Towards the Objective Analysis of Clouds from Satellite Imagery Data

View More View Less
  • 1 National Center for Atmospheric Research, Boulder, CO 80307
© Get Permissions
Full access

Abstract

We present an objective analysis scheme for deriving cloud properties from satellite imagery data for oceanic regions. The scheme is based on the spatial coherence method. As this method is applicable only to simple layered systems, we introduce a default estimate of the cloud cover when the systems become complex as in fronts and tropical disturbances. The steps of the scheme are as follows: 1) identify cloud-free regions and cloud layers within (250 km)2 frames; 2) for each (60 km)2 subframe evaluate the statistics of the radiance field needed to retrieve cloud cover; 3) cumulate the subframe statistics in a given geographical region for several days and construct from the cumulated cloud-free radiances a climatology for that region and time period; 4) derive for each (60 km)2 subframe instantaneous estimates of the cloud-free radiances and cloud properties at the time of satellite passing; 5) composite these (60 km)2 subframe results to form the desired space and time averages. We apply the analysis scheme to derive the cloud cover from NOAA-7 AVHRR GAC data for the orbits of three days and nights over the Pacific basin (0–50°N, 135°W–170°E). We find: 1) the statistics of the radiance field used to obtain the cloud cover represent a 15-fold reduction over the input data volume; 2) clouds will satisfy the conditions for spatial coherence retrievals typically for 30–50% of the (250 km)2 frames and for 50% of the (60 km)2 subframes; 3) the majority of (250 km)2 frames contain more than one identifiable layer of clouds; 4) less than 3% of the (60 km)2 subframes exhibit three identifiable layers suggesting that methods for treating one and two-layered systems on the mesoscale should prove adequate for the majority of maritime cloud cases; 5) the typical uncertainty of an instantaneous cloud cover estimate for a (250 km)2 frame is ΔAc ∼ 0.14. Owing to cancellation of random errors, we expect the uncertainty in the corresponding monthly mean cloud cover to be considerably smaller. In preparing satellite data for analysis, one first reads and converts the bit stream into calibrated radiances. Once the data are in the form of calibrated radiances, the additional computer time required to analyze cloud properties is approximately equal to the computer time needed to read and convert the satellite bit stream.

Abstract

We present an objective analysis scheme for deriving cloud properties from satellite imagery data for oceanic regions. The scheme is based on the spatial coherence method. As this method is applicable only to simple layered systems, we introduce a default estimate of the cloud cover when the systems become complex as in fronts and tropical disturbances. The steps of the scheme are as follows: 1) identify cloud-free regions and cloud layers within (250 km)2 frames; 2) for each (60 km)2 subframe evaluate the statistics of the radiance field needed to retrieve cloud cover; 3) cumulate the subframe statistics in a given geographical region for several days and construct from the cumulated cloud-free radiances a climatology for that region and time period; 4) derive for each (60 km)2 subframe instantaneous estimates of the cloud-free radiances and cloud properties at the time of satellite passing; 5) composite these (60 km)2 subframe results to form the desired space and time averages. We apply the analysis scheme to derive the cloud cover from NOAA-7 AVHRR GAC data for the orbits of three days and nights over the Pacific basin (0–50°N, 135°W–170°E). We find: 1) the statistics of the radiance field used to obtain the cloud cover represent a 15-fold reduction over the input data volume; 2) clouds will satisfy the conditions for spatial coherence retrievals typically for 30–50% of the (250 km)2 frames and for 50% of the (60 km)2 subframes; 3) the majority of (250 km)2 frames contain more than one identifiable layer of clouds; 4) less than 3% of the (60 km)2 subframes exhibit three identifiable layers suggesting that methods for treating one and two-layered systems on the mesoscale should prove adequate for the majority of maritime cloud cases; 5) the typical uncertainty of an instantaneous cloud cover estimate for a (250 km)2 frame is ΔAc ∼ 0.14. Owing to cancellation of random errors, we expect the uncertainty in the corresponding monthly mean cloud cover to be considerably smaller. In preparing satellite data for analysis, one first reads and converts the bit stream into calibrated radiances. Once the data are in the form of calibrated radiances, the additional computer time required to analyze cloud properties is approximately equal to the computer time needed to read and convert the satellite bit stream.

Save