RadarScope vs GR LevelX - resolution etc

[Skip Talbot]

Smoothing is an easy feature to incorporate into your app if you set it up for that purpose. If you've played any sort of console or computer game you've probably seen this countless times already. Walk up to a wall or look at the ground in your 3D shooter, and you'll see the texture gets a little blurry. From afar it looks great, but it's been smoothed if you get close to it. It's called texture filtering. When the image's pixels become larger than the screen's pixels, the graphics hardware in your xbox or PC interpolates between the pixels to produce a color in between the these two image pixels. Your phone has the same graphics hardware in it, so in theory it could be easy as enabling texture filtering if the radar is being presented as a texture that's drawn with the graphics hardware.

 

[ChristofferB]

Smoothing is an easy feature to incorporate into your app if you set it up for that purpose. If you've played any sort of console or computer game you've probably seen this countless times already. Walk up to a wall or look at the ground in your 3D shooter, and you'll see the texture gets a little blurry. From afar it looks great, but it's been smoothed if you get close to it. It's called texture filtering. When the image's pixels become larger than the screen's pixels, the graphics hardware in your xbox or PC interpolates between the pixels to produce a color in between the these two image pixels. Your phone has the same graphics hardware in it, so in theory it could be easy as enabling texture filtering if the radar is being presented as a texture that's drawn with the graphics hardware.

Yeah, I've noticed that and have been using that to some extent but it's not really the same thing.

Regarding, GRLevel. I think I may end up buying that if I go chasing as a part of a chase team at some point. At the moment I have to go with organized tours because I am nowhere near the knowledge to be able to do it myself. I use RadarScope solely to "shadow chase" from home and I will be using it to check out the storms myself on my chase this year. So, road networks etc isn't all that useful for me at the moment although it would be cool to see in order to think how I would have driven.

Does the lack of hail size markers also mean that I won't be seeing reported tornadoes, rotating wall clouds etc in RadarScope?

 

[Brandon Jenkins]

Yeah, I've noticed that and have been using that to some extent but it's not really the same thing.

Regarding, GRLevel. I think I may end up buying that if I go chasing as a part of a chase team at some point. At the moment I have to go with organized tours because I am nowhere near the knowledge to be able to do it myself. I use RadarScope solely to "shadow chase" from home and I will be using it to check out the storms myself on my chase this year. So, road networks etc isn't all that useful for me at the moment although it would be cool to see in order to think how I would have driven.

Does the lack of hail size markers also mean that I won't be seeing reported tornadoes, rotating wall clouds etc in RadarScope?

You can see storm reports in radarscope with an allisonhouse subscription, just not the meso/hail markers.

 

[Rob H]

Hail markers are usually POH/POSH from the SCIT information which can be wildly inaccurate. I've core punched 2.5" hail marker storms and haven't even seen pea-sized. If you want to avoid hail your best bet is to learn how to forecast the three types of hail: inconsequential, could bust windshield, will bust windshield. Then use the dual-pol HCA product to find the hail core. Or stay back a little further

SCIT has your hail markers, meso markers, storm tracks, etc. You can also display Spotter Network reports of hail, or LSRs (Local Storm Reports) for hail - both of which would be actual observed sizes. Reports of wall clouds are usually Spotter Network reports, and funnels/tornadoes can be SN again, or LSRs. So it's kind of a mish-mash of where the data comes from depending on what exactly you want to see.

Long story short, I don't know what RadarScope displays. You'll want to check and see if it supports SCIT, SN reports, and LSRs I suppose.

 

[Mike Gibson]

Smoothing is an easy feature to incorporate into your app if you set it up for that purpose. If you've played any sort of console or computer game you've probably seen this countless times already. Walk up to a wall or look at the ground in your 3D shooter, and you'll see the texture gets a little blurry. From afar it looks great, but it's been smoothed if you get close to it. It's called texture filtering. When the image's pixels become larger than the screen's pixels, the graphics hardware in your xbox or PC interpolates between the pixels to produce a color in between the these two image pixels. Your phone has the same graphics hardware in it, so in theory it could be easy as enabling texture filtering if the radar is being presented as a texture that's drawn with the graphics hardware.

What you're referring to is color interpolation, where RGB values are interpolated from one gridpoint to another. That sort of interpolation works well for photos and other natural imagery but it most definitely does not work well for data. The GR programs interpolate between data values then look up the interpolated result in the color table.

You can turn smoothing on or off per-product in the GR programs, depending on your preference. I turn it on for reflectivity and reflectivity-derived products because I believe it gives a more realistic reconstruction of what's really out there. Have you ever seen a storm that is composed of 1° by 1 km solid blocks of hydrometeors? I haven't. If one radar gate is 60 dbz and the next one is 30 dbz, it's reasonable to assume that halfway between them the value is somewhere around 45 dbz. You'll note that when the Level III data transitioned from legacy 1km gates to 250m gates, the echoes did not retain a blocky 1km appearance (when not smoothed). The increased resolution showed that the previous 1km values indeed transitioned through intermediate 250m values -- just as smoothing of the 1km gates had shown previously.

Finally, there's the issue of which representation of reflectivity should be smoothed (Z or dbZ). The GR programs have always smoothed dbz values. See the following paper for information on why that's the correct method for interpolating reflectivity:

http://www.cimms.ou.edu/~lakshman/Papers/zordbz.pdf

This too was confirmed in the transition from 1km to 250 m gates in the L3 data stream. Interpolating in Z then converting back to dbZ for color classification results in a "bloated" visual output, which was not seen when the data transitioned to 250 m gates.

Mike Gibson
(author of the GR programs)

PS. While I emphasize the transition from 1km to 250m gates in the L3 data in this post, those results were already known by comparing L3 data to the higher resolution L2 data collected at the time.

 

[Zac Flamig]

That's correct. GRLevel2 and GRLevel3 have nice smoothing algorithms that can actually be worth using on reflectivity, for reasons Skip described well. No other radar software I'm aware of has it. PYKL3 has terrible smoothing that just seems to blur the data (the author is very anti-smoothing), while RadarScope has none at all.

WeatherWall Mobile on iOS and WeatherWall on OSX both have smoothing btw.

Also, as Skip mentioned, if you're near a tornado/storm don't be waiting for radar updates to tell you what the storm is doing, use your eyes. On May 31st the tornado went from moving away from me to hitting me in ~60s, much faster than you're ever going to get a radar update. Of course, even if you can see the tornado changing directions it doesn't matter much if you get caught in the wrong spot.

 

[Paul Knightley]

For my '2 cents' I love the RadarScope app - it's very simple, easy to use, and works a treat. I quite like the smoothing option on GRLevelX, but it doesn't bother me not to have it. Beside, the WDT data on RS gives you 'super-resolution' which is more than enough.

Look out the window, though - that's where the best clues are!