Preliminary algorithm for retrieving Sea Ice Concentration from the Microwave Radiometer on SAC-D/Aquarius
Abstract
This paper presents the first version of the algorithm to estimate Sea Ice Concentration from brightness temperatures measured by the MWR (MicroWave Radiometer) on board of SAC-D/Aquarius satellite, launched in June 2011. The MWR is a three channel microwave radiometer, 23.8GHz (H-pol) and 36.5GHz (V-pol and H-pol), with 8 beams for channel of about 50 km in size. The corresponding 8 horns of each channel are in a pushbroom configuration with two different incident angles: 52° and 58°. The algorithm for ice estimation presented here is based on the Bootstrap and Nasa Team (NT) ones, that use gradient and polarization ratios of the obtained brightness temperatures for their calculations. Usually these algorithms use the band of 19GHz for calculating the polarization ratio (PR), while for the gradient ratio (GR) use 37GHz and 19GHz in V-pol, and additionally other bands such as 23GHz and 85GHz are used as clime filters to improve the accuracy of the estimations obtained. In this way they obtain a distribuition of data for polar areas very similar to a triangle in a PR vs GR plot where it is possible to estimate not only the ice concentration but also classify the ice type. For the MWR we adapted this idea to the characteristics of the sensor. We use the 36.5GHz channel for calculating PR and 36.5GHz and 23.8GHz in H-pol for calculating GR. The combination of those MWR data produces a (PR,GR) distribution very noisy as we could verified using Windsat data and comparing the obtained (PR,GR) plots using NT and our definition of these quantities. As the 23.8GHz and 36.5 GHz bands are very sensitives to water vapor variation and sea roughness respectively, the use combined of these bands is very sensitive to both environmental variables. To reduce this effect we develop our algorithm as auto calibrated one. We summarized the way as the algorithm is applied to each beam: 1) We make a scatter plot in a PR-GR diagram with the data of one day corresponding to higher latitudes than 60°. 2) We determine the (PR,GR) points corresponding to Open Ocean, First Year Ice and Old Ice by the means of a procedure that takes into account the density and the triangular form of the ideal distribution of this kind of data. 3) We estimate the brightness temperature corresponding to the three mentioned above points and depurate them using statistical procedures. 4) With these three points we build the typical triangle like NT algorithm. 5) The isoconcentration lines are obtained using a lineal forward calculation if the (PR,GR) points from the brightness temperatures assuming as input different ice concentrations. 6) For each beam the calculated (PR,GR) point is put into the diagram described above and its relative location determines its corresponding ice concentration. In order to improve the accuracy of our estimations we present some test using GDAS data and statistical algorithms applied to obtain a daily ice concentration product. Additionally, a comparison between our outputs and the corresponding estimations from NSIDC is presented.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2011
- Bibcode:
- 2011AGUFMOS23A1625M
- Keywords:
-
- 0750 CRYOSPHERE / Sea ice;
- 4275 OCEANOGRAPHY: GENERAL / Remote sensing and electromagnetic processes