Frequently asked questions

Convection: plumes of rising air, or thermals, condense into cumulus clouds when there is enough moisture in the atmosphere.

Shallow convection produces circulations low in the atmosphere that form cumulus clouds (photo 1). It will not necessarily produce showers.

Deeper, more vigorous convection can produce showery weather, from small, short-lived afternoon rain showers (photos 2–4) to deep, towering cumulus that become a thunderstorm.

Photos © R. Hankers.

Not all clouds hold enough moisture to show up on radar; fair-weather cumulus clouds, for example, do not. If there is enough instability, clouds keep growing vertically, condensing enough moisture to form radar echoes. When a radar echo reaches 35 dBZ (light to moderate rain), convective initiation is achieved.

The radar image, from the Hytop, Alabama (KHTX) radar, shows mature convection that has developed into a line of storms, and new areas where convective initiation is occurring.

Radar image from the UCAR/RAP real-time weather data site.

• Geostationary satellites look down on the Earth from a fixed location.
• This gives imagery of the continental United States every 5–15 minutes.
• Cloud growth can take hours for typical summer afternoon clouds, or 15–30 minutes for strong thunderstorm development.
• Satellites observe the clouds as they form, 5–90 minutes before radar echoes appear, depending on the environment.

The diagram shows that satellite detection starts earlier in a cloud’s growth than radar detection, so a forecast that uses satellite data has a longer lead time.

• Visible and infrared data from GOES East and West are ingested by the algorithm.
• Immature convective clouds are identified and tracked as cloud objects, shown as colored dots in the image.
• Infrared temperatures from three channels are averaged for each cloud object.
• These averages, together with the change in size of the cloud objects and the cloud types, give nine satellite “interest” fields that the algorithm uses.

The diagram shows a cloud top cooling from 0 °C to −10 °C as the cloud grows between two times.

• Model data is used to describe the environment a cloud is growing in.
• Currently, 15 model fields are used to diagnose the convective environment.
• All of this information goes into a logistic regression model, which produces a probability of convective initiation.

The lower images are example SATCAST probability maps for the western and eastern United States.