Cloud Thermodynamic Phase Retrievals Using MODIS Multispectral Data

W. Paul Menzel: PI

Bryan A. Baum      Richard A. Frey      Kathy I. Strabala

This page provides information on the inference of cloud thermodynamic phase in satellite imagery. Included on these pages are selected examples of satellite image analyses, information regarding narrowband radiative transfer modeling, efforts to improve the cloud phase retrievals, and validation of the algorithms.


Cloud phase is inferred from measurements at 8.5 and 11 microns (see references below). Radiative transfer simulations indicate that the brightness temperature difference between the 8.5- and 11-micron bands (hereafter denoted as BTD[8.5-11]) tends to be positive in sign for ice clouds that have an infrared optical thickness greater than approximately 0.5. Water clouds of relatively high optical thickness tend to exhibit highly negative BTD[8.5-11] values of less than -2K. The BTD[8.5-11] values are quite sensitive to atmospheric absorption, especially by water vapor, and also to the surface emittance properties. Clear-sky BTD[8.5-11] values tend to be negative because the surface emittance at 8.5 microns tends to be much lower than at 11 microns, especially over non-vegetated surfaces. The BTD value for low-level water clouds tends to become more negative as the water vapor loading increases. While a relatively small effect, multiple scattering is included in radiative transfer simulations of the BTD[8.5-11]. Small particles tend to increase the BTD[8.5-11] values relative to large particles because of increased scattering.

Advantages: As this is an IR-based method, it may be applied to both daytime/nighttime data. Also, the method is easily implemented and maintained in an operational environment.

Disadvantages: There are three primary issues that cause difficulties for the IR phase method.

  • optically thin cirrus is very difficult to detect, much less discern the phase unambiguously.
  • the ability to discriminate between ice and water particles at cloud top is much reduced when clouds may contain primarily supercooled water droplets or perhaps a mixture of both ice and water (i.e., mixed phase). Single-layered clouds of wide spatial extent having cloud-top temperatures in the range between 240K and 270K are prevalent in the storm tracks in both the Northern and Southern Hemispheres.
  • retrievals are problematic when multilayered, overlapping cloud layers are present and the uppermost cloud layer is optically thin. For example, we have noticed that when optically thin cirrus overlies a lower-level water cloud, the MODIS IR cloud phase algorithm tends to classify the pixel as being either "mixed phase" or "uncertain phase".

Efforts are ongoing to ameliorate these known problems.

In operational practice, each pixel identified as being cloudy is classified as being in one of 4 categories:

  • ice phase
  • water phase
  • mixed phase
  • uncertain

To download the MODIS cloud phase subroutine (in fortran): IRphase.f


Baum, B. A., P. F. Soulen, K. I. Strabala, M. D. King, S. A. Ackerman, W. P. Menzel, and P. Yang, 2000: Remote sensing of cloud properties using MODIS Airborne Simulator imagery during SUCCESS. II. Cloud thermodynamic phase. J. Geophys. Res., 105, 11,781-11,792. View

Platnick, S., M. D. King, S. A. Ackerman, W. P. Menzel, B. A. Baum, J. C. Riedi, and R. A. Frey: The MODIS cloud products: Algorithms and examples from Terra. IEEE Transactions on geoscience and remote sensing, Special Aqua Issue. View

Strabala, K. I., S. A. Ackerman and W. P. Menzel, 1994: Cloud properties inferred from 8-12 micron data. J. Appl. Meteorol. 2, 212-229.

Question: How much of the time is the cloud thermodynamic phase retrieval inferring either ice or water with an IR-only approach?

Here is a daytime image prepared from MODIS Terra data collected on November 5, 2000:

Here is a nighttime global image from the same day:

Further insight as to these results: From detailed inspection of many granules on this day where the cloud phase was indeterminate, we have come to these conclusions.

  • In the higher latitudes, cloud top temperatures lie between 240K and 273K, and the clouds are optically thick. It is extremely difficult to infer cloud phase unambiguously for these clouds. We are investigating the use of visible and near-infrared bands to supplement the retrieval during daytime hours.
  • In the lower latitudes, the primary issue seems to be that of multilayered clouds, i.e., optically thin cirrus overlying lower-level clouds. We are working on a detection algorithm for the presence of these overlapping cloud layers.

Detailed example from Platnick et al.: The MODIS cloud products: Algorithms and examples from Terra. In press, IEEE Transactions on geoscience and remote sensing, Special Aqua Issue. View

Note that in the lower portion of the image, off the coast of Chile and over the ocean, there is a patch of "uncertain". Upon further inspection (see image below), we think the uncertainty is caused by thin cirrus overlying a low-level water cloud. In the false color image (below, right) showing a detailed portion of the scene, low clouds appear as blue and the cirrus as pink.

For further information on our approach for identifying where thin cirrus overlies lower-level clouds, see Multilayered Clouds.

Page created on January 28, 2003.

Back to Bryan A. Baum's Home Page