Divergance & Curl Maps from Level 2 Data

[Robert Edmonds]

Hello all,

Since I'm still working out the oddities in the upper air data at my website, I decided to put that on the back burner and move forward with another idea...

Recently, I added divergence and curl maps derived from level 2 data for the lowest elevation scan. I have added an about section for these maps on my website which should explain a little bit about these maps, although I am sure I will expand on the section. I partly came up with this idea after reading Vasquez's book, Weather Forecasting Handbook. The book has a nice section on interpreting radar which might also help to explain some ideas behind these maps.

So how does this help, you might ask... The curl map gives a measure of rotation, and the div, or divergence, map gives a measure of the divergence and/or convergence. So, a cyclonic rotating storm will show up red because this produces positive curl on the maps. The amount of convergence can indicate whether a storm is strengthening or weakening. So, a storm with convergence, or negative divergence, will show up as blue on the divergence maps.

The maps are currently updated every ~10mins. My computer currently processes these maps for ~8 sites. The radar sites which will have the maps produced for the day will usually change day to day. You can even email me and ask for a site to have the maps produced for a day. If for nothing else go take a look at those level 2 reflectivity images.

The maps can be found at www.wxvelocity.com/radar.html,
and the link which explains a little bit about this can be found at http://wxvelocity.com/about_div_curl.html

Again I would love to hear about any comments or suggestions you have for the site (I'm still am not sold on the range I should use for the color bar). Also if you find a broken link please email me... Since it is new there will probably be some kinks to work out.

From KFTG today...
Reflectivity Map

Divergence Map

 

[L.B. LaForce]

Those are pretty neat maps you have made there. I assume your using the level 2 data feed from iastate, and then deriving these products from them? The reflectivity maps remind me of the DOW's radar, or even like the image at the top of this page: http://www.gel.cmich.edu/met/

Great job, and keep it up!

EDIT: KLVX would be pretty neat today, as that MCS/Derecho heads SSE-ward (not that its gonna have much, if any rotation, but it would be neat to see the products from this site).

 

[Robert Edmonds]

EDIT: KLVX would be pretty neat today, as that MCS/Derecho heads SSE-ward (not that its gonna have much, if any rotation, but it would be neat to see the products from this site).

It has been added...
http://www.wxvelocity.com/kind.html

edit: doh, notice I added kind instead of klvx... oh well it's all automated now actually

 

[Jeff Duda]

Impressive, indeed, but how'd you use just radial velocity to create those maps? You need two components of velocity in each case and the velocity product on radar only gives you one at any given point. I'm curious.

 

[Robert Edmonds]

Impressive, indeed, but how'd you use just radial velocity to create those maps? You need two components of velocity in each case and the velocity product on radar only gives you one at any given point. I'm curious.

That's correct, I explain in the about section that because they are derived from the radial velocity maps, they do not give a complete picture of the true velocity and divergence... However, divergence and curl is sometimes detectable if the winds are oriented properly with the radar.

http://www.wxvelocity.com/about_div_curl.html

If you notice sometimes curl and divergence is detected, in the simulated images. However, fake divergence and curl usually shows up closer to the radar.

 

[Jeff Duda]

Alright...do you know why the artificial vorticity and divergence are being pegged by the radar? Is it due to the extreme changes of radial angle at close ranges near the site?

 

[Robert Edmonds]

Alright...do you know why the artificial vorticity and divergence are being pegged by the radar? Is it due to the extreme changes of radial angle at close ranges near the site?

That's correct too. The fake curl and divergence usually shows up closer to the radar because the radial vector (i.e. the arrow pointing away from the radar) will spatially change more rapidly (i.e. same idea that it appears buildings closer to you move more than those out on the horizon when driving).

 

[Kiel Ortega]

Robert, are you familiar with the LLSD technique which NSSL has been using for a few years now? http://ams.confex.com/ams/32BC31R5C/techprogram/paper_64420.htm

I am guessing your maps are trying to achieve something similar (?).

 

[Robert Edmonds]

Robert, are you familiar with the LLSD technique which NSSL has been using for a few years now? http://ams.confex.com/ams/32BC31R5C/techprogram/paper_64420.htm

I am guessing your maps are trying to achieve something similar (?).

Thanks for bringing this up. I generally developed this without any knowledge of how others go about doing this before. This is because I adapted allot of the code from some other side projects I had going. So I had allot of the pieces but never put them together. The last few weeks I've been more focused on the coding. I have one last little piece that I am making, then I definitely will have to read up and see how what I've developed compares with other schemes. So, thanks for bringing up that link. From a quick read of the abstract it does appear that what they and I are trying to do is similar. In particular I am interested in some of the filtering schemes, I might in the future modify the one I currently employ.

Edit: I am implementing an automated radar site selection algorithm, so any disruption is likely the result of its implimentation... seems to be working

 

[Robert Edmonds]

I expanded the about section which includes some images of what is produced from the program with various wind fields.

http://wxvelocity.com/about_div_curl.html

 

[beaudodson]

Thanks for your hard work. Interesting topic and thread. Look forward to reading more about your study and the results. I think these type of studies can be beneficial to everyone that is looking for more than just the "normal" radar data.

Good work!

 

[Robert Edmonds]

Thought I might share.
I went back and grabbed the May 3, 1999 data. This is the resulting images for one of the storms (you're probably seen the reflectivity image before). The top left is the reflectivity, and the top right is the radial velocity. The bottom left is the curl; notice the red where the cyclonic rotation is located. The bottom right is the divergence; notice strong convergence where the strong rotation is located.

Also, as a result of going back over past data I have changed the color range slightly for the images on my website.

 

[Robert Edmonds]

After looking over archived data, there are sometimes its clear there's rotation/convergence and sometimes it's not so clear using the maps. However, playing with data of the Moore, OK storm I was able to draw out the path fairly well of the tornado using the convergence and rotation maps. Multiplying the rotation and convergence maps together, I was able to use the information to "draw" out this path. Doing a rough overlay with the damage path I was fairly happy with how well they matched up...

There are some issues, but developing all this over the summer as a side project, I'm happy with what I was able to get done starting from scratch. I at least believe I developed the stuff to the point I can keep it running on the side even with classes starting up. However, this I'm sure will remain a work in progress.

Edit: Thanks to Keil for bringing up the link to past research, I have now also implemented what will hopefully be a better filter to the div & curl maps and the upper air maps.

 

[Greg Higgins]

"farrelly" is this something new or did you mean "fairly"?

 

[Robert Edmonds]

"farrelly" is this something new or did you mean "fairly"?

I guess I better be banned for not spelling correctly ... Sorry it was late when I posted.