Given the significant impacts to a number of aviation hubs around the U.S. yesterday, there was a great deal of collaboration and discussion, especially within the GOES-R group. It was a great day to really get into the demonstration products and there were a lot of very good comments and feedback. While these will be covered more specifically within the final report of the experiment, a brief idea of the feedback we received is highlighted here.
HRRR Lightning Threat: In the morning the HRRR composite lightning threat (threat 3) was showing a large number of very marginal threat values over the Northeast with the ongoing stratiform precipitation, and it was noted by several forecasters that the product seemed to be "overdoing it" a bit. Such a large area of scattered lightning threats would indicate a broad area of convection, when in reality it wasn't plausible given the lack of instability. There was some discussion of whether this was due to the way in which the composite threat is made; i.e., the forecaster knew there were no convective cores associated with the ongoing stratiform precipitation, which is where threat 1 comes from, and postulated that these marginal threats were all due to threat 2.
However, we also used the HRRR threat 3 for the anticipated CI in the lower Great Lakes, and it was showing several more significant areas of higher lightning threats in eastern OH for early in the afternoon. While the convection did develop into a line, the higher lightning threats did correspond relatively well to the most intense radar echoes.
UW-CIMSS Cloud-Top Cooling (CTC): One forecaster immediately mentioned how much utility this product would have at the Convective SIGMET (CSIG) desk, particularly in the western Dakotas. In this area the radar coverage is much more sparse and as such, the forecaster mentioned relying heavily on only lightning data. It was mentioned that the CTC algorithm would greatly aid in convection forecasts for radar sparse areas such as this.
There was also a long discussion on the algorithm itself, on how the detections are characterized. Yesterday in particular there were times in which several detections within a cumulus field were set off and cells did develop. However there were other cells that escaped detection, but still developed as well. The question from the forecasters was, "How do the characteristics of the detected cells differ from those that were not detected?" The thought was that it may depend heavily on the environment in which the development is occurring.
Overall, the forecasters noted the the algorithm did extremely well for isolated single cell and supercell situations. However, with embedded and multi-cell development they noticed how the algorithm missed detections in areas with thicker cirrus clouds (after convective initiation began). Yesterday this issue was particularly noticeable as our areas of interest contain ongoing convection and expanding cirrus shields.
SATCAST: Yesterday afternoon the forecasters noted a group of higher strength of signal (SOS) detections in northern TX, which did result in the development of a number convective cells within the cumulus field of interest. It was noted that in the visible and SATCAST overlay used, there were many SOS detections in this field, which may be useful in picking out particular clouds of interest; however, the large number of detections made it difficult to see the cloud features underneath.
No comments:
Post a Comment