National Ice Center Sea Ice Climatology Introduction: The Environmental Working Group (EWG)-Arctic Climatology Project, tasked with producing an Arctic sea ice atlas, identified nineteen years (1972-1990) of these highly detailed classified ice analyses as a scientifically significant database. NIC originally produced these data as hard copy charts. It was proposed in 1996 that these sea ice charts be re-analyzed, quality controlled and then converted into digital data by manual digitizing in a Geographic Information System (GIS). The analyses did not necessarily cover the entire Arctic for any given week, but it was obvious that the added detail resulting from larger scale mapping procedures in the classified charts would prove beneficial to the final climatological database.
Original NIC classified sea ice charts were rarely completed by analysts as coherent, smoothed sea ice analyses like those traditionally produced in the unclassified realm. More typically, sea ice analysts simply compiled "pieces" of sea ice information from a wide variety of national security sources and plotted these data as point observations, tabular information or annotational information. Because of this, NIC was required to undertake an enormous effort of "re-analyzing" these charts to produce the most realistic and complete "picture" of sea ice conditions for a given "work chart".
Data Analysis: The objective of the analysis effort was to review, analyze, quality control and record metadata for 608 classified ice charts. The effort encompassed nineteen years of ice analysis charts that covered the Arctic marginal seas, the North Pacific, Sea of Okhotsk and Sea of Japan. Ice analysts worked primarily with two types of basic ice analysis charts: weekly work charts and publication charts from previously printed annual ice atlases based upon the NIC 7-day ice chart series. The most representative charts for mid-month ice conditions were selected. The following items were reviewed as part of the process.
- Continuity of ice conditions between sequential ice analysis charts
NIC required that final corrected charts be in a condition which was suitable for manual digitizing, with the overall goal of developing an Arctic sea ice database derived from classified data.
NRL established several standard basemaps to accommodate the large variety of classified charts (various projections and scales) early in the project. NRL produced these basemaps using NIMA’s World Vector Shoreline (WVS). The digitizing team registered the re-analyzed charts in ARC/INFO’s Arcedit module. Once digitized, the digitizing team built topology for and cleaned the polygon coverages. NRL then re-projected the ice charts to polar stereographic tangent at 60N, 180W, which allowed the technician to "mapjoin" all charts representing a given week, creating a "composite" view of the Arctic for a representative week for each month for the years 1972 – 1990. The team then used a custom-built "ice tagging" tool to add polygon attributes to the polygons. These attributes described the ice conditions for that polygon based on the analyst’s original annotations on the working chart.
NIC then used a GIS data extraction routine, written in Arc Macro Language (.aml). This routine specifically extracted the "total concentration" (CT) value from the polygons. With this done, NIC then converted the GIS coverages to binary gridded fields (ARC/INFO format), using the GRID module in ARC/INFO. They created grids based on the CT value alone. The output grid cell size was 12.5 km (the units of the coverages were meters in a polar stereographic projection). An identical "CT extraction" routine allowed NIC to create total concentration grids from the unclassified SIGRID data as well, then facilitating the merging of data sets derived from both classified and unclassified sources, and ensuring complete Arctic coverage for a given representative monthly gridded sea ice chart. This technique facilitated the computation of grids depicting climatological attributes based on the values from each grid from a month, from all the available years.
Climatological Grids Derived From Merged Charts: In April, 1999, the National Ice Center (NIC) completed production of the climatological data sets extracted from previously classified sea ice data in support of the Gore-Primakov Commission’s Environmental Working Group. NIC determined that the most useful climatological product derived from the EWG data set would be monthly Arctic grids. NIC created a complete monthly data set by merging the monthly grids based upon the weekly ice charts with derived grids from classified sources. Legacy classified and unclassified Arctic sea ice analyses were re-analyzed, revised and quality controlled, where necessary, in an effort to provide the most accurate depiction of the annual cycle of sea ice formation and decay. These are monthly climatological gridded fields depicting (1) median total ice concentration and (2) the occurrence of sea ice.
NIC employed GRID again to compute the climatology from all available gridded fields. The first statistic computed was "probability of occurrence of any ice", which is an enumeration of the number of years for a given week that ice of any concentration occurred in a given pixel (total number of years being 19). The possible grid values for this output grid are the integers from 0 through 19. The next statistic, median total ice concentration, was obtained by computing the median CT value for a given pixel for all available grids for a given month (i.e. half the data in the 19-value sample fall on or above this median value, half fall on or below this value). (See Appendix B for file naming convention). NIC is working with the National Security Agency to seek the release of monthly gridded fields of total ice concentration created from the merged data set.
Quality Assurance of Climatology Grids: NIC revised the resulting climatology by deleting or joining any small sea ice anomalies to the ice edge or within the ice field itself and merged them with surrounding attributes. NIC smoothed continuous regions of sea ice concentration and probabilities to more accurately depict ice edge, marginal ice zone, and the concentrated ice pack. Obvious errors were compared to original charts and subjectively edited as deemed fit. In particular, NIC revised fast ice extent using knowledge of local ice regions that were not known in early ice analysis years.