By inspecting environmental satellite imagery, analysts from a NOAA NESDIS satellite product group created a Northern Hemisphere snow and ice map from November 1966 until the U. S. National Ice Center (NIC) took over production in March 2008. The next two sections, NIC Derivation Techniques and NESDIS Satellite Product Group Derivation Techniques, describe how the data are processed by the two groups.
NIC Derivation Techniques (March 2008 to present)
Method
When NIC began producing the IMS product in 2008, there were changes in the way sea ice was mapped to create the daily product. Snow mapping, however, did not change, and is as described in the section in the NESDIS Satellite Product Group Derivation Techniques section.
For sea-ice mapping, the IMS product is manually generated by an analyst looking at all available satellite imagery, at output from satellite ice mapping algorithms, and at other data sources. The analyst begins with a map from the previous day to initialize the process. Input satellite data and fields are sampled to a standard 6144 x 6144 grid (~4 km per grid cell). The analyst then integrates all data sources to create the best representation of sea-ice cover at a 4 km resolution.
A cell is considered ice-covered if more than 40 percent of the 4 km cell is covered with ice, regardless of the ice thickness or ice type. Users familiar with satellite passive microwave sea-ice concentration products that NSIDC distributes, in which a 15 percent concentration threshold is used to define the ice edge, should be aware that the IMS 40 percent threshold is not a threshold applied to passive microwave sea-ice concentration or to any single ice concentration product. Rather, analysts start with yesterday's ice product, judging by eye if a cell or cells are more than 40% ice covered based on available data, and edit appropriately. Analysts use data that are of sufficient resolution to allow the concentration to be reasonably well determined in this way. The completed IMS product is then automatically saved in ASCII, GeoTIFF, and GIF formats.
A further description of the IMS system is available on the NIC web site at their About the NIC IMS Products web page.
In December 2014, NIC began creating a 1km as well as the 4 km and 24 km products. It is produced in the same manner as the 4 km product (S. Helfrich, personal communication).
Generally speaking, there is a new IMS product every day. However, occasionally, the IMS product is not updated for particular regions even though the IMS file date has been incremented by one day. NIC confirmed that this happens when analysts do not have enough information to change the analysis for a region. They may have some data that could be used; but unless the data are sufficient, they will persist the ice boundaries. This can produce areas for which the ice edge seemingly does not change for a day or more when it may actually be changing. NIC would like to flag these areas but are unable to do so at present (S. Helfrich, personal communication to F. Fetterer, October 2016).
Data Sources
For determining sea-ice coverage, IMS analysts consider derived ice charts, modeled ice conditions, and surface observations, as well as visible, passive microwave, and active microwave satellite resources. The use of data sources varies by the timeliness of the data, the resolution of the data, weather conditions, and the time of year. While there is no hierarchy in data sources used for determining ice conditions, analysts tend to prefer visible band imagery. Other satellite sources, such as passive microwave imagers and sounders, Synthetic Aperture Radar (SAR), and scatterometry, are favored when visible imagery is absent or obscured by clouds. In the Arctic, this is quite common. Operational ice charts produced at NIC or by other ice services, modeled ice data, ship observations, oceanographic data, and atmospheric conditions are also considered when satellite sources are analyzed in order to provide context and to support the analyst's interpretation regarding the presence or absence of ice.
Data sources for the IMS product have changed since the 4 km resolution was introduced in February 2004. New satellites and other sources have been introduced to replace those that are no longer available. Metadata that records which imagery was used to generate the snow and ice maps are not kept at this time. Helfrich et al. (2007) include an estimate of the percentage of imagery used from each source. The estimate was made before production was moved from the NOAA satellite products group to NIC in 2008. Table 11 lists the majority of the different input data sources used in IMS production.
Analysts display available sources on large workstation screens, where each day's IMS product is made. As of 2010, the primary visible band imagery that analysts use comes from the Moderate Resolution Imaging Spectrometer (MODIS). Other visible satellite data sources, in 2010, include Advanced Very High Resolution Radiometer visible band (AVHRR-VIS), Geostationary Operational Environmental Satellite (GOES) Imager, Spinning Enhanced Visible and Infrared Imager (SEVIRI), and the Multi-functional Transport Satellite (MTSAT) Imager. AMSR-E 89 GHz brightness temperature at 6.25 m resolution was an important passive microwave data source when the satellite was in service (~2002-2011). Analysts directly interpret areas with high 89 GHz brightness temperatures as areas covered by ice when this interpretation is supported by information from other sources. Other passive microwave sources include the Special Sensor Microwave Imager (SSM/I) derived ice concentrations and Advanced Microwave Sounding Unit (AMSU) derived ice concentrations. When analysts use passive microwave derived ice concentrations, they usually interpret 40 percent to 60 percent concentration in the passive microwave product to be equivalent to about 7/10+ coverage on NIC ice charts.
The automated NOAA ice cover output that make use of AVHRR, SSM/IS, GOES imager, and SEVIRI at the same IMS resolution are also examined as an objective evaluation of ice conditions (Sean Helfrich, citing the work of Peter Romanov, personal communication 08 November 2010). SAR imagery from RADARSAT-2, European Remote Sensing Satellite-2 (ERS-2), Advanced Land Observing Satellite (ALOS) Phased Array type L-band Synthetic Aperture Radar (PALSAR), and Envisat Advanced Synthetic Aperture Radar (ASAR) are used but are not analyzed on the same workstation screen with other IMS data sources. SAR data are examined on adjoining NIC Sea Ice Prediction and Analysis System (SIPAS) workstations and referenced as the IMS analysis is produced. The SIPAS workstations are those used by NIC to produce their weekly regional operational analysis products that use the SIGRID-3 sea ice chart designations.
Derived ice conditions from ice charts and ice edge products are also considered. The US, Canadian, Norwegian, Danish, Russian, German, Swedish, and Japanese ice charting agencies also serve as data sources in the absence of direct satellite data or in areas where only passive microwave derived ice data is available. Passive microwave derived data are especially suspect when conditions are right for surface melting or where ice is new and thin. The use of ice charts is limited due to the infrequency of ice chart production. Operational regional ice charts, that is, those that use the WMO egg code to describe ice conditions, are not available every day. The NIC ice edge products are examined thoroughly but are not used directly for the IMS product due to differences in mission ice identification requirements for each product. The NIC and Canadian Ice Service (CIS) ice edge products attempt to delineate not only where ice is present, but also where any ice is likely, regardless of the concentrations, for safety of navigation. The IMS product attempts to demarcate each 4 km x 4 km grid cell that appears to have more than 40 percent ice concentration as ice covered. An ice edge line created from an IMS field will not match an ice edge line from the regular NIC ice edge product. Helfrich et al. (2007) have information on how NIC operational charts tend to differ from NIC operational IMS products.
Modeled ice conditions from the National Centers for Environmental Protection (NCEP) Marine Modeling and Analysis Branch (MMAB) and coupled Numerical Weather Prediction (NWP) models are also available for analysis, though these are generally used only for context and to understand where areas favor ice formation. Ship reports and ice buoys also enhance the analysis by providing limited ground truth, boundary layer weather conditions, and ice motion information.
New data sources for IMS production are continuously introduced as other advanced visible and passive microwave satellite data, coupled ice models, and more SAR imagery become available. NIC wishes to incorporate ancillary information about data quality in future versions of the IMS. Ideas for this include adding a variable to flag when a cell was not updated so that users know, on a cell-by-cell basis, if the surface type assignment of ice or open water for a given cell was updated or simply carried over from the last analysis. As of January 2017, however, NIC does not have the resources to realize these plans.
NESDIS Satellite Product Group Derivation Techniques (through February 2008)
Through February 2008, the IMS product was manually created by a NOAA NESDIS satellite-product-group analyst looking at all available satellite imagery, automated snow mapping algorithms, and other ancillary data. NESDIS analysts drew snow maps on workstations that displayed these data products and satellite imagery. The visible imagery of the Polar Operational Environmental Satellites (POES) and geostationary orbiting environmental satellites were primary. Moderate Resolution Imaging Spectrometer (MODIS) imagery was used as well. In addition, ground weather observations from many countries were used. Microwave products from POES Advanced Microwave Sounding Unit (AMSU) and the Department of Defense (DOD) DMSP were incorporated into the daily snow and ice chart because, even though they are at relatively low resolution, they allow a view through clouds. A weekly sea ice analysis from NIC, the United States Air Force Snow and Ice Analysis Product, and snow products from the National Operational Hydrologic Remote Sensing Center (NOHRSC) were made available to the analyst, as well as several automated snow detection layers developed by NOAA NESDIS satellite product group and the NOAA National Centers for Environmental Prediction (NCEP). For sea ice, analysts rely first on visible imagery and radar data, then on passive microwave data, followed by the NIC analysis product, depending on the timeliness of the data, the resolution of the data, and the time of year.
The analyst began with a previous day's map to initialize the process. Input satellite data and fields were resampled to the two IMS grids available then: 6144 x 6144 grid (~4 km/pixel) matrix and 1024 x 1024 grid (~24 km per pixel). All resolutions were saved in ASCII format and the 4 km product was saved as GeoTIFF and GIF formats. The ASCII files were built to NCEP specifications because NCEP was the primary user of this product at the time. NCEP created a Binary Universal Form for the Representation of meteorological data (BUFR) format and a GRIdded Binary (GRIB) format from the ASCII output, but these formats are not archived at NSIDC.
Processing Steps at NIC
IMS processing can be broken down into four generalized steps (Helfrich et al., 2019):
- A preprocessing system takes all input products and imagery from their native formats and resolutions and converts them into the IMS formats and resolutions and places them on a internal server.
- An IMS GUI system picks up the processed data from the server at intervals throughout the day and displays the data on the IMS projection.
- The analysts tag locations as snow covered and ice covered over the entire Northern Hemisphere. The GUI system also generates snow depth, ice thickness, and time since last observation via code that analysts can also alter before the analysis is exported.
- Scripts produce final products and distribute final products to proper destinations.
Quality Assessment
The quality of the snow- and ice-cover charts will depend on the availability of clear sky imagery, the georegistration of that imagery, the quality of other input data sources, and the experience of the OSPS analyst. This is a manually created product which uses multiple images to map the snow/ice regions. Surface data is also made available to the analysts to aid with real-time quality control. Regions covered by cloud during the 24-hour analysis period are generally mapped as persistence, taking lower resolution passive microwave data and surface observations into account where possible. Other than grid points in the square array which, from a hemispheric view, fall off the sphere and are flagged as 0 (outside the northern hemisphere), there should be no missing values over the mapped hemisphere.
Sub-grid scale features may not be detected. The documentation for the Northern Hemisphere EASE-Grid Weekly Snow Cover and Sea Ice Extent (Armstrong and Brodzik 2002) and for snow products at Rutgers University Global Snow Lab includes more information on quality assessment, including the following from the Global Snow Lab:
"Despite the shortwave limitations [ ...], the NOAA maps are quite reliable at many times and in many regions. These include regions where, 1) skies are frequently clear, commonly in Spring near the snowline, 2) solar zenith angles are relatively low and illumination is high, 3) the snow cover is reasonably stable or changes slowly, and 4) pronounced local and regional signatures are present owing to the distribution of vegetation, lakes and rivers. Under these conditions, the satellite-derived product will be superior to maps of snow extent gleaned from station data, particularly in mountainous and sparsely inhabited regions. Another advantage of the NOAA snow maps is their portrayal of regionally-representative snow extent, whereas maps based on ground station reports may be biased, due to the preferred position of weather stations in valleys and in places affected by urban heat islands, such as airports."
See the NSIDC Sea Ice Index: Interpretation Resources for Sea Ice Trends and Anomalies for a general discussion of passive microwave imagery for sea ice extent. Note that while the NOAA IMS product makes use of both passive microwave and visible band imagery to map ice extent, the NSIDC product, Northern Hemisphere EASE-Grid Weekly Snow Cover and Sea Ice Extent, uses only passive microwave for ice extent.