You are on page 1of 7

Are Crowd-Sourced Noctilucent Cloud Observations Reliable?

John Rowlands, July 2011.

Networks of geographically-dispersed NLC observers have operated, most notably within Russia, North America and Europe, for many decades. Since the internet became widely available to the public in the mid-1990s, the establishment of and rapid communications from such networks has become extremely easy. Leading the way in the gathering of NLC observations on line has been the NLC Observers' Network, established and maintained by Mr. Tom McEwan of Glengarnock, Scotland. Mr. McEwan's diligent and understated efforts over many years to allow instant submission of NLC observations are commendable and clearly undertaken in the spirit of increasing scientific understanding of NLC. The data from the archive, which is freely available to all from the web site (http://www.kersland.plus.com/), has been used for research purposes by a number of professional atmospheric scientists and has been acknowledged in several scientific papers. It is also very important to note that Mr. McEwan has shown a clear desire to ensure the observations submitted to him are made properly and as accurately as possible. Notable in this regard is his expression of concern in a post to the NLC Observers' Forum of June 2008, which may be read at: http://nlcnet.proboards.com/index.cgi? board=talk&action=display&thread=172 Whilst the NLC Observers' Network data is without doubt quite unique and very valuable, it is entirely reasonable to examine whether 'crowd-sourced' NLC observational data from this and other sources is, in fact, reliable such that scientists may draw proper conclusions from using that data. The following discusses what may be a series of relevant questions for anyone concerned with the scientific use of crowd-sourced NLC observational data. It in no way suggests any improper motivation of recklessness in the gathering of such data by the limited number of sites that do so. Most of the questions presented here have not yet led to an examination of the data for evidence as to whether the concerns are justified in practice. However, scientists should satisfy themselves of the answers before using data in scientific papers submitted to journals. An Internet-Based Free-For-All By its very nature, crowd-sourced NLC observations submitted via a standard automated online form can be made by anyone, regardless of experience, motivation or accuracy. There is no ready mechanism that can determine how reliable the observers and therefore the observations actually are, other than some community familiarity with the few dedicated and reliable observers. Therefore, it is possible that significant errors in the NLC reports exist within the data. It is difficult to assess, without good knowledge of the observers - which is typically absent - how extensive such errors may be. Of particular interest in terms of reliability has been an episode spanning two consecutive observing seasons. NLC typically occur in a highly-repeatable annually-occurring season that, from UK latitudes, generally spans the very early June to very early August period. At least one observer began submitting reports, including photographs of NLC that fell in the latter half of May where there were no corroborating observations from elsewhere, which

immediately raised questions about their validity. In some cases, reliable negative observations (where NLC were absent from a sky clear of tropospheric cloud) were made at the time of the claimed positive sightings. Whilst it would be foolish to dismiss without proper consideration any 'out of season' NLC, the images supplied during both years clearly held no evidence whatsoever of NLC and in most instances were readily dismissed as confusing contrast effects from light cirrus or haze against a twilit sky giving the appearance of bright streaks superficially similar to NLC. When the observations were tagged with a 'needs confirmation' label or similar, the observers in question took offence in the following manner, posted to the NLC Observers' Forum on June 07, 2008: I for one do not like 'confirmation required' under all of my May NLC sightings on the site! Clearly, it is a matter of great worry for any observer concerned with the proper validation of data that such a view was expressed. The user in question has not submitted observations subsequent to this event, raising further concerns about the very motivation in making the observations at all. Whether this was a case of over-zealousness in 'being the first' to sight NLC or simply a persistent case of misidentification remains unclear. Ability of Observers The ability of some observers must be called into question. For example, from 57 total observations during June 2000, 11 (19%) made no attempt at classification of cloud forms present. By June 2011, the situation had apparently improved but still showed a significant number of non-classifications; 9 of 99 total (9%). I suggest it may be reasonable to use such non-classifications as indicative of less reliable observations, either because they have not made the effort to do so or are unfamiliar with the classification system, possibly reinforcing the concern over reliability. Their observations are most likely genuine and made in a sincere spirit, but professional scientists intent on using such data without proper consideration of its reliability clearly would be acting without due care. Fake Observations In rare cases, NLC reports may be entirely fabricated. A celebrated recent case is that of an Iranian observation purportedly made from near Mount Sabalan in the north west of that country (latitude 38 degrees north). Although seemingly-genuine NLC reports from locations as far south as Portugal have been made in recent years, the high elevation of NLC and other elements of the image caused concern. The image showed extensive NLC in a bright twilit dawn sky, with a lake in the foreground displaying static water ripples, most likely from a thrown stone, with the initial water turbulence at the centre of the ripples still evident. Distinct reflections of NLC within this turbulence were also suspicious. Although a high-ISO setting on the camera could have led to a fairly short exposure of perhaps less than a second, the ripples do nevertheless appear suspiciously distinct. An article then appeared on 17th October 2008 at http://www.spektrum.de/artikel/970892 in which Professor Dr. Franz Josef Lbken and Dr. Gerd Baumgarten of the Leibniz Institute for Atmospheric Studies stated the image was, in their view, a falsification. They

submitted as powerful evidence for their claim a genuine NLC image taken in southern Sweden that had earlier appeared on spaceweather.com, and which matched exactly the NLC features seen in the Iranian image. The suggestion was that the Iranian image was in fact a composite. An examination of the NLC appearing in the two images demonstrates conclusively that the falsification claim was correct. Nevertheless, the Iranian image spread very widely on the internet and is still presented as genuine on the majority of web sites.

Figure 1. The Iranian fake NLC image (left). Water ripples are crisply captured despite the likely need for a long exposure of a second or more. The high elevation of NLC from such a southerly latitude was suspicious. The eventual discovery by the Leibniz Institute of an image with the precise-same NLC features and tropospheric foreground cloud showed conclusively that the Iranian image was a composite fake. The image was widely reported as a genuine NLC sighting on very many internet sites. Original, genuine NLC image from southern Sweden, bottom right. Crop of the same genuine image showing the portion of sky used in the Iranian fake, top right.

Human Factors Evening & Morning Sectors. The recording of NLC is also very likely to be influenced by human factors, of which extreme tiredness is the most obvious example. In the month or so either side of the summer solstice, there is a significant 'dark ages' period between the fading of any evening sector NLC and the subsequent appearance of a morning sector NLC. This can make continuing observations throughout the night very demanding, perhaps requiring a brief period of sleep. It is clear that few find the prospect of such a punishing regime attractive. A number of comments left on internet forums and social networking sites suggest that a significant number of observers will examine the evening sky for signs of NLC, which may

anyway not be very obvious to the inexperienced eye, and upon seeing none, will report a negative sighting (if they bother to report anything at all), and not make any effort to see whether the morning sector to which NLC first-sighting is heavily biased (see Figure 2) reveals any NLC. It is apparent that the most extensive NLC invariably occur in the morning sector, where a diurnal or semi-diurnal thermal tide lowers mesospheric temperature and thus leads to or enhances NLC displays.

Figure 2. Timings of first NLC appearance as seen from Anglesey, summer 2010. A strong bias towards the morning sector is evident. It is suggested that such a bias leads to a premature 'giving up' amongst some observers, who wrongly conclude that no evening NLC means no morning NLC will appear, for which they consequently do not make the effort to observe. Common texts on NLC promote this adverse behaviour by not highlighting this bias in first-appearance timing; indeed, the above plot may be the first of its kind to be published.

Human Factors The Working Week. Another very obvious human factor to consider is whether the working week has any influence on the timing and number of observations. It is to be expected that the need to maintain a vigil for NLC up until about 3am or slightly longer, allowing only four hours' sleep before work will be a strong inhibitor of observations within that period. Human Factors - 'Swarming' Another human effect is that surrounding 'swarming' of observers. This was evident during a BBC production involving a Facebook page concerning NLC during 2010, and again in a continuation, with no BBC involvement, during 2011. The effect is one whereby a few dedicated observers might notice NLC, who then report to an internet-based site. The news quickly gathers pace and within a few days, a surge in people making observations seems to take place. Whilst the interest and enthusiasm is commendable and to be encouraged, from a datareliability point of view, it is to be expected that many of these individuals will have little or no experience in observing NLC to a scientifically-reliable standard. There have been notable examples of observers submitting images of cirrus during the middle of the day and others wondering publicly on Facebook how anyone knows NLC are covering the sky when it is dark, having earlier portrayed themselves as capable observers. Indeed it is evident that a not insignificant number of contributors are intent on developing for

themselves a degree of respectability, skill and reliability that they have not earned. Swarming may also give rise to the appearance, unless carefully treated, of an increasing frequency of NLC occurrence, and/or to exaggerated reports concerning azimuth, elevation and brightness that may skew scientific analyses.

Figure 3. 'Swarming' of interest in, and most likely observers of NLC on Facebook. This may lead to skewing of results in any scientific use of crowd-sourced NLC data unless the authors are aware of the effect. Graph shows comments made on Facebook NLC site 2011 for the period June 03 to July 02 2011. The spikes follow NLC displays of note. A clear potential consequence is for under-observation during periods when NLC are less frequent or less extensive.

A Case of Misidentification Some scientific authors have cited data from spaceweather.com, a very popular web resource which draws heavily on image contributions from its considerable audience. Whilst the majority of observations are likely to be reliable insofar as the images submitted are genuine, there are some concerns. On one occasion in 2011, a large image with the headline Noctilucent Clouds Over Europe was displayed for at least two days on the spaceweather.com homepage. The owner of the site may have been entirely unaware of the error, although an e-mail to highlight the misidentification failed to attract a reply, correction or removal of the image until the next natural update some days later. The contributor of that image, Mr. John Houghton, who has kindly agreed for this case to be discussed and his image used in this paper, stated that the NLC were visible on 31 st May 2011 even before sunset. That assertion, together with an examination of the photograph, clearly shows that it was a case of misidentifying cirrostratus as NLC. NLC occur only when the sun has set to a considerable level below the horizon typically -6 to -16 degrees. Due to their tenuous nature, the presence of the sun in the sky will easily overpower NLC reflections. Reliance without very careful checking on submissions to such sites is therefore fraught with dangers. Note also that the spaceweather.com claim in the same report that NLC first appeared after the eruption of Krakatoa does not take into account a persuasive observation made by Thomas Romney Robinson (see: http://climate.arm.ac.uk/publications/noct-paper-rev.pdf), then-director of Armagh Observatory in Northern Ireland dating from 1850 33 years prior to Krakatoa's 1883 eruption.

Figure 4. A beautiful image of cirrostratus that the author, John Houghton who has kindly agreed to the use of his image for this paper, genuinely believed was noctilucent cloud. Spaceweather.com nevertheless ran the image as one depicting 'Noctilucent Cloud Over Europe'. Mr. Houghton, in granting his permission for the image to be used, pointedly stated: If it would help to illustrate to the scientific community the perils of accepting unsubstantiated information sourced from the Internet as fact, then please go ahead and use this as an example. (Houghton to Rowlands, Pers. Comm. 04/07/2011.)

General Problems Not Specific to the Internet. Brightness estimates are notoriously difficult to standardise across a world where one observer may be contending with background brightness, most notably caused by light pollution, whilst another may be observing from a completely dark-sky site. Whilst some of the 5-point scale makes reference to how starkly NLC stand out from the background sky, there is also a reference at brightness 5 to 'illuminating objects facing [the NLC display]'. In light polluted areas, determining whether an object is being illuminating by NLC is most likely impossible, leading to the question: how do observers in such situations properly assess NLC brightness? Conversely, from very dark sky sites, the summer twilight causes buildings, cliff faces etc. to be permanently illuminated, again calling this method into question. A rather tongue-in-cheek solution (the 'bright-o-scope') was constructed by the author during 2010, and is illustrated below. It limits stray light entering the device, which examines only a small area of sky pointed directly at the NLC display. Numbers printed in a diminishing-sized font become clearly visible at different brightnesses, making a more objective assessment that is also more comparable between sites possible. The effectiveness of the device has not yet been widely tested.

Figure 5. A zero-cost, possibly effective solution to improving the accuracy of NLC brightness estimates. The toilet roll tube limits the area of sky being examined, reduces extraneous light ingress. A small card with numbers increasing in font size from 1 to 10 is fixed at the rear, the idea being that smaller numbers become clearly legible with increasing NLC brightness. It has yet to be tested properly.

Location? Location? Location? Finally, consideration as to the location of observers must be given. It is clear that, ignoring ship and aircraft-based observations, the vast majority of NLC observers are based in Europe, with a small number based in the USA and Russia. Whilst this does not of itself necessarily lead to bias in the conclusions that may be drawn from the observations, the increase in reports from more southerly states of the USA and Europe when conditions lead to particularly striking NLC displays must cause us to wonder whether there is under-observation of the sky at other times from these locations. As earlier indicated, if there is only a very low probability of sighting NLC from these locations, it is unlikely that the considerable effort involving the loss of very many hours of sleep in order to maintain a regular vigil for the two summer months will be made. Therefore, lowelevation and non-extensive NLC may well be appearing more often than currently reported.

You might also like