The Angry Corrie 69: Nov 2006-Feb 2007


Steam-driven altimeter no more

INTERESTING DEVELOPMENTS on the Knight's Peak front, the best part of a decade after the controversy about its status blew up. To recap, Knight's Peak is at NG471254 on Landranger 32 and Explorer 411, fourth pinnacle on the north ridge of Sgurr nan Gillean. Its height has long been a matter for debate, as some argued that it was a 3000ft summit and hence a Munro Top (Robin N Campbell has long held such a position), but the steep broken terrain led to difficulties in surveying, and for many years the consensus appeared to be that KP was just below 3000ft.

Then, in 1997, the revised edition of Munro's Tables listed KP as a Munro Top. The height given, 914m, didn't come from any Ordnance Survey measurements, but from an on-site altimeter reading taken by some SMC members. Not only was the methodology controversial, but the height claimed implied a remarkable degree of accuracy. A height of 914.3m is under 3000ft (hence Beinn Dearg in Torridon and Foinaven can be legitimately listed as 914m Corbetts: the assumption is that they're somewhere between 913.5m and 914.3m), whereas 914.6m would definitely be rounded up to 915m. The OS sometimes round "point-five" readings upward, sometimes down, so the height given for KP in 1997 implied that the SMC believed it to be 914.4m or 914.5m.

This led to scepticism from Ken Stewart (TAC35 pp10-11) and Alan Blanco (TAC36 pp4-5), and when TACit Press published Corbett Tops and Corbetteers in 1999, KP was listed there as a 912m Corbett Top, giving it a strange dual status depending on whether the SMC or TACit list was being used.

Things went relatively quiet for a number of years, but it was known that the SMC was attempting to obtain a height for KP from the OS. This has now been forthcoming. On 8 August, former SMC president Ken Crocket posted this on the relative hills newsgroup (http://groups.yahoo.com/group/rhb/message/11329):

My contact in the OS has just informed me of the results of a re-survey of Knight's Peak. And the answer is 914.95m. This will of course be rounded up to 915m, and this new height will be published in future editions of the Explorer map. The re-survey used high order GPS and Photogrammetry and is definitive.

This led to various queries from other members of the rhb forum, such that Crocket added this clarification on 10 August:

I have asked re OS methodology regarding Knight's Peak and received the following answer [...] "...the photo model was controlled using sub 0.1m accuracy GPS (ie points on the ground were fixed that could be identified on the imagery - GPS was not taken to the summit). The accuracy of the imagery heighting using this method is quoted as +/-0.5m for the Z (height) value."

So the next OS map will see KP fitted out with a 915m spot height, although quite when remains to be seen. And 914.95m is unequivocally above 3000ft (it's 3001ft 9in), so doubt over status has been removed, even allowing for the +/-0.5m accuracy. As Ken Crocket noted, "It's definitely a Munro Top!" This doesn't change the total of Munro Tops: there are still 511 (284 main Munros plus 227 subsidiaries). Corbett Tops however drop from 669 to 668, although there are no plans for a new edition of the booklet.

TAC would be interested to hear "from the floor" about all this, so feel free to write in. For now, though, here are a couple of comments from informed parties with a particular interest in such matters:

Chris Crocker, of Statistical Topics in Hillwalking, http://www.biber.fsnet.co.uk/

The 912m spot height on the OS's 1:10000 digital height product Land-Form PROFILE would have been by photogrammetry from the metric survey, which is +/-3.3m according to Brian Harley's Ordnance Survey Maps, A Descriptive Manual. I can't be sure of the confidence level on this error as it's not an official OS figure, but the general OS rule when quoting accuracy is to use +/-3 times standard error plus bias, which gives well over 99% confidence for normally distributed errors.

I spoke to an OS guy about spot heights years ago and he said that in addition to measurement error it was often difficult to spot the highest point of a summit on the screen. Perhaps this was a problem with KP being so spiky.

The +/-0.5m quoted by Ken Crocket doesn't come with a confidence level, and the standard error isn't given either, so it's impossible to qualify it. The method of measurement is clearly more accurate than the default method used for spot heights.

I wouldn't infer anything about claimed accuracy of the SMC's 914m figure from 1997, as their "precision altimeter" would have given a height to the nearest metre. I remember that when the draft Tables came out, the height was given as 924m. Perhaps they failed to correct for temperature - few modern-day users of altimeters know about this, and at 3000ft the correction will nearly always be negative. It seems unlikely that so controversial a promotion would suffer a misprinted height even in a draft document, so did the SMC suspect their measurement was dubious but decide it was too late to retract the new top, and therefore reduce the height to 914m in a damage-limitation exercise?

Pete Child, chief cartographer, Harvey Maps

We are interested to see this new figure and express mild surprise at the increase in height, but we bow to the OS's higher precision equipment. The (small-scale) photos we used in 1997 appeared to indicate a height lower than 914m, but we could not be more precise due to the rocky nature of the top. Larger-scale photography can be very accurate, but it is doubtful that it could get to within 5cm accuracy.

A high order GPS could get to this level of precision, particularly in plan position, assuming the accuracy of the USA satellites - but can we rely on them? After all, wasn't it NASA that got metric and imperial measurements mixed up on an early Apollo mission?!

It will be interesting to see if any of the other nearby summit heights - eg Sgurr nan Gillean, Am Basteir - have become higher. Maybe an OS surveyor will be dispatched to climb all the hills checking heights. How many changes to the lists might we see then?


TAC 69 Index