Feature ArticlesExamining Change in the Marine Environment by Satellite
By Gary Borstad
Senior Remote Sensing Analyst
ASL Environmental Sciences Inc.
Tracking change in remote parts of the globe is a challenge that is increasingly well met by satellite imaging. This is particularly true of the marine environment, where change occurs over a wide range of scales, both temporal and spatial. While traditional in-situ observations are unavoidably constrained to point or occasionally linear measurements in space (ship transects) or time (moored buoys), satellite imaging offers synoptic coverage on a regular basis, at resolutions that can be on the order of one kilometer and one day.
Earth observing satellites have been in operation since the early 1960s. The earliest satellites were designed chiefly for weather applications, but beginning in the 1970s, land and ocean observing satellites appeared with the launch of the first of the Landsat series in 1972 and the Coastal Zone Color Scanner (CZCS) in 1978. Passive microwave imaging began in 1972 with Nimbus 5 and the capability of monitoring sea ice. Since then, newer generations of sensors have increased the capabilities of these early instruments with better sensitivity, higher resolution and improved algorithms, as well as expanded the array of Earth-observation applications.
This now 30-plus year archive of satellite imagery enables users to examine long-term change and variability in the marine environment. In some cases the same sensor or series of sensors (e.g., Landsat from 1972 to the present and the Advanced Very High Resolution Radiometer [AVHRR] from 1985 to the present) have provided continuous or near-continuous datasets over their lifetimes.
In other cases, different sensors have evolved with improving capabilities to monitor specific parameters (e.g., CZCS, the Sea-Viewing Wide Field-of-View Sensor [SeaWiFS] and the Moderate Resolution Imaging Spectroradiometer). For the latter in particular, data providers have been careful to address intercalibration issues and to correct for these where possible.
Merged datasets now exist for some important parameters such as sea-surface temperature (SST) and ocean chlorophyll in an attempt to provide high-quality datasets for temporal analysis.
This article presents three examples of long-term time-series analysis that examine vegetation change in the coastal zone, scales of SST and chlorophyll variability in the northeast Pacific, and the varying regional effects of seasonal, annual and climate change on ice in the western Arctic.
River Delta Vegetation Losses
Anecdotal reports from Inuvialuit hunters from the Anderson River area in the Canadian Northwest Territories have described declines in vegetation on the river delta over the last two decades. This delta, part of a federal migratory bird sanctuary, is among the most important areas for bird breeding and summering in the Northwest Territories, and loss or degradation of habitat is an important conservation issue.
To quantify and confirm these reports, images acquired by Landsat satellite series sensors were used to assess vegetation changes over the 31-year period from 1972 to 2003. The Normalized Difference Vegetation Index (NDVI), a quantitative index of chlorophyll-containing plant biomass or vegetative cover, was calculated for each image in the time series.
Long-term habitat losses were clearly apparent in the river delta. In the outer islands, an important nesting area for lesser snow geese (Chen caerulescens caerulescens) and black brant (Branta bernicla nigricans), mean NDVI declined by 37.5 percent, with losses of habitat area of 45 percent. In the western delta, these declines were 11.9 percent and 18 percent, respectively. Visual observations indicate that dead shrubs and barren mud flats now occupy significant parts of the delta that were apparently well vegetated by willows, grasses and sedges, and other herbaceous plants in the past. Soil samples taken from areas with degraded habitat are highly saline, suggesting saltwater flooding as a potential cause of vegetation loss, most likely due to a climate effect. These changes could have an impact on the migratory bird populations of this area.
In contrast to the situation in the delta where vegetation loss was prevalent, the analysis showed that the NDVI for upland tundra increased by 21.1 percent between 1972 and 2003.
There was also a significant correlation between NDVI and spring temperature found for the uplands area. These findings are in agreement with other studies carried out since the 1980s in Siberia and Alaska that used AVHRR data to describe a 25-year period of more or less constant increase in NDVI. However, the longer Landsat time series used here reveals the effects of a cold period in the 1970s that was missed in the shorter AVHRR time series. Remote sensing time series are only now becoming long enough to adequately capture the extreme interannual variability in regions such as the Arctic.
Northeast Pacific Ocean
The west coast of North America is an oceanographically complex area, dominated by the downwelling Alaska Current regime in the north, the upwelling California Current regime in the south and transitional waters off southern British Columbia. There is a strong seasonality to the upwelling/downwelling conditions, with a summer northward shift of the Alaska Current regime that results in increased upwelling off British Columbia and concomitant effects on nutrient availability and primary productivity. The region is also subject to significant freshwater inflow, as well as episodic events such as El Niño and decadal cycles such as the Pacific Decadal Oscillation (PDO).
Satellite observations of SST from AVHRR Pathfinder for 1985 to 2008 and chlorophyll from SeaWiFS for 1998 to 2008 were used to capture this spatial and temporal variability. Average annual SST patterns were similar throughout the Alaska Current dominated zone, but there was considerable regional variation of both timing and amplitude in the chlorophyll cycles. The Strait of Georgia/Johnstone Strait zone was noticeably productive, with two to three times the average standing stock of chlorophyll in other zones. This oceanographically distinct area lies between Vancouver Island and the mainland and as such is removed from some of the outer coast influences, but it is strongly affected by outflow from the Fraser River.
As well as spatial variability, an examination of SeaWiFS chlorophyll for the years 1998 to 2008 revealed considerable interannual variability. The years 2007 and 2008 were unusually productive in both the inner straits and along the outer west coast of Vancouver Island, and they also lacked the characteristic spring/fall seasonality associated with these zones. In contrast, the year 2005 was a poor chlorophyll year on the west coast, with very low spring production and lower than average fall production; in the straits, the same year exhibited a strong, early spring bloom but low fall chlorophyll levels. Evidence of higher order effects were seen up the food chain: 2005 was considered to be a bad year for west coast salmonids, but Georgia Strait salmonids entering the sea early in the spring performed well.
With a 24-year archive, satellite SST extends far enough back in time to begin to show decadal scale variability, including trends related to ocean indices like the PDO and the Oceanic Niño Index, which can have far-reaching effects. The 1997 to 1998 El Niño event shows clearly in the West Coast SST anomaly, as do the alternate cooling (1999 to 2002, 2008) and warming (2003 to 2006) phases of the PDO.
Arctic Ice Concentrations
Sea ice is an essential component of the environment in the Arctic, with profound effects on physical oceanography, biological productivity and marine mammals. There has been a considerable focus on the sharp decline in summer sea ice coverage over the past decade, with record low values recorded in 2002 and 2005 followed by a further reduction of nearly 25 percent in the summer of 2007. The long-term trend amounts to a reduction in sea ice coverage of just more than 11 percent per decade in late summer. However, Arctic ecosystems are defined and best understood on regional scales rather than over the full Arctic Ocean region.
Since 1968, Environment Canada's Canadian Ice Service (CIS) has compiled digital ice charts for the western Arctic at weekly time intervals from June to December and with a longer time interval in winter and spring. These ice charts are derived from multiple data sources, including airborne and ship-based ice observers and satellite datasets like Radarsat-1 and 2, which provide all-weather coverage at high spatial resolution.
An examination of the CIS time series shows large regional differences in long-term trends in sea ice along the fabled Northwest Passage, with much larger declines occurring in the west by comparison with the central and eastern subregions.
Also of interest is the large degree of interannual variability. In many subregions, the total change in ice concentration over the 41-year data record was less than the standard deviation. The subregions and times of year when the long-term trend was comparable to or exceeded the standard deviation of total ice concentration were limited to late summer in the Alaskan Beaufort Sea, early summer in Amundsen Gulf and midsummer in Viscount Melville Sound. For all other subregions and dates, the long-term trend was considerably less than the 41-year variability.
This high degree of variability by subregion indicates that the changing Arctic ice climate, which is widely believed to be associated with climate change, is occurring in a very uneven manner by subregion and time of year.
The increasing availability of satellite data over the past 30 years has vastly improved understanding of the ocean and remote areas like the Arctic. From the growing time series, scientists are becoming able to resolve spatial and temporal scales of change and to identify anomalous events and their varying local effects, as shown in the examples presented here. As technology improves, researchers can expect to add to the number of observable parameters, along with better measurement for existing parameters in terms of accuracy, resolution and coverage on temporal and spatial scales.
The Anderson River vegetation study was performed under contract to the Canadian Wildlife Service by Borstad Associates Ltd. (Sidney, Canada). Compilation and analysis of the northeast Pacific Ocean SST and chlorophyll datasets was funded by the Canadian Space Agency Earth Observation Government Related Initiatives Program, under contract to the Department of Fisheries and Oceans. The authors also acknowledge the long-term commitment of the CIS to providing high-quality ice datasets over vast areas of the Arctic Ocean and its marginal seas.
For a full list of reference, please contact Dr. Gary Borstad at firstname.lastname@example.org.
Dr. Gary Borstad is vice president and director of remote sensing at ASL Environmental Sciences Inc. He was founder and president of G.A. Borstad Associates Ltd., which recently merged with ASL, and has been an active consultant and researcher in satellite and airborne remote sensing since 1971.
Leslie Brown is senior remote sensing analyst at ASL Environmental Sciences Inc. She holds a master's degree in oceanography and has more than 30 years' experience in remote sensing and oceanography.
David Fissel graduated from the University of British Columbia with a M.Sc. in physical oceanography in 1975. As a founding partner and now president of ASL Environmental Sciences Inc., he has managed more than 250 oceanographic consulting projects carried out in Canada and around the world.