Remote sensing of Antarctic clouds with infrared and passive microwave sensors
The importance of the polar regions and of clouds for the global climate is widely recognized. But clouds in polar regions are difficult to detect in infrared satellite images because the brightness temperatures of the surface and clouds are very similar. Sensors operating at visible wavelengths are...
| Published in: | Meteorologische Zeitschrift |
|---|---|
| Main Authors: | , |
| Format: | Article in Journal/Newspaper |
| Language: | English |
| Published: |
Borntraeger
2002
|
| Subjects: | |
| Online Access: | https://doi.org/10.1127/0941-2948/2002/0011-0021 https://doaj.org/article/907016270af84936bbb5f59fc7a0dfe3 |
| Summary: | The importance of the polar regions and of clouds for the global climate is widely recognized. But clouds in polar regions are difficult to detect in infrared satellite images because the brightness temperatures of the surface and clouds are very similar. Sensors operating at visible wavelengths are less suited because they need daylight and can therefore not operate in the polar night. Three methods have been developed and tested to retrieve clouds over polar regions. Only data of infrared and passive microwave sensors of the DMSP satellites are used, the algorithms are therefore independent of the daylight and can also be applied in the polar night. The first approach classifies single infrared images of the sensor OLS over the Weddell Sea into 'cloud', 'sea ice' and 'open water'. First, the images are subdivided into segments. For each segment, four texture parameters are evaluated. The algorithm uses a Learning Vector Classification Neural Network. The trained network classifies test data from March 1992 with an accuracy of 88%. Inclusion of SSM/I sea ice information improves this result to 91%. The second, qualitative method detects nonstratiform clouds over the Antarctic continent using differences of nearly simultaneous infrared images collected by different DMSP spacecraft. If observed under different incidence angles, nonstratiform clouds show different infrared brightness temperatures. Validations with visible data and with data of the passive microwave sounder SSM/T2 also onboard the DMSP platforms show good agreement. Thirdly, a multi sensor approach is outlined including the first two methods and in addition the SSM/I-derived atmospheric parameters total water vapor, liquid water path (LWP) and the cloud signature, a recently proposed quantity for cloud characterization over sea ice. The cloud features high/middle/low, stratiform/nonstratiform and thickness (expressed as LWP or cloud signature) can be used for cloud description. The method is demonstrated in a case study with data recorded on the ... |
|---|