Global information of ice water path (IWP) in ice clouds is urgently needed for testing of global climate models (GCMs) and other applications, but satellite retrievals of IWP are still in the developing stages, and tend to have large uncertainties (e.g. factor of 3 or more). A new retrieval method is presented here which may have relatively low uncertainties, using one or more thermal channels.

Recent research indicates that ice cloud radiative properties depend on more than effective diameter (Deff) and ice water content (IWC). The size distribution shape, or dispersion, such as the degree of bimodality, is also an important factor at thermal wavelengths. Failure to account for this can lead to substantial retrieval errors. Hence, we have developed a retrieval scheme that accounts for the size distribution shape in mid-latitude and tropical cirrus.

Previously, ice particle shape appeared to represent a large source of uncertainty in IWP retrievals. Based on the methodology and new information on ice particle mass- and area-dimensional relationships, shape uncertainties now appear considerably less. By using a channel centered near 3.9 microns, absorption is largely volume dependent. This reduces particle shape uncertainties, which are associated with area dependent absorption.

For a given ice particle shape, preliminary tests varying particle size, IWC and cloud depth, as well as accounting for surface emission and instrument noise, suggest the scheme is accurate to about ±15% over an IWP range of 1 to 100 g m-2, and about ±25% over an IWP range of 100 to 200 g m-2, except for very thin cirrus having unusually large crystals or multiple cloud layers. This is based on thousands of radiation transfer model retrieval simulations.

By far the greatest contribution to our retrieval uncertainty is from ice particle size and shape. These uncertainties combined produce an estimated IWP uncertainty (standard deviation) of ±20%, based solely on the 3.9 micron channel and an IWP range of 1 to 100 g m-2. The 3.9 micron channel is well suited for nighttime retrievals, but not for daytime.

The retrieval scheme will be further tested with in situ data and ground based retrievals from two ARM cirrus IOPs, where GOES-8 channels at 6.7 and 10.8 microns were used for daytime conditions. These cirrus cases were sampled with the Counterflow Virtual Impactor (CVI), which gives a direct measurement of IWC and IWP.

Since the AVHRR instrument has a channel near 3.9 microns, and was carried on current and previous NOAA satellites, global trend analyses of IWP can be carried out over the last 20 years.

Finally, this scheme is equally well suited for water clouds for determining LWP, although uncertainties should be lower since absorption in water clouds at 3.9 microns is about half that for ice clouds, and there is no uncertainty due to particle shape.