GEMPAK User Stories

User Story 1:

1. Precipitation processes

• The topic is introduced by asking the students to think about why some rather impressive looking clouds don’t produce precipitation, even if they are low clouds, while at other times precipitation falls from rather thin looking clouds.
  
• Then in lab, the students use GARP to identify areas of stratiform precipitation on the radar, and the use the satellite with the mouse-over temperature readout to measure the pixel cloud-top temperatures (with acknowledgement of uncertainty and error in these estimates). They define the range of cloud-top temperatures at which stratiform precipitation is observed at the surface.
  
• They also identify using VIS and IR imagery locations with cloud, but no precipitation.
  
• Then in class, we discuss their labs. I talk about the Bergeron process and collision-coalescence [for stratiform precipitation- understanding that it doesn’t work the same in convective clouds]. We discuss the sensitivity of precipitation occurrence to the presence of ice in stratiform clouds. To date, the students haven’t found examples of stratiform precipitation with cloud tops above freezing, or even above -10 to -15C or so.

2. Sounding Manipulation

• The students generally already know that thunderstorms are most likely in the afternoon, and that stability must decrease then. But to quantify this, we have a lab where the students use NSHARP to load up a 12Z sounding either from a historical case or a current one. They, they use the mouse to modify the sound with the model or NWS-predicted high for the day to see – do we reach the CCL? How much does CAPE increase through the generation of a well-mixed PBL? Then, I have them modify the dew point, for example to account for the evaporation of dew and evapotranspiration, etc. They realize that the CAPE can be more sensitive to dew point than to temperature with this exercise.
• Students are able to see in a direct and quantitative way how the boundary layer and diurnal trends can link to the occurrence of convective storms.
• Improvements- a zoom capability would allow the students to more easily modify the soundings. As it is, sometimes points get “left behind” by the mouse, and student values vary widely.

There are many more lessons like this that I’ve developed over the years, and these are very condensed accounts of them, but it’s a start.

User Story 2:  Senior level synoptic and mesoscale forecasting laboratories ( 2 examples at Iowa State):

We archive the .gem files for upper air, surface, and some models for good cases demonstrating some phenomenon, and thus easy ability to archive the data students will need is essential.  As an example of a lab exercise using real-time data to understand tropical storms and hurricanes, I have students use satellite data with lat/lon overlaid and pick out regions of substantial cloud cover over the tropical Atlantic and Gulf, and tell me if they appear to be organizing in some way.  The students then are told to predict using GFS model data if the system is likely to develop in the next 48 hours, and where it would go.  They are also asked to do the same for the 48-96 h period, and then to explain how much threat they feel a system would be to the U.S.   To do this lab, students use garp to display the satellite images along with enhancements, and they use gdplot to view model predictions of 500 mb flow, 300-850 mb shear, and 700 mb relative humidity.

In another lab using archived data, I have them view images of momentum surfaces (m) and theta-E to diagnose regions where CSI exists.  They also do this using moist potential vorticity overlaid with relative humidity and theta-E contours.  I have used gempak cross-section programs to do this, but in recent years have noticed garp can do this (with a trick to create moist potential vorticity instead of potential vorticity).

User Story 3: HOOT project

The largest University of Oklahoma group who uses GEMPAK is HOOT.  HOOT is a student run forecast product development team, consisting of both undergraduate and graduate students.  HOOT uses C-Shell scripts to process 1000's of images for their website (http://hoot.ou.edu). The primary programs used by the HOOT project are gdplot2, gdplot3, gpmap, snprof, snmap and sfmap.  As HOOT relies on timely processing of data for the generation of a large number of products, initialization of the analysis program is critical. A command line interface for scripting is preferred, as their products are produced on a remote server and they do not wish to run a persistent x-session.

User Story 4:

These are points that were made during an interview of a high end Gempak user who was asked to identify existing capabilities that are critical to be included in AWIPS II.

1. High Degree of Control

• Setting contour intervals
• Shading
• Line width/type (dotted/dashed etc.)

2. GUI should be able to access archived data

3. Ability to interpolate to p/v surfaces

 4. Ability to add/modify code.  Subroutines should be transparent so users can modify or use in other programs independent of NAWIPS.

 5. Ability to read/write to/from Fortran

 6. Software should be able to run on standard hardware NOT high end hardware.  Software should be just as fast & efficient as current Gempak