D-value Reveals Radiosonde Discrepancy/Inaccuracy
-
I'd like to start by saying it's distinctly possible that most other users know this and I'm just slow joining the party, but I thought I would share, in case it helps someone.
When viewing radiosonde measurements in Windy (fantastic product that I love), there have been several times that I've questioned the elevation and pressure pair, that is displayed when you mouse over (on the web) or touch and slide (on iOS) the radiosonde measurement view (highlighted by the magenta oval in the image below). The reason I've questioned it is because on many occasions my quick calculations for D-value (i.e., true altitude - pressure altitude) often didn't add up to what I expected at various locations and under various atmospheric conditions. I ran into this again last week, so I decided to dive into a little deeper.
What I found was that for fm35/netCDF format sonde data from NOAA MADIS, the D-value can be significantly different at the mandatory levels compared to all other levels. I spot-checked this at a few different sites around the world and they all appear to have the same situation.
Then I wondered if maybe the issue was isolated to the fm35 format from NOAA MADIS. So I started poking around to find radiosonde sites with other format and/or from other sources. Eventually I landed at Site ID 96471 (Kota Kinabalu, Malaysia) which had both fm94/BUFR from NOAA GTS and fm35/netCDF from NOAA MADIS (nice apples to apples comparison). When I calculated the D-values for both formats, sure enough, the fm94 looked smooth and correct at all levels, and matched the mandatory levels from the fm35, while all other fm35 levels had significant deltas. The non-mandatory levels in the fm35 appear to follow the same general shape, but with an offset (see graph below). I spot-checked a few other sites around the world that have fm94 data available from a couple different sources and they all appear to have what I would say is correct/accurate data at all levels (not just mandatory levels).
I’ve convinced myself that this is a source data issue, not a Windy issue, and I suspect that it likely has something to do with the different radiosonde formats (though I'm definitely no expert on radiosonde formats). So I guess the point of this post is a word of caution for anyone that may be using the Windy radiosonde measurements view (and likely any other method of viewing fm35 data, for that matter) for a quick, back of the napkin calculation of D-value. It would appear that if the radiosonde format for a given dataset is fm35/netCDF, only the mandatory levels can be trusted.
Given what I perceive to be an fm35 format deficiency or inaccuracy at non-mandatory levels, it would be helpful if the radiosonde measurement view in the iOS app would show what data format a user is viewing. If the user had the option to select a past measurement, similar to the dropdown on the full website, that would be cool too, but now I’m just being a bit of a beggar.
Thanks for reading my lengthy ramblings,
skywalker04aby -
I'm not sure what happened to the images in the original post, which is a little sad. if an admin can modify the post to include the below, that would be great, otherwise, here they are. Try #2.
