Assimilation of SAR Ice and Open-Water Retrievals in Sea Ice Numerical Prediction System

The Environment and Climate Change Canada’s (ECCC) Regional Ice-Ocean Prediction System (RIOPS) is an operational short-range numerical ice forecasting system covering the entire Arctic region. The assimilation component of the system utilizes multiple satellite data sources in order to produce accurate sea ice concentration analyses and generate improved short-term forecasts. Currently, SSM/I, SSMIS, AMSR2 passive microwave, and ASCAT observations as well as optical AVHRR data and Ice Charts manually produced by the Canadian Ice Service (CIS) are used. However, passive microwave and scatterometer observations have relatively coarse (25–50 km) resolution, optical data are dependent on solar illumination and clouds, and the CIS products only cover a relatively small portion of the RIOPS geographical domain.

Spaceborne synthetic aperture radar (SAR) high resolution (at or below 50 m) images from the ongoing satellite missions such as Canadian RADARSAT-2 currently provide the most reliable information on sea ice conditions. Dual-polarization RADARSAT-2 HH-HV ScanSAR imagery is the main source of data for operational production of Ice Charts at CIS. The upcoming Canadian RADARSAT constellation mission (RCM) equipped with three SAR platforms will further increase the amount of SAR data over the Arctic region. Therefore, there is a strong interest in making use of the large quantities of SAR observation in the sea ice analysis component of RIOPS.

The main objective of this study is to assess the impact of automatically derived information from SAR data in the sea ice concentration analysis system over a 12-month period.

More than 7,000 RADARSAT-2 HH-HV images acquired over the Canadian and adjacent waters being monitored by CIS were collected for year 2013. Our previously developed technique for automated detection of ice and open water from SAR with very high accuracy of 99.8% [1], [2] was utilized in order to derive ice and open water retrievals from the collected SAR images. Each retrieval represents a binary (ice or water) value at the scale of 2.05 km × 2.05 km which is smaller than the average resolution of RIOPS (~5 km). In order to assess the impact of the SAR retrievals in the sea ice analysis system, the following data assimilation experiments over the year 2013 were conducted:

1. Control experiment with all currently used satellite data sources (i.e. SSM/I, SSMIS, AMSR2, AVHRR), except CIS manually derived ice charts from SAR data (which indirectly represent SAR data).
2. Charts experiment. Similar to Control experiment, but with CIS Ice Charts included.
3. Six SAR experiments with different options. Similar to Control experiment, but with our SAR retrievals included.

Verification of sea ice concentration analyses was conducted against three data sources: (1) CIS weekly Ice Charts, (2) US National Ice Center (NIC) biweekly Ice Charts, and (3) Interactive Multisensor Snow and Ice Mapping System (IMS) ice extent product. Our results demonstrate that impact of SAR retrievals in sea ice analysis system is similar to the impact produced by the CIS Ice Charts. Main improvements in the SAR runs were found near land. This is similar to the spatial distribution observed with Charts run, as mostly the same images were used for Ice Chart production. More detailed verification results for different Arctic regions and seasons will be discussed in the presentation.

With the launch of the three-satellite RCM in 2018, a significantly larger volume of SAR observations over the Arctic region will become available. Our retrieval techniques will be adapted to the RCM data stream. The data assimilation results obtained for RADARSAT-2 data suggest that the increased coverage of RCM will considerably improve the accuracy of automated sea ice analysis component of RIOPS.

[1] A. S. Komarov, and M. Buehner, “Adaptive probability thresholding in automated ice and open water detection from RADARSAT-2 images,” IEEE Geoscience and Remote Sensing Letters, vol. 15, no. 4, pp. 552-556, April 2018.

[2] A. S. Komarov, and M. Buehner, “Automated detection of ice and open water from dual-polarization RADARSAT-2 images for data assimilation,” IEEE Transactions on Geoscience and Remote Sensing, vol. 55, no. 10, pp. 5755-5769, October 2017.