Introduction

Arctic sea ice is a key component of the polar climate system. Typically, it is monitored remotely from space, from the air, and by direct observations from ships and floating ice islands. Physical parameters that are monitored include temporal and geographic extent, ice area, ice age, and ice thickness. Additionally, the movement of sea ice in response to wind and ocean currents is documented through the collection and analysis of imagery time series.

Since late 1978, scientists have systematically monitored the extent of Arctic sea ice cover from space by using passive microwave satellite imagery collected by the U.S. Defense Meteorological Satellite Program (DMSP) instruments (Peng and others, 2013). Sea ice and other Earth surface features emit measurable low-level electromagnetic energy in the microwave range of the spectrum. The DMSP sensors detect this naturally emitted passive microwave radiation. Compared to clouds and ocean water, sea ice emits significantly more microwave radiation. Consequently, the passive microwave energy produced by sea ice, which penetrates clouds, is used to detect sea ice, both day and night, regardless of cloud cover.

According to the National Snow and Ice Data Center’s “Cryosphere Glossary” (NSIDC, 2023), ice extent is the “total area covered by some amount of ice, including open water between ice floes.” Ice extent is typically reported in square kilometers. A typical threshold between “ice-covered” or “not ice-covered” is 15 percent.

The 15-percent minimum cutoff is used to define the ice edge because it provides the most consistent agreement between satellite and ground observations (NSIDC, 2022b). The sea ice area, typically measured in millions of square kilometers (10⁶ km²), is the part of the sea ice extent that is covered by ice at any point in time. Maximum sea ice coverage usually occurs in March, and minimum sea ice coverage occurs in September before colder weather begins new ice formation.

U.S. agencies that monitor the extent of Arctic sea ice coverage are the National Snow and Ice Data Center (NSIDC), the National Aeronautics and Space Administration (NASA), and the U.S. National Ice Center (NIC). The NIC is a partnership among the U.S. Navy, the National Oceanic and Atmospheric Administration, and the U.S. Coast Guard.

Space-based monitoring of Arctic sea ice began with the launch of NASA’s Scanning Multichannel Microwave Radiometer (SMMR) on October 25, 1978. Consequently, 1979 was the first full year of sea ice satellite observations. During the more than 43 years of satellite monitoring of Arctic sea ice, there have been vast decreases in both the maximum and minimum extents of ice coverage (fig. 1). The absolute minimum Arctic sea ice extent recorded was 3.4×10⁶ km² in September 2012 (NSIDC, 2022a). The NSIDC’s “Charctic Interactive Sea Ice Graph” (NSIDC, 2022a) depicts ice extents of 3.404×10⁶ km² on September 15, 2012; 3.389×10⁶ km² on September 16, 2012; 3.387×10⁶ km² on September 17, 2012; and 3.418×10⁶ km² on September 18, 2012. A 2012 NSIDC publication had originally identified September 16 as being the all-time minimum, with an extent of 3.418×10⁶ km² (NSIDC, 2012).

Since 2012, the annual minimum Arctic sea ice extent has been significantly less than the 30-year average baseline. This 30-year record, which is based on the average ice extent for the period from 1981 to 2010, is frequently used because it provides a consistent sea ice extent baseline for year-to-year comparisons. According to the National Centers for Environmental Information (NCEI, 2021) of the National Oceanic and Atmospheric Administration, 30 years is considered a standard baseline period for weather and climate. Beginning in 1978, the more than 43-year-long satellite record of ice observations is long enough to provide a space-based observational baseline period as well (Meier and others, 2017; NSIDC, 2022a).

For example, based on satellite observations (NSIDC, 2022a), Arctic sea ice daily minimum extent decreased from 7.6×10⁶ km² on September 13, 1980, to 4.1×10⁶ km² on September 18, 2007, and then to 3.4×10⁶ km² on September 17, 2012. Since then, the annual minimum has fluctuated between 5.1×10⁶ km² on September 13, 2013, and 3.8×10⁶ km² on September 18, 2020. The 2020 minimum sea ice extent was the second lowest extent measured since modern record keeping began in 1978. The extents in 2007 and 2016 are tied for the third lowest Arctic sea ice extent at 4.1×10⁶ km² (NSIDC, 2022a). The 2022 minimum (4.7×10⁶ km² on September 18) is tied for tenth lowest with 2017 and 2018. To put this area and loss into perspective, the size of the conterminous United States is about 8.1×10⁶ km². Figure 2 depicts the monthly extent of Arctic sea ice in September 1980, 2012, and 2022.

This report describes the Global Fiducials Program (GFP) and how Global Fiducials imagery has been used to document Arctic sea ice dynamics. The origins and history of the GFP are described and examples are presented depicting how the GFP’s imagery products are uniquely used to document complex sea ice behavior at spatial resolutions previously not publicly available.

Global Fiducials Program and Library

Through the Global Fiducials Program (GFP), the U.S. Geological Survey (USGS) is helping the global science community to study rapidly changing Arctic sea ice behavior by providing free and unrestricted imagery time series of both moving ice containing drift buoys and static sites. The time series document annual and inter-annual variability. The imagery is available at a Global Fiducials Library (GFL) website, the Global Fiducials Library Data Access Portal (USGS GFLDAP) at https://www.usgs.gov/global-fiducials-library-data-access-portal, which is described below.

The GFP was created in the mid-1990s by the Central Intelligence Agency (CIA) through its MEDEA Program in conjunction with a Government Applications Task Force made up of 13 other U.S. Federal agencies, including the USGS (Baker and Zall, 2020). The term “MEDEA” is not an acronym. Rather, it was chosen to complement the name of another U.S. Government advisory group, “JASON,” which comes from the Greek myth related to Jason, Medea, and the Argonauts (Baker and Zall, 2020).

The agencies recognized that exploring the effects of global environmental change required earth science data that could be derived from extremely high resolution United States National Imagery Systems (USNIS) imagery. Such imagery could be used for systematic monitoring of dynamic and environmentally critical areas of the Earth in five major disciplines, which are described at the GFLDAP: ocean processes, land use/land cover, atmospheric processes, ice and snow dynamics, and geologic processes. The study of sea ice is a subset of both the ocean processes discipline and the ice and snow dynamics discipline.

Initially, MEDEA’s data acquisitions were based on input from a group of about 60 U.S. academic scientists who were selected by MEDEA Program managers and who obtained high-level security clearances. Beginning in 1997, Federal agency scientists were also able to nominate new locations for data acquisition.

Individual data-acquisition locations were referred to as “fiducial sites” or “fiducial points.” Each serves as a benchmark for long-term, ongoing monitoring of global environmental change (Baker and Zall, 2020, p. 23).

Following its inception, MEDEA’s GFP identified about 500 worldwide fiducial sites (Dozier, 1997), 6 of which were focused on Arctic sea ice. Collection of USNIS electro-optical images from the sites began in the mid-1990s, with thousands of images being collected (Molnia and others, 2018; Baker and Zall, 2020, p. 23). Many images were not usable because of clouds and haze. To date, most of the collected images remain classified and have not as yet been released to the public.

MEDEA’s Global Fiducials Program existed during 1993–2000 and from 2008 to May 2015, when it was abolished by the CIA. The USGS-led Civil Applications Committee (CAC) directed the GFP from 2000 to 2008 and twice joined with MEDEA to co-lead the program (1997–2000 and 2008–2015). Since 2015, the CAC’s Global Fiducials Working Group (GFWG) has solely led the GFP.

In 1997, the CAC established the Global Fiducials Library (GFL) to serve as a centralized archive for GFP imagery and to provide a mechanism for searching and accessing these data. The “About” page of the USGS GFLDAP site (U.S. Geological Survey, 2022) describes the GFL as follows: “The Global Fiducials Library (GFL) is a long-term archive of images from U.S. National Imagery Systems which represents a long-term periodic record for selected scientifically important sites. The GFL was created to be the collection, archive and data management component of the Global Fiducials Program (GFP).” The “long-term periodic record” consists of imagery time series (that is, sets of images taken sequentially of the same location) from the GFP that document natural hazards and disasters, global change, land resources issues, ecosystem dynamics, and many other types of Earth surface processes. The GFL contains imagery of domestic GFP sites, international GFP sites, Antarctic sites, and Arctic sea ice sites.

Since 2015, the GFWG has solely managed Global Fiducials activities, including the identification of new fiducial sites. Like MEDEA’s fiducial points, a fiducial site is a geographic location that is used as a benchmark for the long-term monitoring of processes, both natural and anthropogenic, associated with the causes and effects of global environmental change. Fiducials are marks or points of reference applied to images to present a fixed standard of reference; in the GFP, they refer to the identification of a place on Earth’s surface. The GFP maintains long-term data records for specific places on Earth’s surface, and the records are made available for scientific investigations. Fiducial sites are associated with Earth processes and environmentally sensitive areas that are being monitored so that scientists can better understand and model the dynamic systems and changes that are occurring. The sites are located around the globe.

According to a report by the U.S. Joint Chiefs of Staff (2017, p. G-9), an imagery-derived product (IDP) is “[a]ny representation made from US classified satellite imagery that is not a direct copy of the original image itself.” An IDP may be analog or digital. A literal IDP is an image-like product that is not a direct copy of the original image, whereas a nonliteral IDP is (1) a graphic product such as a map, a line drawing, a chart, or a graph or (2) metadata or other statistical data derived from imagery. Literal IDPs (LIDPs) have had their pixels digitally manipulated so that the pixels’ original characteristics are irreversibly changed while the appearance of the image remains unchanged. This manipulation is done to prevent adversaries from determining the capabilities of the sensors that collected the original image.

To date, more than 4,400 LIDPs, produced from USNIS images collected in support of the GFP, have been made available for unrestricted public distribution and scientific analysis at a resolution of about a meter. They are available from the USGS GFLDAP at https://www.usgs.gov/global-fiducials-library-data-access-portal. Of these LIDPs, 1,751 are sea ice images collected between 1999 and 2014. Many are accompanied by nonliteral IDPs. Examples of these IDPs appear in later sections of this report.

The CAC represents Federal civil agency interests and requirements in the use of Global Fiducials imagery (Molnia and others, 2012). The CAC facilitates the appropriate Federal civil uses of overhead remote sensing technologies and data collected by military and intelligence agencies and commercial sources. A primary function is ensuring that the personal privacy and civil rights of U.S. persons are not violated by the use of GFP imagery. The CAC is operated and staffed by the USGS on behalf of the U.S. Department of the Interior and its interagency partners (Opstal and Rogers, 2022).

The USGS GFLDAP provides unrestricted access to GFP imagery from more than 150 fiducial sites. Users may search the web portal for existing imagery and preview that imagery and its associated metadata. The USGS GFLDAP web portal (fig. 3) has a fiducial site location map and a brief description about each site. Fiducials imagery is available to preview through an imagery gallery, the Global Fiducials Library Image Gallery website, which is a part of the USGS long-term archive at https://gfl.usgs.gov/imagegallery/?PTAGNAME=ArcticBuoys&img:2573:208.

Images collected over land are orthorectified to provide terrain-corrected, accurately positioned imagery. This is not done for sea ice images. Images of Arctic buoys and static sea ice sites are orthocorrected to a constant elevation because the sites are at sea level. Arctic sea ice buoy LIDPs have a spatial resolution of 1.3 meters (m). Metadata are also made available for all GFL sea ice imagery products, including Sun elevation and Sun azimuth for the Arctic datasets. All other GFL sites, including the static sea ice sites, have a spatial resolution of 1.0 m.

Studies of four types of GFP Arctic sea ice sites have produced imagery that is in the GFL. Descriptions and images from the four types of sites are listed below:

Why Study Sea Ice Processes?

The extent of Arctic sea ice influences Earth’s climate and albedo. Large areas of ice and snow reflect much of the Sun’s energy back into space, helping to cool the Earth. Darker ocean water or substantial sea ice areas covered with melt ponds (fig. 4A, B) absorb the Sun’s energy, which increases ocean water temperature or accelerates sea ice melting. The more open water that is exposed, the more energy that is absorbed by the ocean. The increase in temperature significantly affects the survival of the sea ice cover during summer. Understanding the mechanisms and triggers of melt pond formation processes becomes critical for choosing data to use in modeling climate changes and the effects of melt ponds on Arctic sea ice extent. Storms and wind-driven waves also play a role in the Arctic sea ice breakup and melting.

In general terms, the extent of sea ice in the Arctic increases in size during the fall and winter months (October through March) when little or no solar radiation reaches the ocean’s surface. This lack of solar radiation permits new first-year ice to form and thicken and, in some locations, to mix with existing ice. During the spring and summer months, the thickness and extent of the sea ice begin to decrease from melting as solar radiation reaching the ocean’s surface increases. If the sea ice does not thicken enough over the winter, it may completely melt away during summer. If the ice grows thick enough during the winter, it will experience thinning during the summer, but it may not completely melt away. In this scenario, the ice will remain until the following winter, when it will again grow and thicken. The ice formed during more than one winter is classified as multiyear ice. During winter, snow may accumulate on the surface of sea ice.

During spring and summer, snow remaining on top of the ice melts, and the resulting meltwater and additional water from melting sea ice accumulate in cracks and depressions on the sea ice surface, forming melt ponds. These melt ponds absorb more heat from the Sun’s radiation compared to the surrounding ice and snow, which typically reflects the sunlight. As the water in the melt ponds warms, the ponds become larger in surface area and deepen, melting into the underlying ice. As the melt ponds become deeper, they can melt through the entire thickness of the sea ice. Channels may form that connect melt ponds and drain meltwater from one melt pond to another.

Imagery in the GFL is of high enough resolution to allow mapping of melt ponds to get estimates of how much of the sea ice, on average, is covered by the darker-than-ice ponds. A melt pond dataset, compiled by Fetterer and others (2008), was based on GFL imagery from the GFL Arctic static ice sites.

Arctic Sea Ice Imagery Available From the Global Fiducials Library

On June 1, 2022, the USGS GFLDAP website displayed 1,751 Arctic sea ice images, collected between 1999 and 2014, that were available for free download. These images represent a subset of the more than 3,138 sea ice images that GFP records indicate were collected during that period, exclusive of SHEBA imagery. The images collection includes the following: 844 images collected from the six static sites, 1,734 images collected of sea ice containing drift buoys, and about 520 images collected of Marginal Ice Zone, Ice Edge, and Seasonal Ice Zone Region sites. The GFP plans for all sea ice imagery collected since 2014 to be added to the GFL website once LIDPs are produced and public release approval is received.

Imagery From the Surface Heat Budget of the Arctic Ocean (SHEBA) Project

The first Arctic sea ice studies began in 1893, when the Norwegian ship Fram, under the command of Fridtjof Nansen, was frozen into the Arctic Ocean as part of an attempt to reach the geographical North Pole (Nansen, 1897). This first systematic effort to monitor Arctic sea ice resulted in significant information about Arctic sea ice dynamics. Seventy-nine years later, the Arctic Ice Dynamics Joint Experiment (AIDJEX) was conducted. According to the NSIDC (2022c), the AIDJEX program was the first major sea ice experiment constructed specifically to answer emerging questions about sea ice movement and changes in the western Arctic Ocean in response to ocean and atmosphere forcing. Initial field deployment in 1972 was followed by expanded field activities in 1975 and 1976. This exceptional study was accomplished without a systematic space-based or aerial high-resolution imagery collection component or capability.

In 1997, 104 years after the Fram expedition, the Surface Heat Budget of the Arctic Ocean (SHEBA) Project began; it was funded by the National Science Foundation (NSF) of the United States (National Science Foundation, 1998). The Canadian icebreaker Des Groseilliers was purposely frozen into sea ice of the Beaufort Sea. The field component of the SHEBA Project lasted for 404 days, from August 25, 1997, to October 3, 1998, with the icebreaker serving as both an experiment platform and a base for scientific researchers. Scientists from across the globe were able to reside on the ship, perform experiments, and conduct field excursions to study and observe sea ice dynamics with the goal of gaining a better understanding of global climate change (Uttal and others, 2002).

Although all participants refer to the activity as “SHEBA,” multiple titles are used when the project name is fully spelled out. The NSF (National Science Foundation, 1998, 2000), which funded the experiment, referred to it as the “Surface Heat Budget of the Arctic Ocean” Project. The National Center for Atmospheric Research (NCAR; undated) also referred to it as the “Surface Heat Budget of the Arctic Ocean” Project, whereas the NSIDC, which cohosts the data, referred to it as the “Surface Heat Balance of the Arctic” Experiment (Fetterer and Untersteiner, 2000). The format used in this report, “Surface Heat Budget of the Arctic Ocean” Project, follows that used by the NSF.

The GFP’s monitoring of sea ice by using high-resolution USNIS satellite imagery began in September 1997. More than 100 USNIS high-resolution images were acquired of the ship and the ice adjacent to it as the ship drifted across the Beaufort Sea. When degraded and released, this GFP imagery was the only publicly available imagery with high enough resolution to document the complex changes that occurred in the sea ice during the period of observation. The resolutions of images from Landsat (15–30 m resolution), commercial satellites collecting electro-optical images (typically 3–5 m resolution), and similar Canadian and European satellites (such as RADARSAT) that were operational at the time of the SHEBA Project were too low to display the small and subtle changes that occurred on the sea ice surface. Median melt pond diameters observed during the SHEBA Project were about 4 m. The SHEBA Project provided the first high-resolution imagery to monitor the behavior of the same ensemble of Arctic sea ice for a significant period of time.

The Des Groseilliers transmitted the latitude and longitude of its changing position as it drifted (fig. 5). The hourly receipt of the ship’s coordinates permitted the USGS to anticipate future ship locations and to schedule time-sensitive new collections designed to reimage the large area (about 15×40 km) of the same sea ice that surrounded the drifting ship. Although most imagery showed significant cloud cover, 57 generally cloud free, optical-band reconnaissance images were reprocessed, and LIDPs were approved for public release by the National Imagery and Mapping Agency (now the National Geospatial-Intelligence Agency) at the request of the NSF. One of these images is presented in figure 6A. Figure 6B depicts the area outlined in red on figure 6A and shows the icebreaker. This publicly released imagery is housed in the GFL, and the LIDPs are available from the NSIDC (Fetterer and Untersteiner, 1998, 2000).

Imagery From Six Static Sea Ice Sites in the Arctic Ocean Monitored by the Global Fiducials Program

Following the successful imaging done during the SHEBA Project, in 1999, the GFP began to systematically monitor other Arctic sea ice by using imagery from the USNIS. Initially, six static locations were selected in order to better understand sea ice dynamics, distribution, age, melt history, and annual variability. The sites were “static” in that a finite location (fiducial point) was imaged and whatever ice, if any, that was present was documented. The static sea ice sites (fig. 7) were selected so that the quantity, age, and condition of sea ice could be repeatedly monitored at these unique locations. Four sites had collections over 15-year periods, while two were imaged for 10 years. The publicly released LIDPs have a spatial resolution of about 1.3 m and depict areas that are 2×2 km to 6×6 km in size.

The static sea ice sites in the Arctic Ocean (presented from west to east) are described below, and the latitude and longitude are given for each site. The descriptions of the static sites are derived from unpublished MEDEA and CAC documents produced during the GFP’s site selection process.

Chukchi Sea site (lat 70° N., long 170° W.)—The Chukchi Sea site is located about 225 km northwest of Point Hope, Alaska, and about 515 km south-southwest of the city of Utqiagvik (formerly Barrow), Alaska. Historically, the area containing the Chukchi Sea site was near the boundary between older multiyear ice and thinner first-year ice that tended to melt away every summer. Examination of satellite imagery and NSIDC-hosted data by author Molnia has shown that the age and extent of sea ice in this region have significantly declined since the late 1970s. At this site, 80 images were collected between 2005 and 2014.

Barrow, Alaska, site (lat 71° N., long 156° W.)—The Barrow, Alaska, site includes the part of Alaska’s North Slope Borough containing the city of Utqiagvik (formerly Barrow) and extends about 5 km from the shore over the adjacent continental shelf. It is in a location where sea ice forms along the western Alaska coast in winter and generally melts or breaks away by mid-July. Changes in the timing of coastal sea ice breakup and the location of offshore sea ice have significant local ecological, biological, and human effects. The image time series collected at this site portrays changes in the timing of coastal sea ice breakup and provides information on the properties of smaller areas of ice. Figure 8 depicts the variation that occurred in two consecutive years, comparing images collected a year apart in July 2006 and July 2007. At this site, 97 images were collected between 2005 and 2014.

Beaufort Sea site (lat 73° N., long 150° W.)—The Beaufort Sea site is located about 165 km north-northwest of Prudhoe Bay, Alaska. This region has been the site of many field studies since the International Geophysical Year (1957–1958). The site was typically near the edge of the ice pack, but annual ice conditions vary, as is shown by the two images in figure 9. In the August 2001 image, more than 75 percent of the field of view is ice covered. In the August 2007 image, more than 75 percent of the area is ice free. At this site, 163 images were collected between 1999 and 2014.

Canada Basin site (lat 85° N., long 120° W.)—The Canada Basin site is located about 1,725 km north-northeast of Prudhoe Bay, Alaska, and about 500 km from the closest land on the coast of Ellesmere Island. The Canada Basin site typically has the oldest and thickest ice with the longest residence time in the Arctic Ocean. The GFP Arctic sea ice time series is one of the most significant datasets for documenting the types of change that are affecting this older ice area. At this site, 165 images were collected between 1999 and 2014.

Fram Strait site (lat 85° N., long 0°)—The Fram Strait site is located about 410 km northeast of the northernmost point of land in Greenland. The Fram Strait, located between Greenland and Spitsbergen, Norway, is the primary exit route of sea ice from the Arctic Ocean into the Greenland Sea. During the period of observation of the Fram Strait site, the age and quantity of ice observed declined. At this site, 167 images were collected between 1999 and 2014.

East Siberian Sea site (lat 82° N., long 150° E.)—The East Siberian Sea site is located about 1,100 km north-northeast of the closest point of land in Siberia, Russia. The region containing the site produces most of the Arctic Ocean’s first-year ice. On the basis of the observed reduction in ice age and duration, this area appears to be sensitive to inter-annual oceanic changes and atmospheric forcing. This sensitivity is borne out by the dramatic decreases in summer ice extent during most of the past two decades. At this site, 212 images were collected between 1999 and 2014.

Table 1 summarizes the number of images that were collected at each of the six static sea ice sites in the Arctic Ocean.

Table 1.    

Number of images collected at six static sea ice sites in the Arctic Ocean between 1999 and 2014 and released to the Global Fiducials Library.

[Data from the U.S. Geological Survey’s Global Fiducials Library Data Access Portal (https://www.usgs.gov/global-fiducials-library-data-access-portal), May 15, 2022. Sites are listed from west to east, and locations are shown in fig. 7]

Table 1.    Number of images collected at six static sea ice sites in the Arctic Ocean between 1999 and 2014 and released to the Global Fiducials Library.

Site name	Duration	Number of images
Chukchi Sea site	2005–2014	80
Barrow, Alaska, site	2005–2014	97
Beaufort Sea site	1999–2014	163
Canada Basin site	1999–2014	165
Fram Strait site	1999–2014	167
East Siberian Sea site	1999–2014	212
Total		884

Drifting Buoys Deployed by the International Arctic Buoy Program

The MEDEA Program recommended that the USGS track buoys that were deployed by the International Arctic Buoy Programme (IABP), which began in 1991 and which is coordinated by the University of Washington’s Polar Science Center. Twenty different research and operational institutions from nine different countries collaborate to maintain a network of drifting buoys in the Arctic Ocean. According to the University of Washington, Polar Science Center (2022b), the buoy network provides “meteorological and oceanographic data for real-time operational requirements and research purposes including support to the World Climate Research Programme (WCRP) and the World Weather Watch (WWW) Programme.”

As was the case with the SHEBA Project more than a decade earlier, in which the icebreaker’s position was frequently reported to the USGS, tracking multiple GPS-equipped buoys anchored in sea ice provided an opportunity to simultaneously collect high-resolution imagery to monitor the behavior of the same ensemble of Arctic sea ice for a significant period of time at multiple locations. Ultimately, in the newer study, 38 different drift buoy targets were successfully monitored and imaged between 2009 and 2014, with as many as 9 being observed and imaged during a single summer.

The IABP website (University of Washington, Polar Science Center, 2022a) indicates that the IAPB deploys several types of GPS-equipped buoys: (1) ice mass-balance buoys (IMBs), which measure meteorological and sea ice and snow parameters (Richter-Menge and others, 2006; Polashenski and others, 2011; Planck and others, 2019); (2) ice tethered profiler (ITP) buoys, which measure salinity and upper ocean temperature (Woods Hole Oceanographic Institution, 2022); and (3) Upper layer Temperature of the Polar Oceans (UpTempO) buoys, which measure salinity and temperature of the upper 25–60 meters of the water column (University of Washington, Polar Science Center, 2022d).

Figure 10 is a diagram depicting instrumentation of the IMB type of buoy. Figure 11 presents a photograph that shows an IMB buoy being prepared for deployment.

Methods for Tracking Buoys

Until 2010, the Advanced Research and Global Observation Satellite (ARGOS) system was used to transmit data and buoy positions to the USGS. That year, the U.S. Army Cold Regions Research and Engineering Laboratory (CRREL), a U.S. logistics provider for the IABP, switched from ARGOS data transmission satellites to Iridium Communications transmission satellites. In 2011, GPS receivers and transmitters were added to the buoys equipped with Iridium Communications.

The IABP managers assisted the USGS by setting up a file transfer protocol (FTP) website that allowed the USGS to download buoy GPS data. With access to the buoy GPS data files, the USGS developed a methodology to convert the GPS data text files to a comma separated values (CSV) format and create an ArcGIS shapefile. The ArcGIS shapefile was then plotted within Esri’s ArcMap to plot out the latest buoy positions. By plotting the buoy track, USGS personnel saw how the wind and ocean currents affected buoy and associated ice floe movement. These plots aided in predicting where to task the next target location. Tasking is the formal process of determining and submitting the specific information necessary to correctly identify the location where a new image needs to be collected.

In the summer of 2008, the USGS successfully captured imagery of ice floes by using a target size of 4×4 nautical miles (nmi). This imagery demonstrated to the USGS, MEDEA, and the IABP that the USGS-developed methodology worked. In 2009, target sizes were modified to 10×10 nmi to maximize the amount of time the drifting buoy would be within the target area.

In 2009, the MEDEA Program asked the USGS to track the same ice floes over an entire season from April through September, over multiple sites (Wilson and others, 2012). In 2011, Python scripts were incorporated into the methodology to more quickly convert buoy GPS position text files to ArcGIS shapefiles. This development allowed the USGS to annually monitor and track more GPS buoys in support of the MEDEA Program.

The MEDEA Program scientists involved in cryosphere studies determined which areas of the Arctic needed to be studied. The IABP managers provided the locations of the GPS buoys. Together they determined which buoys in the IABP network the USGS would track and image. The IABP managers set up a website where selected Arctic buoy GPS data were updated hourly and stored. This was the primary source of information used by the USGS to track the buoys.

A buoy’s GPS text file contains the buoy’s ARGOS identification number, year, day, time, and latitude and longitude (fig. 12). This file information was updated hourly and was available on an FTP site for USGS collection managers to download and process into a geographic information system (GIS) format. These data were plotted to construct an ever-growing buoy track. The ArcGIS shapefile of the buoy track was used to estimate the displacement and direction of travel of the buoy.

The GPS buoy track of hourly records was plotted daily and was color coded by day (fig. 13). For example, all Monday positions were the same color, whereas the Tuesday positions were a different color, and so on.

The buoy movement was caused by prevailing winds and ocean currents pushing against the ice pack and ice floes. Because of this driving mechanism, the ice movement could be erratic. Alternatively, the buoy could travel for days in a straight line. To account for these variables, USGS collection managers plotted the buoy track daily.

Next, the USGS used the buoy track plot to estimate a position for a target area. The challenging part was to position this target so that the buoy and therefore the surrounding sea ice would remain inside the target area for the next 2–3 days (fig. 14A, B). Estimates of the target position depended on the displacement rate and the direction in which the buoy and ice were moving. Sometimes, Arctic storms caused a buoy to move very quickly out of the target zone. The time of year also greatly affected buoy movements. For instance, in the spring when the sea ice was basically a large continuous sheet of pack ice, buoy and ice floe movements tended to be slower. As the year progressed, melt ponds and leads formed, and the ice began to melt and break apart. In late summer and early fall when sea ice extent was close to its minimum, there was more open water between ice floes. Buoys were pushed much greater distances, often at higher velocities. Plotting the GPS buoy track led to determining the displacement rate of the buoy, as the tighter the GPS spacing was between the hourly records (dots on a map), the slower the buoy was moving. Conversely, the farther apart the hourly GPS records, the faster the buoy and ice were moving.

Figure 14C demonstrates the successful collection of an actual image captured 2 days after tasking. Receiving a cloud-free image of ice inside the projected target area constituted a successful collection. One reason that the USGS plotted the buoy track on an hourly and daily basis was to make sure that the buoy was still well inside the targeted area. If the buoy moved outside of the target zone, or even close to the edge of the target area, a new target position was calculated, plotted, and submitted for collection of future images. If a collected image had more clouds than was acceptable, the buoy track was checked against the target position, recalculated if necessary, and resubmitted for collection.

Occasionally, a GPS buoy stopped transmitting. This situation was caused by several different issues. The most likely cause was that the battery powering the GPS unit had run out of electrical power. In this case, the USGS contacted the IABP and MEDEA to identify an alternative buoy to target. Sometimes a buoy got caught up on an ice ridge, which caused the GPS transmission antenna to be tilted at an angle, resulting in the satellite relay not receiving the GPS signal. Sometimes the signal was blocked. In that situation, the USGS monitored the GPS website to see if the buoy resumed transmitting GPS data. If it did not transmit within a week, the USGS consulted with the IABP and MEDEA to switch collection to a new buoy. After the USGS downloaded an image, quality-assurance and quality-control steps were performed to confirm that the same ice floes were visible for each buoy being tracked.

For example, an image of the ice containing Buoy 132070 was taken on June 21, 2012 (fig. 15A), and compared to an image taken 32 days later on July 23, 2012 (fig. 15B), to make sure that the large dark ice mass, an unnamed ice island, was imaged in both scenes. It was. An oblique aerial photograph of the unnamed ice island had been taken more than a year earlier, on April 15, 2011. It is shown on figure 15C. The foreknowledge of the presence of the ice island was a factor in deciding the site of buoy placement.

Usually, ice islands are not composed of sea ice. They are tabular icebergs of glacier origin, found in the Arctic Ocean, and they usually form by breaking off from an ice shelf. Ice islands typically are 30–50 m thick and occupy areas ranging from a few thousand square meters to 500 km² (NSIDC, 2022b). Glacier ice and sea ice have different chemical and physical properties, including salinity, color, and surface texture.

Arctic ice islands typically calve from Canadian High Arctic ice shelves, primarily located along the north coast of Ellesmere Island. These Canadian High Arctic ice shelves are the only ice shelves in North America, and they are the most extensive ice shelves in the northern polar region. Ellesmere Island ice shelves originate from both glacier ice and sea ice (Jeffries, 1992).

Imagery Products From Sea Ice Buoys

Between 2009 and 2014, the USGS successfully tracked 38 different sea ice buoys and collected satellite images of the changes in the adjacent sea ice. The LIDPs of many of the resulting imagery products are available on the USGS GFLDAP website for free download and for unrestricted use. Table 2 summarizes the annual number of observed buoys and images collected.

Table 2.    

Number of images collected from 38 buoys drifting in the Arctic Ocean between 2009 and 2014 and released to the Global Fiducials Library.

[Data from the U.S. Geological Survey’s Global Fiducials Library Data Access Portal (https://www.usgs.gov/global-fiducials-library-data-access-portal),May 15, 2022]

Table 2.    Number of images collected from 38 buoys drifting in the Arctic Ocean between 2009 and 2014 and released to the Global Fiducials Library.

Year	Buoys tracked	Images collected
2009	4	213
2010	7	163
2011	3	166
2012	9	251
2013	9	648
2014	6	293
Total	38	1,734

The tracks of 38 buoys monitored between 2009 and 2014 are shown on figure 16. This figure depicts the general location, the path, and the complexity of the route that each individual buoy traveled during its period of observation. The USGS successfully imaged ice movement across hundreds or, in some cases, thousands of kilometers of the Arctic Ocean for periods from less than 2 months to, in one instance (IAPB Buoy 125530), more than 17 months. A description of some uses of drift buoy observations was presented in Kwok and Untersteiner (2011).

Typically, imagery was analyzed by the USGS scientists, cooperators, and MEDEA scientists to determine ice fracture patterns, sea ice ridge heights, ice cover percentages, seasonal development and coverage of melt ponds, evolution of ice concentrations, floe size distribution, lateral melting, and other variables that are used to monitor sea ice extent and behavior and to refine and develop climate models. Resulting ice floe imagery has been added to the GFL for 38 buoys tracked from 2009 through 2014.

Case studies depicting the imagery for two drift buoys are presented below. These images demonstrated the timing and extent of melt pond development and allowed scientists to conduct comparative analysis studies to better understand seasonal changes and physical ice changes and processes that affected the same ice floes during the period of observation.

Case Study 1—Advanced Research and Global Observation Satellite Buoy 132070

ARGOS Buoy 132070, located on an unnamed ice island shown in figure 15C, was monitored between June 20, 2012, and September 24, 2012, and imaged between June 21, 2012, and September 21, 2012. During that 97-day monitoring period, Buoy 132070 drifted 1,570 km. The USGS GFLDAP website page for this buoy contains 15 LIDP images derived from images collected through September 21, 2012. This Beaufort Sea ice island was about 2.7 km long by 1.3 km wide when first imaged. Figure 17 is a map showing the westerly track the ice island and buoy followed from north of the United States-Canada border to north of the Bering Strait.

Figure 18 contains eight LIDP images derived from USNIS electro-optical images of the ice containing ARGOS Buoy 132070; the images were collected during the 48-day period between June 21 and August 7, 2012. All have been degraded for public release. Each image depicts an area that was 3.1 km from east to west by 3.6 km from north to south. The images show the changes that the unnamed ice island and adjacent sea ice experienced as they drifted to the west during the summer of 2012.

As shown in the image collected on June 21 (6/21 on fig. 18), the ice island, which had a dark-gray surface cut by subparallel fractures or crevasses, was oriented nearly north-south. The area surrounding it was composed of tightly packed angular ice floes with only four small areas of open water. Much of this area was snow covered with a mottled appearance, probably due to the presence of small melt ponds.

The image collected 15 days later, on July 5 (7/05 on fig. 18), showed that the ice island had rotated nearly 90° to the east. The former northern end was pointing to the northeast. The surrounding area was still composed of tightly packed angular ice floes. The southwestern corner of the area imaged had a large area of open water. Most of the snow was gone, and the number of small melt ponds had increased. A lead was beginning to form at what previously was the south end of the ice island.

The image collected 5 days later, on July 10 (7/10 on fig. 18), showed that the continued rotation of the ice island had resulted in the former northern end pointing to the southeast. Most of the surrounding area was still composed of tightly packed angular ice floes, but an area of open water measuring about 1.0×1.5 km had developed on the west side of the ice island. Most of the remaining snow was located between ice floes. The number of small melt ponds had continued to increase.

The image collected 7 days later, on July 17 (7/17 on fig. 18), showed that the former northern end was again pointing to the northeast. The direction of its rotation was unknown. Only the western part of the surrounding area was composed of tightly packed angular ice floes. The entire eastern area was open water. The size of the melt ponds had continued to increase.

The image collected 3 days later, on July 20 (7/20 on fig. 18), showed that the former northern end was pointing to the southeast, and the island’s area was decreasing due to calving and melting. Most of the western part of the surrounding area was composed of tightly packed angular ice floes with several large areas of open water. The entire eastern area was open water. The area of the melt ponds had continued to increase.

The image collected 3 days later, on July 23 (7/23 on fig. 18), showed that the former northern end was pointing to the south-southeast, yet the ice island had changed very little. Most of the western part of the surrounding area was composed of tightly packed angular ice floes with fewer areas of open water than on July 20. The entire eastern area was still open water. The area of the melt ponds had continued to increase.

The image collected 9 days later, on August 1 (8/01 on fig. 18), showed that more than half of the ice island present on July 23 had calved and melted away. Only a few small ice floes remained. The remaining area was open water.

The image collected 6 days later, on August 7 (8/07 on fig. 18), showed that the remaining piece of the ice island had continued to calve and melt. The remaining fragment, which was now about 0.7×1.9 km, had rotated nearly 90° to the west. Only a few small ice floes remained.

Although images used here extend through August 7, imaging continued through September 21, and position monitoring of the few remaining fragments of the ice island continued until September 24, 2012.

Case Study 2—World Meteorological Organization Buoy 47611

The World Meteorological Organization (WMO) Buoy 47611, located in the Beaufort Sea, was monitored from June 20, 2009, to September 24, 2009, and imaged from June 27, 2009, to September 14, 2009. During that 79-day imaging period, the buoy drifted nearly 800 km. The USGS GFLDAP website page for this buoy contains 49 unique LIDP images derived from images collected by the USNIS through September 14, 2009.

Figure 19 is a map showing the westerly track that WMO Buoy 47611 followed as it drifted from northeast of the United States-Canada border to north of Alaska.

Figure 20 contains 10 LIDP images derived from USNIS electro-optical satellite images of the ice containing WMO Buoy 47611 that were collected between June 27, 2009, and August 11, 2009. All have been degraded for public release. Each image depicts an area that was approximately 6.4 km from east to west by 6.8 km from north to south.

The images show the changes that the monitored sea ice experienced as it drifted to the west during the 48-day period depicted in the figure. On June 27, 2009 (6/27 on fig. 20), the monitored area contained more than a dozen large subrounded to angular composite ice floes. They ranged in maximum dimension from less than 0.25 km to more than 3.0 km. The two largest ice floes were labeled “A” and “B.” Only a part of floe B can be seen. Less than 15 percent of the area was open water. Some of the ice was snow covered with a mottled appearance, probably due to the presence of small melt ponds. Cloud shadows and small areas of sunshine provide a false hummocky texture to parts of the image.

The image collected 10 days later, on July 7, 2009 (7/07 on fig. 20), showed that most of the large composite ice floes described above changed minimally. A north-south fracture in composite ice floe A had resulted in the separation of its eastern third. To facilitate tracking, the two fragments were labeled “A1” and “A2.” Some of the smaller floes had fractured, producing many smaller fragments. Overall, the area of open water had increased slightly. The greatest change was northeast of floe A2. Cloud shadows and small areas of sunshine continued to provide a false hummocky texture to parts of the image. A large floe on the northwest corner of the image was labeled “C.”

The image collected 2 days later on July 9, 2009 (7/09 on fig. 20), showed an area that was cloud free. As in earlier images, most of the large composite ice floes showed minimal changes. Little change had occurred at the north-south fracture separating floes A1 and A2. Additional smaller floes had fractured, producing more fragments. The area of open water had continued to increase. The greatest change remained northeast of floe A2.

The image collected 5 days later on July 14, 2009 (7/14 on fig. 20), showed an area that was more than 90-percent cloud covered. However, the clouds were thin to the extent that they ranged between transparent and translucent. The most significant change that is visible from the previous image is at the north-south fracture separating floe fragments A1 and A2. Fragment A2 had begun to rotate toward the southeast, and as much as 0.2 km of separation had occurred at the north end of the fracture.

The image collected the next day, July 15, 2009 (7/15 on fig. 20), showed an area that was 100-percent cloud free. The most significant change was still the result of the continuing rotation of floe A2 along the north-south fracture separating floes A1 and A2. Maximum separation was about 0.6 km. Fragment A2 had drifted northward as it rotated toward the southeast. Melt ponds were the most significant surface feature on the ice floes.

The image collected on July 22, 2009 (7/22 on fig. 20), showed significant cloud cover and a band of lost data. In the 7 days between images, ice floes A1 and A2 each had broken into multiple fragments, although the fragments remained in close proximity to each other. The entire ice mass had rotated clockwise about 15°.

The image collected 3 days later, on July 25, 2009 (7/25 on fig. 20), showed an area that was cloud free. There was very little difference between ice floes A1 and A2 and their daughter products shown in this image compared with the previous image. Similarly, floes B and C showed minimal change since they were first described.

The image collected 1 day later, on July 26, 2009 (7/26 on fig. 20), had a band of cloud cover in mid-image. Compared to the previous image, this image shows very little difference between ice floes A1 and A2 and their daughter products. Similarly, floes B and C showed very little change.

The image collected 5 days later, on July 31, 2009 (7/31 on fig. 20), is cloud free. Surface snow cover and melt pond area were significantly less than previously shown. Since July 26, 2009, the amount of open water had increased by about 15 percent. Only small changes had occurred to ice floes A1 and A2 and their daughter products. While ice floe B continued to show very little change, the southwest side of ice floe C had fractured.

The final image in this sequence, collected 11 days later on August 11, 2009 (8/11 on fig. 20), is also cloud free. Open water covered about 40 percent of the area shown. Since July 31, all of the labeled ice floes except C had developed many fractures. The length of the largest fragment of floe B exceeded 3.0 km. Subsequent imagery showed the continued decrease in size of all ice floes. The final image, not shown here, was collected 34 days later, on September 14, 2009.

Summary

Since 1997, the Global Fiducials Program (GFP) has collected remotely sensed high-resolution United States National Imagery Systems (USNIS) imagery to monitor Arctic sea ice behavior. Initial imagery collection efforts documented the Surface Heat Budget of the Arctic Ocean (SHEBA) Project, in which the Canadian icebreaker, Des Groseilliers was purposely frozen into sea ice of the Beaufort Sea. This Arctic sea ice project spanned 404 days, from August 25, 1997, to October 3, 1998, with the icebreaker serving as a platform for experiments and a base for scientific research. More than 100 USNIS images were acquired of the ship and the ice adjacent to it as the ship drifted across the Beaufort Sea.

Fifty-seven (57) of the images were publicly released as literal imagery-derived products (LIDPs) having a resolution of 1.3 meters. An LIDP is an image-like product that is not a direct copy of the original image. LIDPs have had their pixels digitally manipulated so that the pixels’ original characteristics are irreversibly changed while the appearance of the image remains unchanged. This manipulation is done to prevent adversaries from determining the capabilities of the sensors that collected the original image.

Beginning in 1999, six static sea ice sites in the Arctic Ocean were imaged for periods as long as 15 years in order to better understand sea ice dynamics, distribution, age, melt history, and annual variability. The sites were located (1) in the Chukchi Sea; (2) offshore of Barrow (now Utqiagvik), Alaska; (3) in the Beaufort Sea; (4) in the Canada Basin; (5) in the Fram Strait; and (6) in the East Siberian Sea. Between 1999 and 2014, the number of static site images approved for public release was 884.

In 2009, the U.S. Geological Survey (USGS) began to track and repeatedly capture imagery of the same ice over time as it drifted across the Arctic Ocean. The study method was to track telemetering buoys at various locations across the Arctic. Between 2009 and 2014, 38 buoys were monitored for periods ranging from less than 2 months to more than 17 months. The buoys and associated sea ice drifted across hundreds or, in some cases, thousands of kilometers of the Arctic Ocean during the periods of collection. A total of 1,734 LIDP images derived from USNIS images collected from tracking 38 unique buoys have been approved for public release.

In 2013 and 2014, in support of the Seasonal Ice Zone Reconnaissance Surveys (SIZRS) Program, the USGS imaged sites in the Beaufort and Chukchi Seas located at every degree of latitude between 70° and 80° N. along a north-south transect. This was done to track and understand the interplay among the ice, atmosphere, and ocean contributing to the rapid decline in summer ice extent that has occurred in recent years. In support of this goal, more than 300 sea ice LIDP images have been approved for public release.

For each of these efforts, resulting sea ice imagery products approved for public release are posted on two websites hosted by the USGS for the Global Fiducials Program: