About using dual-polarization multi-frequency microwave observations for improved identification of thin ice and retrieval of its thickness
Zusammenfassung der Projektergebnisse
Helicopter-borne scatterometer data and air-borne SAR data are investigated to enhance knowledge about microwave frequency and polarization combinations to be used preferably for identification and thickness retrieval of thin sea ice, i.e sea ice with a thickness below about 0.3 m. The scatterometer data were obtained with Multi³Scat instrument of the University of Hamburg in the central Arctic Ocean during August / September 2007, an unusual warm year with delayed freeze-up. The SAR data were obtained in March 1998 during the “Arctic Radiation and Turbulence Interaction Study” northwest of Svalbard in the Fram Strait under typical winter conditions. Based on the scatterometer data a thin ice classification algorithm was developed, capable to separate nilas and grey ice from open water and old (thick) ice. The classification was developed and evaluated with co-incident visual imagery and flight logs. In a next step a simple radar backscatter model was developed. With this model thin ice thickness can be estimated with the scatterometer data as input using a statistical approach. Both the classification and the thickness retrieval approach were applied to scatterometer data and to satellite data. The air-borne SAR data were in processed form. These were investigated first with a series of visual image interpretation approaches before different automated classification approaches were applied. In contrast to the Multi³Scat data only two frequencies at L- and C-Band (1.25 GHz and 5.3 GHz) are available here. However, in addition to co-polarization data here also cross-polarized data as well as the phase information can be used for the interpretation and classification. The main results of the project are: • Usage of multiple frequencies and polarizations is beneficial for the discrimination of thin ice from thick ice and open water under both typical winter and fall conditions. • A set of functional relationships between co-polarized radar backscatter at S- to Ku- Band and incidence angle has been developed for ice types nilas, grey ice and old ice under fall freeze-up conditions. • The potential of using a simple radar backscatter model as a tool to derive thin ice thickness from radar backscatter measurements is limited, particularly during the unusual warm fall freeze-up conditions encountered during the Multi³Scat measurements. Ice thickness retrievals works more promising and more accurate using the above frequencies for shallow than steep surface incidence angles. • The ice thickness retrieval approach reproduced the thickness of the thinnest ice observed during the Multi³Scat measurements within 0.02 m as compared to in-situ observations. • The major limitation of the developed simple ice thickness retrieval approach is its dependency on a correct classification of thin ice. Because the classification is developed for fall conditions in the Arctic it is neither working under Arctic or Antarctic winter conditions; it also shows problems in polynya areas covered by frazil and grease ice. • The results of a theoretical modelling study to relate sea ice thickness with the microwave surface emissivity at frequencies from about 1 to 100 GHz indicate a sharp increase of emissivity with ice thickness over the first 0.2 m at most frequencies used currently by satellite microwave sensors like AMSR-E. Above that thickness the penetration depth into saline sea ice is not large enough to cause any thickness related change in the emissivity.
Projektbezogene Publikationen (Auswahl)
-
Multi³Scat - A helicopter-based scatterometer for snow-cover and sea-ice investigations. Geosci. Rem. Sens. Lett., 6(4)
Kern, S., M. Brath, R. Fontes, M. Gade, K.-W. Gurgel, L. Kaleschke, G. Spreen, S. Schulz, A. Winderlich, and D. Stammer
-
. Retrieving ice concentration from SMOS. Geosci. Rem. Sens. Lett., 8, 2, 283-287
Mills, P., and G. Heygster
-
Sea Ice Classification During Freeze-up Conditions with Multifrequency Scatterometer Data. Trans. Geosci. Rem. Sens., 51(6-1), 3336-3353
Brath, M., S. Kern, and D. Stammer
