In a recent article titled 'Detecting Putative Sporadic E Propagation in WSPRNet Spot Records ' for TAPR, Jeanette Zhou, KN6DAD outlined how she used data from the WSPRnet website to investigate Sporadic-E propagation.
She writes... "WSPRNet propagation data were downloaded from http://wsprnet.org/drupal/downloads and solar indices were from the German Research Centre for Geosciences (GFZ) (gfz-potsdam.de) in February, 2021.
WSPRNet spot records of 28, 50, and 144 MHz transmissions during 2020 were then used for the analysis, combining related spot records with the same timestamp and sender call sign to remove duplicates.
Putative sporadic E propagations were identified as when a transmission’s distance was about 1200 km in the above identified bands; the total number of records used for the analysis was 1,118,989. "
She then used the data to general the chart above which clearly shows the Summer peak of the Sporadic-E season although I suspect that many of the reports on 144 MHz in July, August & October were due to tropospheric propagation. The 144 MHz data may also be skewed from the fact that there aren't that many people using WSPR on 144 MHz in Europe or North America.
It was interesting that she found that a high Kp index resulted in poor Sporadic-E conditions.
"In conclusion, the putative sporadic E propagation in the WSPRNet spot records seems to be consistent with known seasonal variations of sporadic E propagation and corresponding geographic distribution, which suggests that these records provide a large number of observations that could be used to study sporadic E formation.
Exploratory data analysis also suggests that Kp index may be negatively correlated with recorded sporadic E propagation."
This corelates with the theory that quiet geomagnetic conditions are better for Sp-E formation.
In summary... I found this article interesting not only because it shows the Sporadic-E season but also it shows the perils of using WSPR spots for analysis.
For example, look at how poor the numbers are for April on 28 MHz and yet we know that there are always plenty of Sporadic-E openings at the end of April.
The reason there are so few WSPR spots for April is because there are so few people using it on the 10m band. Once the propagation starts getting better in May then a lot more people start using WSPR on the band.
In other words, a lot of the data is due to human activity rather than actual conditions. If a band looks dead then people are less likely to have WSPR transmitters active which means many openings generate no reports.
A million plus WSPR records sounds impressive but a lot of the data in there will be skewed by assumptions about the propagation mode, a lack of WSPR stations on the higher VHF bands and human behaviour.
WSPR reports can probably be used for propagation analysis but you would need to be very careful about what data you select to use.
The full article can be seen on Page 7 of the PDF document HERE
1 comment:
Your caution about these results is well-placed.
There are so many factors to take into account when using this kind of 'citizen science' database that it would take another paper to discuss them all.
Anyone who's used WSPR for propagation studies will realise, above all, that there is (a) a highly variable, herd-mentality dominated aspect to the number, continuity and quality of stations on any band, notably, as you rightly say, the higher bands. Even on 20m, it's often extremely difficult to find more than a handful of stations who are operating continuously - and even then, we usually know little about their RFI/take-off aspect, etc, etc. There have eve been cases of deliberate misreporting of output power.
Some, perhaps most of the pattern she has presented will suffer from observer bias, going back to the 'herd mentality' of operating on a given band when others have done the work of testing it first. As soon as the band opens, up goes the cluster report, and the herd follows.
I'm always open to new findings, but I've used public data like this in other fields before and instantly found it seriously lacking. That Kp is negatively correlated to Es is interesting - but may also be the wrong measure. Ap would seem to be more appropriate. When I see geomagnetic disturbances (and so high Kp), I see Es formations (and PMSE, as proxies of an intimately-related phenomenon), some 6-12 hours later.
All in all, interesting, but almost certainly fails to understand the pitfalls in the data and seems to use inappropriate tests.
Post a Comment