As a part of an NC State University and NWS CSTAR project, Bryce Tyner from NC State has developed a mechanism to adjust the currently implemented modified Rankine vortex interpolation method used in the TCMWindTool. The TCMWindTool is used by forecasters at Weather Forecast Offices to create forecast grids of winds in a 2.5km gridded forecast database from the guidance contained in the TCM forecast product provided by the National Hurricane Center. The new method subtracts off an empirically based error function from the modified Rankine vortex interpolated wind field. Additional information on how the error function was developed is available in a previous blog post – Potential New Interpolation Method in TCMWindTool.
To evaluate the new modified Rankine with error function vortex interpolated wind field, comparison images of the H*wind analysis (H*Wind), the currently used Modified Rankine (MR) and the new Modified Rankine with Error Function (MREF) were created for selected tropical cyclones. These images are available online at the following URL: https://sites.google.com/a/ncsu.edu/modified-_rankine_improvements/
Collaborators were asked to provide comments and feedback about the potential improvement by the new MREF over the current MR. Here are a few of the specific comments:
• In the right front quadrant where the strongest and most damaging winds occur, the MREF is better in two respects; it results in a quicker decrease in winds just outside of the eye which is more representative in many cases and it spreads out of the moderate strength winds further to the northeast which is also more representative.
• In the left quadrants, it is more conservative with the wind magnitudes, which we find is often the case with the storms that move northward near the Gulf Stream.
• Finally, every storm is different, but usually the winds are simply too strong in the current MR method and the fact that the MREF decreases the area of the maximum winds is a positive.
In addition, I wanted to share some specific example images and some interpretation of the imagery.
Igor (2010) provides an example of a case where the MREF is a significant improvement over the MR by reducing the wind magnitude across all quadrants, most notably in the southwest quadrant. In this case, the MREF performed much better than the MR.
Ike (2008) is a very good example of the helpful adjustment the error function provides in reducing the area of the strongest winds around the center. Note the reduction in the area of winds greater than 75kts shown in the red between the MR and the MREF.
Rita (2005) is an example of the helpful adjustment the error function provides in reducing the large area of the strongest winds around the circulation center. In this case, the MREF performed better than the MR.
Dennis (2005) is another good example of the helpful adjustment the error function makes to wind magnitudes across all quadrants. Note how the MREF magnitudes are much less than the MR across all of the quadrants. However, the error function is likely responsible for the low bias in the northwest quadrant where the H*Wind analysis notes the maximum winds and in the northeast quadrant where an excessive region of moderate winds is indicated. In addition, because the error function is not applied in the region of the radius of maximum winds, an artificial wind maximum ring is shown around the storm center in the MREF.
In a climatologically unusual wind field such as with Earl (2010) which had the strongest winds in the southeast quadrant, the MREF still applies the climatology based bias using a more typical wind field. This in effect dilutes the unusual wind field contained in the TCM (which had the strongest winds in the southeast) and results in more uniform wind field. In this case, the MREF performed worse than the MR.
In Irene (2011), the error function in the MREF resulted in a low bias in the wind field, especially in the eastern quadrants while the MR had a notable high bias. Neither the MREF nor MR performed well but the MR might be preferred.
Felix (2007) was a compact storm with an unusual wind field. The MREF had another ring of uniform strong winds around the center and a low bias outside the radius of maximum winds. The MR had a better wind field that matched the H*Wind analysis much closer.