This session was relatively typical of what has been recently experienced on 630-meters. The noise level was slightly elevated due to terrestrial storms in the Midwest of the US and participation was down slightly (71 MF WSPR stations observed at 0245z) but reports were generally very good overnight with many single digit S/N levels experienced here in Texas at WG2XIQ. I feel certain these observations were repeated by a number of stations around North America and the world.
Neil, W0YSE/7 / WG2XSV, experienced a very good session with a new personal record. He details the session and the accomplishment below:
The geomagnetic field remained quiet through the session with a slightly south-pointing Bz and slightly elevated solar wind velocities. I do not believe that the elevated solar activity that was expected actually materialized as expected. Perhaps the roller coaster has stopped for the moment.
Phil, VE3CIQ, notes the impact of the longer days on his operating but he made the most of his operating time and provided the following details:
Ken, K5DNL / WG2XXM, reports that he decoded twelve WSPR stations and was decoded by 35 unique stations. He decoded WH2XCR 17 times, best S/N at -18 dB. Ken also decoded Ken, W8RUT / WI2XFI, who was only operating at 200 mW ERP, 31 times.
Larry, W7IUV / WH2XGP, reports down band conditions compared to the previous session. He decoded eleven WSPR stations and was decoded by 34 unique stations with no VK stations during this session.
Joe, NU6O / WI2XBQ, reports that band conditions to the East were quite good from his station in Northern California.
KE0FMX represents the sole new receive station observed in North America during this session. Welcome aboard!
Regional and continental WSPR breakdowns follow:
There were no trans-Atlantic or trans-African WSPR reports during this session. UA0SNV was present but no reports were found in the WSPRnet database.
Eden, ZF1EJ, and Roger, ZF1RC, continue to provide reports to stations in and around North America, including reports for WH2XCR in Hawaii.
In Alaska, Laurence, KL7L / WE2XPQ, experienced another typical session, with reports along the west coast of North American and Hawaii. Nothing wrong with consistency!
In Hawaii, Merv, K9FD/KH6 / WH2XCR, experienced a strong session with Australia, including two-way reports with VK3ELV and VK4YB and reception reports from VK2DDI and VK2XGJ. Merv noted that an approaching weather system might provide the necessary impact to allow the path to Japan to open again based on previous observations of band behavior but it does not appear that all of the elements were present for a JA path during this session. The eastern portions of North America made a strong showing. The commonality of longitude is interesting for these eastern stations. That may just be coincidence in stations that were present.
In Australia, Phil,VK3ELV, and Roger, VK4YB, experienced two-way reports with WH2XCR. Phil received additional reports from JH3XCU and TNUKJPM.
Jim, W5EST, provides a discussion on a new way of looking at long paths, entitled, “SLIDING WINDOW APPROACH TO 2015-2016 SEASONALITY: LONG PATHS”:
“I’m planning to do a sliding window approach to describe long path seasonality, starting with transatlantic (TA) 630m WSPR decodes.
What’s a sliding window approach?Step 1, one gathers counts of nightly decodes over the season as the raw information. If no TA decodes occurred on a given night, then the TA count for that night is zero. Step 2, for each given night calculate the average of the decodes over a given number of nights in a time “window” that includes the given night. For instance, the time window can be several days or even weeks long centered on the given night. Step 3, take the average over the window for a given night, and then advance the window one night and repeat the averaging process, night by night until all of the season’s data is averaged for all the nights. It’s basically a windowed running average.
The sliding window approach spreads some decodes from nights that yield more decodes onto nights of fewer or no decodes. The result no longer precisely describes the particular season from which the averaging was done. However, the approach recognizes that in another season those nights that had fewer or no decodes this season may be more fertile. Accordingly, the sliding window approach is justified when one is trying to get a sense of likelihood of decodes from one season to next. The goal is to get a sliding window average pattern from one season’s data that wouldn’t be too different from the sliding window average pattern that you would obtain from another season’s data.
A sliding window average performs like a really-low, low pass filter. From a statistical point of view, one pictures the 630m TA season as having nightly long path probability that slowly varies as the season progresses. Factors like storms and operator activity decisions noisily affect the actual numbers of decodes that occur on each given night.
You could say the receiving stations are sampling the 630m long path band conditions from night to night. The nightly numbers that we count arise from sampling a seasonal distribution that has slowly varying underlying long path probability or trend. Samples statistically fluctuate. Averaging them helps diminish those fluctuations that make the numbers noisily depart from an underlying trend that we want to reveal.
Because of the averaging, the seasonal data of a given year also gets spread out several days earlier and later than the year’s actual season. I suggest that’s okay to impute some probability of TA earlier and later. Let’s push the boundaries of previous experience.
You could put on an imaginary statistician’s thick glasses and green visor, and get really technical, and object to the sliding window. By way of objection, recognize that the WSPR decoder is delivering decodes only when the signal is above the decode threshold, and even then with a varying probability. The technical statistical word for this is “censoring.” If you call the “signal” here the ongoing numbers of nightly TA decodes, a sliding window approach does not just take some average of signal that has a little noise on it. The signal here is zero a lot of time and subject to large censored variations due to many different circumstances. That said, I want to start somewhere and learn lessons along the way.
Indeed, any content contribution from somebody with a rigorous statistical approach or just plain common sense to advance the topic of 630m seasonality is most welcome to blog it here. For instance, what about the 11 year sunspot cycle? Perhaps two or three consecutive years might be comparable enough to each other to justify this sliding window idea, but how should one numerically interpret TA given the sunspot cycle?
For now, my work plan calls for a Step 1A and 1B to count nightly long-path information two different ways and then do Step 2 and Step 3 sliding window averaging on the counts taken each of the two ways.
Step 1A calls for an activity-oriented reception spreadsheet column that counts up every decode of the same TX obtained at different RX stations in the same timeslot and states the entire number of TA decodes for each whole night. The data from this column effectively combines TX and RX station activity and records overall nightly numbers of long path decodes. Over the course of a long path season, this way of counting recognizes that there’s a human factor behind station operation levels that combines with the propagation factor. Human operators decide to turn the station on or not. Each TX human operator decides what WSPR transmit percentage (TxPct) that station will run.
From a physical 630m propagation viewpoint, however, conditions are what they are regardless of how many RX stations accumulate decodes. So for Step 1B, a separate propagation-oriented reception spreadsheet column counts just once the occurrence of a decode in each given time slot from a given TX notwithstanding that the same TX slot may be decoded at two or more RX stations simultaneously. From a digital viewpoint, this column “ORs” the receptions of all the RX stations in each timeslot and shows the nightly total number of timeslots that yielded TA. That also helps reduce the effects of storms on the data and TxPct variations from TX to TX .
Over the course of a TA night, each RX station may receive some time slots that none of the others receive. Regardless of which RX station does a unique decode, that decode counts as one decode in Step 1B for the propagation counting purposes. That’s good, because we can expect long path propagation will shift around geographically. When more RX stations are active, they sample more of that geographic variation in long path propagation. And one can expect that the counts in this propagation-oriented column will be acceptably augmented somewhat if more stations are active to provide more information.
If the 630m season features an extended interval of storms that blanks TA, that long interval may get through the “low pass filter.” To some extent, propagation-oriented results of the sliding window approach will inevitably include some effect of whatever storm season confronts the constellation of 630m RX stations. If so, maybe that’s a good thing because, after all, the data will reflect the weather reality that 630m long path operators actually face.
Here’s where this project stands today: I have the transatlantic TA and N.Am.-VK transpacific TP raw decodes for August, 2015 to early April, 2016. There remains the hard work to count them and tabulate the counts on a spreadsheet and then perform the window averaging. We will see how long this TA sliding window project takes. In a future blog post I’ll plan to report the TA results along with any lessons learned!”
Additions, corrections, clarifications, etc? Send me a message on the Contact page or directly to KB5NJD <at> gmail dot (com)!