Holland et al. (2010) Interpolation Method: Irene (2011)

As discussed in previous posts, the TCMWindTool used to develop wind speed and gust forecasts at the NWS offices currently implements the modified Rankine vortex for wind interpolation between the various wind radii.  After a literature review, a new Holland et al. (2010) model provides some promise to improved wind interpolation.  As mentioned in the paper, the new method is less sensitive to wind radii forecast errors.

TCM best track data was acquired from NHC for Irene (2011) at 0600 UTC 26 August.  Below are plots of the H*Wind surface analyzed wind speeds, modified Rankine vortex, and Holland et al. (2010) inteprolated wind speeds plotted out to the best track analyzed 34 knot maximum wind radii for the various quadrants.  The H*Wind analyzed wind speeds is also shown.

In quadrants one and two, there is very good agreement between the H*Wind analyzed wind speeds and the Holland et al. (2010) interpolated wind speeds.  This is especially apparent outside of the radius of maximum winds to the 34 knot maximum radius.  The modified Rankine vortex method leads to a positive bias when compared to the H*Wind analysis at this time.  Both models over predict the wind speeds up to the radius of maximum winds, likely due to the way this is specified (take as 1/2 the default eye radius of 25 nm + 8 nm).

There is much less agreement when looking at quadrants 3 and 4.  This is likely due to the much weaker wind speeds observed in these quadrants, relative to the maximum sustained winds estimated by NHC and the rapid decay of wind speeds within these quadrants in the H*Wind analyses.  The documentation in the Holland et al. (2010) model, does suggest reduced sensitivity to these best track wind radii errors.  A senstivity analysis is currently being to conducted to validate this claim and will be presented in a later post.

In summary, the Holland et al. (2010) model shows marked improvement over the modified Rankine vortex, at least when the various wind radii are correctly forecasted and verified in the analysis.  I will be looking at additional storms as well as different analysis times to continue with this verification.

(Order of images:  H*Wind Analysis, Quadrant 1 Plot, Quadrant 2 Plot, Quadrant 3 Plot, Quadrant 4 Plot)

This entry was posted in TC Inland and Marine Winds. Bookmark the permalink.

2 Responses to Holland et al. (2010) Interpolation Method: Irene (2011)

  1. Jonathan Blaes @ WFO RAH says:

    Nice post Bryce. A question and a couple of comments. I assume that quadrants 1 and 2 are the northeast and southeast quadrants, is this right? Quadrant 1 looks promising but 3 and 4 seem odd. Not sure how to interpret that data.

    I think it would be interesting and perhaps even more enlightening to see how the Holland and modified Rankine comparison plots appear with 0 or 12 hour wind radii from the NHC TCM product used as the input and not the best track data. This would compare what potential improvement there would be with altering the TCM wind tool.

  2. Bryce says:

    Thanks for the comments, Jonathan. I am still playing with the best way to visualize the improvement using the Holland et al (2010) model. I can easily compare to 0/12 hour wind radii from the NHC TCM product. Upon brief inspection, the wind radii are not much different compared to best track data. I felt using best track data might be more appropriate since we are comparing to H*Wind analyses (more reflective of best track information than forecast data).

    As far as the differences for quadrant 3 and 4 are concerned, this likely has to do with a few factors. In quadrants 1 and 2, looking at the H*Wind analysis at this time (shown in original post), the wind field appears much more symmetric. Recall I plotted AVERAGE wind speed for the H*Wind data in each quadrant at each radius. In quadrants 3 and 4, the wind field is much less symmetric. As a result, this causes the average to deviate significantly from the TCM maximum wind radii data for each quadrant. Based on your insight, Jonathan, perhaps taking the maximum wind at each radius would be better, given the TCM product reflects maximum 30, 50, and 64 knot wind radii in each quadrant. I will follow up when new plots are created.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s