An optimal estimation algorithm to derive Ice and Ocean parameters from AMSR Microwave radiometer observations
Global multispectral microwave radiometer measurements have been available for several decades. However, most current sea ice concentration algorithms still only takes advantage of a very limited subset of the available channels. Here we present a method that allows utilization of all available chan...
| Main Authors: | , , , |
|---|---|
| Format: | Conference Object |
| Language: | English |
| Published: |
2016
|
| Subjects: | |
| Online Access: | https://orbit.dtu.dk/en/publications/7bee77e1-8b06-45e2-aef8-7974e649e401 http://lps16.esa.int/page_session185.php#2383p |
| Summary: | Global multispectral microwave radiometer measurements have been available for several decades. However, most current sea ice concentration algorithms still only takes advantage of a very limited subset of the available channels. Here we present a method that allows utilization of all available channels as well as the combination of data from multiple sources such as microwave radiometry, scatterometry and numerical weather prediction. Optimal estimation is data assimilation without a numerical model for retrieving physical parameters from remote sensing using a multitude of available information. The methodology is observation driven and model innovation is limited to the translation between observation space and physical parameter space Over open water we use a semi-empirical radiative transfer model developed by Meissner & Wentz that estimates the multispectral AMSR brightness temperatures, i.e. horizontal and vertical polarization at channels between 6 and 89 GHz as a function of a limited set of physical parameters, i.e. atmospheric water vapor, cloud liquid water, wind speed, surface and air temperature. This type of model is ideal for optimal estimation applications because of its limited set of free variables. The atmosphere/open water model is adapted to simulate the atmosphere over sea ice and to work over intermediate ice concentrations by linear scaling of the surface emissivity and surface effective temperature. The simulation of the surface brightness temperature of sea ice requires a separate forward model. Important physical parameters include snow layering, scattering in the snow and sea ice, effective temperature and ice concentration. Here we are testing and evaluating different models. Ice emissivity model development and validation is based on time series of data from Ice Mass Balance Buoys which is co-located with AMSR data and ERA Interim data. A priori knowledge of each of the physical parameters is used to constrain the solution and improve the retrieval. We test two different a ... |
|---|