P4.33 Analysis of cloud retrievals from MODIS and their impact on radiative fluxes over Canada in the context of application for the CloudSat mission

Wednesday, 12 July 2006
Grand Terrace (Monona Terrace Community and Convention Center)
Alexander P. Trishchenko, Canada Centre for Remote Sensing, Ottawa, ON, Canada; and K. V. Khlopenkov and H. G. Leighton

CloudSat is designed to provide information about cloud vertical structure from active cloud profiling radar as part of the Aqua-train constellation. For cloud properties in optical and thermal bands the CloudSat processing system relies substantially on the use of data and cloud retrievals from MODIS/Aqua. This paper presents the results of our analysis of cloud retrievals available from MODIS MYD06 and MYD35 cloud data products for typical surface and climate conditions over Canada. We analyzed several cases covering different seasons for general observational geometry and for CloudSat sampling geometry corresponding to along track observations displaced by 215 km to the east of the MODIS image center. Particularly, we focused on winter scenes and cloud retrievals over snow covered areas. These conditions represent a major challenge, due to difficulties in cloud identification and retrievals over snow covered surfaces.

To evaluate the MODIS cloud product we employed the cloud detection algorithm SPARC (Separation of Pixels using Aggregated Rating over Canada) developed at the Canada Centre for Remote Sensing. SPARC is able to produce alternative cloud masks, cloud top height estimation, and other parameters. It was found that agreement between MODIS and SPARC cloud masks is on average around 74% and up to 86% during summer. The total percentage of the pixels that were mistakenly identified as clear-sky by one scheme and overcast cloudy by another scheme was less than 2%.

As MODIS cloud detection is designed to be clear-sky conservative, i.e. sensitive to any indication of cloud contamination (real or false), it may sometimes identify haze or inhomogeneous snow covered surface in the boreal zone as overcast cloud. This may turn cloud flags on and initiate cloud property retrievals assuming dark surface underneath. We analyzed a number of such situations to determine the overall impact on cloud statistics and radiative fluxes for CloudSat observational conditions and general geometry.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner