JP1.21
Detection of thin cirrus using MODIS 1.38 micron reflection
John K. Roskovensky, Univ. of California, Los Angeles, CA; and K. N. Liou
Detection of Thin Cirrus Using MODIS 1.38 µm Reflection
Roskovensky, J. and Liou, K. N., Department of Atmospheric Science, University of California, Los Angeles, California
The 1.38 µm channel on board MODIS may have significant advantage in detecting thin cirrus over existing methods due to its high sensitivity to upper tropospheric clouds and a nearly negligible sensitivity to low level reflectance. To investigate this potential, three different cloud schemes are employed. One based on the MODIS Cloud Mask Product (MOD35) which utilizes both visible and 1.38 µm reflectance thresholds. The second is a modified version of the cloud phase detection scheme presented in the paper by Ou et al (1996) that incorporates a series of visible, near and far infrared reflectance, ratio, and difference tests. The third scheme simply relies on various experimental 1.38 µm reflectance thresholds to detect high cloud and several tests from the other two schemes to identify other cloud types. For validation purposes, six MODIS cases over the ARM SGP site were chosen for their unique and different cloud structures primarily by millimeter wave radar reflectivity from the central facility in Lamont, OK and visible imagery. The cases include sparse thin cirrus, thick and continuous cirrus, overcast mid-level cloud, overcast low-level cloud and broken multilayer cloud. These varied arrangements are useful in studying the 1.38 um response and to test its reliability in identifying thin cirrus. 1.38 µm reflectance thresholds are determined from high probability clear pixels. It is found that observed clear sky 1.38 µm reflectance is not isotropic and is highly dependent on the satellite scan angle. Mean clear sky 1.38 µm reflectance also varies from scene to scene possibly due to varying humidity and aerosol concentration. Initial results indicate that both the second and third cloud schemes find a much larger amount of high cloud than the scheme based on the MODIS Cloud Mask product primarily because it uses a fairly conservative 1.38 µm reflectance threshold value. On the other hand, both schemes also appear to be prone to falsely identify mid and low level clouds as high cloud when compared to radar and the MODIS Cloud Phase product (MOD06). Still, these results give promise that with further analysis of 1.38 µm reflectance, it is possible to develop a more accurate depiction of thin cirrus is possible.
Joint Poster Session 1, Cloud Remote Sensing (Joint between 11th Cloud Physics and 11th Atmospheric Radiation)
Monday, 3 June 2002, 1:00 PM-3:00 PM
Previous paper Next paper