|
||||||
|
|
||||||
  |
 |
|
|
 |
|
|
|
Terms of |
Cheryl Bertoia
U.S. National Ice Center
4251 Suitland Road, FOB 4
Washington, D.C. 20395-5120
(voice) 301-457-5314 ext. 302 (fax) 301-457-5300
e-mail bertoiac@natice.noaa.gov
Sea ice analysis techniques under development at the National Ice Center (NIC) are briefly discussed in order to offer the opportunity for international collaboration in future work.
The NIC established a Science Program in the spring of 1997 when a Senior Scientist was added to the NIC staff. The main goals of the program are to:
One of the first accomplishments of the science team was the creation of a science plan (available at http://www.natice.noaa.gov). This document outlines future goals for scientific development, and ties into the planned evolution of the NIC sea-ice product suite. The NIC science and development team efforts can be divided into three development thrusts: techniques for microwave sensors (synthetic aperture radar and passive microwave), other sensor techniques and transition/ice analysis system development.
RADARSAT In 1998, two sea ice classification systems were delivered. Florence Fetterer developed the Multi-Year Ice Mapping System (MIMS) at the University of Colorado. MIMS is designed to quickly map old ice in uncalibrated SAR images. The technique uses a local dynamic thresholding algorithm and automatically maps and contours multi-year ice [1]. MIMS is currently undergoing test and evaluation at the Naval Research Laboratory at the Stennis Space Center in Mississippi.
The ARKTOS (Advanced Reasoning using Knowledge for Typing Of Sea ice) system is a sea ice classification system that incorporates image processing and knowledge based rules to interpret RADARSAT SAR images [2]. ARKTOS first segments the input image using Watershed region growing technique based on image gradients, and subsequently merges regions based on area, average intensity, and strength of common boundaries [3]. ARKTOS then computes attributes for each contiguous region (henceforth, feature) such as area, average intensity, and shape and texture measures. Given these measurements, facts regarding each feature are formed by quantizing the values into symbols. For example, if the feature's average intensity is less than 50.0, then intensity (feature) = "black". The decision points (or thresholds) were determined and refined via visual inspection of features. Next, during the classification phase, the Dempster-Shafer rule-base engine reads the facts and matches rules by looking for satisfied conditions. After matched rules are fired, the engine combines the evidence and gives the belief and plausibility values for a feature to be in one of four surface types: multi- year ice, first-year ice, open water and unknown. Finally, ARKTOS assigns the feature to the surface type with the highest product of the plausibility and belief values.
SSM/I The NIC science team evaluated the existing suite of sea ice concentration algorithms for the Special Sensor Microwave Imager [4], and modified the operational sea ice algorithm. A passive microwave algorithm was developed which uses a principal components combination of SSM/I brightness temperatures and NIC provided local ice conditions from visible and infrared data to improve global sea ice concentrations [5].
Two post-doctoral fellows were added to the NIC science program in the spring of 1999. An early accomplishment was the implementation of an algorithm to track ice motion using 85 GHz SSM/I [6]. The algorithm provides daily coverage over the Arctic basin and surrounding ice-covered regions during non-melt conditions (October - April). Motion estimates can be improved and their effectiveness extended, via an optimal interpolation (OI) assimilation method, by combining the SSM/I-derived motions with motion estimated from a sea ice model [7]. This OI approach will be used to assimilate SSM/I motions into the Navy's next generation coupled dynamic/thermodynamic sea ice forecasting model.
ENVISAT The NIC is preparing for the advent of Envisat by participating in two ESA approved projects, one with the Canadian Ice Service and one with the Jet Propulsion Laboratory. We plan to study multiple polarization C-band SAR signatures of sea ice for application to operational mapping of sea ice type and concentration. We will initially use JPL's polarimetric AIRSAR system, which has acquired several data sets over sea ice in the Beaufort and Bering Seas to study polarimetric scattering signatures of sea ice at different incidence angles (20 to 50 degrees). After determining the characteristics (relative and absolute backscatter at different polarizations) of sea ice signatures, we will determine the differences among various ice types and open water, expanding observations with an advanced polarimetric sea ice model that was developed at JPL. Eventually, we will develop new techniques for operational sea ice mapping using polarimetric SAR data.
MODIS MODIS is the key instrument aboard the Terra (EOS- AM ) satellite, a new NASA satellite scheduled for launch in mid-November 1999. MODIS will view the entire surface of the Earth every 1-2 days, making observations in 36 co-registered spectral bands, at moderate resolution (0.25 - 1 km). MODIS has great potential to improve automated ice and cloud analysis of the polar regions. The NIC Science Team is working with the Naval Postgraduate School to test the existing NASA Science Team algorithm, then will research better regional thresholds for the polar regions. The existing NASA Science Team ice algorithm provides a binary ice/no-ice product. NIC is working with the University of Colorado to develop a MODIS algorithm to provide ice type information.
Two NIC post-doctoral fellow projects involve study of climate in the Arctic and Antarctic. The first project involves the development of a three-dimensional, coupled ice-atmosphere model. Most Arctic models obtain atmospheric temperature and downward radiative fluxes as input variables from a reanalysis or weather forecast model (as PIPS does). By computing these quantities internally, the proposed model allows the atmosphere to respond to the changing state of the sea ice surface and the sea ice to respond to the changing state of the atmosphere. Thermodynamic coupling between the atmosphere and ocean surface has been shown to have a significant impact on the sensitivity and variability of the climate on global and regional scales. The goal for this model is to examine the impact of thermodynamic coupling on the Arctic climate. Although these results may not apply directly to NIC's interests in sea-ice forecasting, the study will provide valuable experience for the seasonal ice forecasting project scheduled to begin next year.
Another project involves the study of Antarctic ice sheets and their effect on global mean sea level through the blending of TOVS sounding data with surface wind products to estimate averaged precipitation rates over the Southern Ocean and Antarctic continent. These data products will be used to describe interannual moisture variability in the Southern Hemisphere. Other work involves a project to assimilate ice motion vectors into the Navy's next generation coupled dynamic/thermodynamic sea ice model , as described above.
In 1998, the NIC began considering upgrades to the existing suite of image processing and GIS hardware and software. A study was commissioned and resulted in the delivery of documentation regarding requirements for the new system, and capabilities of commercially available software to meet these demands. The NIC is now in the process of re-evaluating the system requirements in preparation for building the new system. An important aspect of the design of the new system was taking into consideration the need to test and easily transition science developments into the operational environment.
[1] Fetterer, Florence, Cheryl Bertoia, and Jing Ping Ye. 1997. Multi-year ice concentration from Radarsat. Proceedings, International Geoscience and Remote Sensing Symposium , Singapore (CD ROM).
[2] Bertoia, C., D. Gineris , K. Partington, L.K. Soh and C. Costas Tsatsoulis. 1999. Transition from research to operations: ARKTOS - A knowledge-based sea ice classification system. Proceedings, International Geoscience and Remote Sensing Symposium, Hamburg, Germany, 28 June - 2 July (CD ROM).
[3] J.M. Gauch. 1999. Image segmentation and analysis via multiscale gradient watersheds. IEEE Transactions on Image Processing 8 (1) (January).
[4]Partington, K. and C. Bertoia. 1999. Evaluation of special sensor microwave imager sea-ice products. Proceedings, International Geoscience and Remote Sensing Symposium, Hamburg, Germany, 28 June - 2 July (CD ROM).
[5] Partington, K. 1999. A data fusion algorithm for mapping sea ice concentrations from special sensor microwave imager data. IEEE Transactions on Geoscience and Remote Sensing 38 (4) (July): 1947-1958, Part 2.
[6] Meier, W. N., 1999. Remotely-sensed sea ice motion from SSM/I imagery: Evaluation of available algorithms for operational applications. NAVICECEN Technical Report (March 18).
[7] Meier, W.N., J.A. Maslanik, and C.W. Fowler. 2000. Error analysis and assimilation of remotely-sensed ice motion within an Arctic sea ice model. Journal of Geophysical. Research-Oceans 105 (C2) (February): 3339-3356.
