| Abstract: |
Space-borne microwave radiometers have proved to be the best remote sensing tool for retrieval of dry snow characteristics, especially the terrestrial snow water equivalent (SWE). The two main land-cover types reducing accuracy of SWE retrieval are forest cover and frozen lakes. The physical characteristics of snow on lake may differ substantially from those for snow on terrain. The emission behavior of the snow/ice/water system is different from that of the snow/soil system and, additionally, existence of water on top of ice strongly affects brightness temperatures (TB). Due to abundance of lakes in northern regions the effect of lakes to the satellite-derived brightness temperatures should be accounted for in order to guarantee reasonable accuracy in SWE retrieval. A literature survey of microwave radiometry of snow on terrain and also of snow on lake ice is available in (Hallikainen et al., 2017) with an additional 19 references in (Hallikainen et al., 2016).
We have conducted a multiyear (2011-2014) airborne microwave radiometer measurement program of snow on lake ice using frequencies of 1.4 to 36.5 GHz, vertical and horizontal polarization (Hallikainen et al., 2016). Our data comprise observations along a test line over two lakes in southern Finland, and include data also for adjacent land areas, whose land-cover ranges from forest to short vegetation. A variety of snow and ice conditions were encountered, covering early winter, mid-winter and late winter conditions, and the melting season.
Previously, we have in our reporting focused on observations and modeling of brightness temperatures for snow on lake ice. In this presentation the observed brightness temperatures for snow on lake ice and snow on terrain are compared, especially at 6.9 and 36.5 GHz. Our observations strongly suggest that, in the 6.9 to 36.5 GHz range, brightness temperature behavior of the snow/terrain system is quite different from that of the snow/ice/water system.
Snow on land: Brightness temperature depends little on frequency in early winter; then, gradually TB starts to decrease with increasing dry snow thickness and snow grain size, and with increasing frequency, due to volume scattering. Forests decrease sensitivity of TB to snow.
Snow on lake ice: In early winter TB is strongly dependent on frequency, decreasing with decreasing frequency for thin snow and thin ice layers (emission from water, no volume scattering). The land/ice difference may be 50 K at 36.5 GHz and over 100 K at 6.9 GHz. In late winter, the frequency dependence of TB starts to reverse due to increased snow grain size (volume scattering). During the melting season, the frequency dependence of refrozen snow on lake ice is half-reversed and land/ice difference at 36.5 GHz may be 60 K and only 10-15 K at 18.9 GHz even for a fairly thin layer of snow on ice and a substantially thicker snow layer on land. This behavior decreases accuracy of SWE retrieval unless accounted for.
References
M. Hallikainen, J. Lemmetyinen, L. Jiang (2017), ”Snow properties from passive microwave,” in Comprehensive Remote Sensing (Editor-in-Chief S. Jiang), Remote Sensing of Hydrological Cycle, Elsevier Publishing Company.
M. Hallikainen, M. Vaaja, J. Seppänen, J. Praks, J. Lemmetyinen (2016), “Snow on lake ice: Overview of a multiyear airborne radiometer data collection program and related modeling efforts,” Proc. MicroRad 2016, pp. 98-103, Espoo, Finlad 11-14 April 2016. |
