales. The four subscenes shown in the lower panel of Figure 1 document two-fold changes in kelp standing crop in less than seven months at this site. Our methodologies for assessing kelp cover and biomass using SPOT imagery are detailed in a manuscript recently submitted for publication in Remote Sensing of the Environment (Cavanaugh et al. submitted).
The ability to remotely characterize giant kelp cover and biomass on local to regional scales opens many new doors for study of kelp ecosystem dynamics by researchers at SBC-LTER. Since January 2006, we have been acquiring SPOT-5 10 m multispectral and 5 m panchromatic imagery on a bimonthly basis for all of the Santa Barbara coast and the northern Channel Islands. With recent NASA support, we plan to use these processed images to assess the spatial and temporal dynamics of giant kelp forests on a regional scale, evaluating the growth, persistence and disturbance in individual forest stands. This assessment will be used in conjunction with studies of kelp spore dispersal and disturbance frequency to build on previous efforts by SBC-LTER researchers to develop a spatially realistic metapopulation model of Southern California kelp forests, which was previously limited by the coarse resolution of historical kelp biomass time series.
This project demonstrates the value and promise of remote sensing data sets for scaling-up detailed ecological assessments to describe larger scale regional ecosystem dynamics. We expect that many researchers within the LTER community use SPOT or similar (i.e., Landsat) imagery in their work. Unfortunately, there is currently a dearth of high-quality Landsat imagery and SPOT data has remained difficult to acquire due to its high cost. In an effort to address these issues, UCSB, partnering with Terra Image USA, has established the SPOT at UCSB program to provide U.S. educational institutions SPOT imagery of the U.S. and Canada at a reasonable cost (more than 85 per cent off retail cost for member institutions). The SPOT at UCSB program provides members access to our growing archive of more than 1.4 million SPOT scenes as well as the ability to task the suite of SPOT satellites for their particular study sites. Imagery from most of the LTER sites are being tasked already and we are interested in additional suggestions of tasking sites. Details of the SPOT at UCSB program (including our online catalog) are available at www.spot.ucsb.edu or by calling 805-893-8475.
David Siegel, Kyle Cavanaugh, Brian Kinlan and Daniel Reed (SBC)
Enlarge this image
