I. Introduction 

    Following the work of Rosenfeld, Lerner, and Woodley (2004), this is a preliminary study of storms which developed on 20 April, 2004 and 22 May, 2004 using GOES-12 data. The Rosenfeld et al. (2004) study used AVHRR data, which has superior spatial resolution (~1 km) but lacks the temporal sampling to study cloud evolution. The purpose here is to:

1) Determine if the spatial resolution of the GOES data is sufficient to resolve the vertical profiles of mid-IR (3.9 micron) solar reflectance from cloud clusters near storms,

2) Explore the time evolution of cloud-top reflectance for individual storms,

3) Further investigate differences in the vertical profiles of solar reflectance and their evolution between storms of varying intensity.

The 3.9 solar reflectance has been estimated using the 3.9 and 11 micron observed brightness temperatures, the solar elevation and solar constant. (The contribution of emitted long wave radiance is subtracted from the observed radiance at 3.9 micron).  It should be noted that the reflectance can be used to estimate an effective radius of cloud-top hydrometers (as in Rosenfeld and Gutman, 1994).  For the initial stage of this study, results are presented in terms of reflectance rather the effective radiance.

For comparison with the plots of Rosenfeld et al., the reflectance can be converted to effective droplet radius using the following table.

Table I

Relationship between cloud-top reflectance of water clouds and droplet radius (for moderate sun-target-sensor angles, ~45 degs). After Rosenfeld and Gutman, (1994).

3. Imagery

    The following images can be viewed in Table 2:

1) Reflectance at 3.9 micron (band 2) ,

2) Brightness temperature at 11 micron (band 4),

3) Visible imagery (band 1).  

The links in Table 2 use the Interactive Composite Editor (ICE) developed by Tom Whittaker at the University of Wisconsin for the NASA Earth Observatory.  In addition to displays of each image, the Java applets allow the computation and display of interactive scatter plots for selected sub-regions of each image. An overview of the ICE is given here.

After clicking on the desired image time in Table 2, a page will be displayed with three small images along the top row.  To create a scatter plot:

1) Click on the visible image for display in the large window.

2) Click the "select region" box in the righthand margin.

3) Move cursor to desired starting point in the large window. Hold the left mouse button down to draw desired region. Release the button to end selection.

4) In the top row, click the 11 micron image, then click the "albedo" image.

5) Click the box marked "Scatter" in the righthand margin.  A new window will appear with a plot of the 11 micron brightness temperature in the vertical axis and albedo (reflectance) in the horizontal axis.

Scatter plots were computed in areas which included at least a portion of the anvil (once developed) and clouds on the inflow side of storm.  The areas chosen were roughly 40,000 square kms (a box 200 km x 200 km).  These are available for viewing in the last three rows in Table 2.
 

4. Preliminary Results

    A comparison of the profiles of reflectance vs. cloud top temperature suggests that the storms have significantly different properties.  The Illinois clouds exhibit the greatest range in reflectance (Fig. 2), .03-.20 microns.  The coldest cloud tops have values of .06-.08.  The Iowa storm clouds (Fig. 3) have nearly as large a range of reflectance as the Illinois storms, however, the coldest tops have smaller values.  The Oklahoma storms have a smaller range of reflectance (Fig. 4).  The reflectance decreases more slowly with 11 micron temperature than in the other two cases. Specifically, in the upper portion of the plots of the Oklahoma storms, there is a linear decrease of reflectance with temperature from 0.2 near -5 C to 0.06 near -50 C.  Rosenfeld et al. (2004) found that a similar slope characterized the most intense or severe storms (see their Fig. 8). The non-severe storms exhibited a more rapid decrease in reflectance with temperature in the upper layers. The profiles of the Illinois storm appear more characteristic of the non-severe storms in Rosenfeld et al., 2004.  An analysis with AVHRR data  should be conducted to determine the validity of these profiles obtained from the GOES data.

    It is useful to compare the cloud top reflectance of the individual storms during their development.  The time series of reflectance are shown in Table 3.

    Once formed, the Illinois storm-top has greater reflectivity than the other storms.  This suggests consistently smaller cloud-top hydrometeors near the top of the Illinois storm (for example, 15 versus 20 microns at 2310 UTC).  If smaller hydrometeors are indictive of more intense updrafts on a local scale, then observations of highly reflective cloud tops may be a useful quantity to monitor.  It would be relatively simple to implement an image enhancement scheme to isolate this cloud property (during daylight hours).  The routine display of storm-tops with this enhancement would enable a more extensive study of cloud top properties versus storm severity.     

References:

Rosenfeld, D., G. Gutman, 1994: Retrieving microphysical properties near the tops of potential rain clouds by multispectral analysis of AVHRR data. Atmospheric Research, 34, 259-283.  
 
Rosenfeld, D., A. Lerner, W.L. Woodley, 2004?: Satellite-retrieved microstructure of clouds as an indicator for the development of severe convective storms. To be published?