UKH

Hey, so I was hoping someone on here might know of a good resource to learn 8 figure grid references. 

I'm solid with 6 figures and nav but am joining an MRT so need to learn.

Thanks in advance 

Alex 

It's exactly the same, just to a greater precision - each 100m square is divided up into 10 again to give you a reference accurate to 10m.

So if the eastings were 1066 would the first 3 (ie106) be 6 figure reference and then the last digit (6) be the sub division of the penultimate digit (6)

Yes.

It will also depend on what map you’re using and whether you’re using GPS or not. If you’re just using 1:25000 map and your eye then the 4th and 8th digits are usually (or always used to be) a 0 or a 5 for simplicity. Obviously if you’re using GPS you’ll get more accuracy. 

But I would check that still holds true. 

We like our MR trainees to be able to use a romer properly to give a reasonably accurate 8 figure grid ref (and to convert accurately - a 10 figure grid ref taken from a GPS doesn't really give anything more than an 8, and in cases where it might matter - top / bottom of crag or river left / right, it's probably better to get a verbal description.

All our team members use GPS on callouts, whether dedicated units or phones. But we also require them to be able to navigate accurately using map and compass - you never know when the battery'll die, and maps are more convenient for planning search areas and strategy.

The OSGB36 grid starts with a 5x5 alpha reference, labelled A-Z from top left to bottom right, missing the letter I. Each of these grid squares is 500km x 500km. This grid covers an area that extends significantly outside the British Isles.

Each of these 5x5 squares is then further divided into 5x5 alpha squares, each of which is therefore 100km by 100km.

So, the two-letter alpha prefix on a grid reference refers to a 100km by 100km square.

If we then add a single digit for each of easting and northing, we divide the two-letter alpha grid square into a 10x10 grid, with each square measuring 10km by 10km.

Add a second digit to each E/N and we get a square 1km by 1km.

Add a third digit, and we get a square 100m by 100m. This is the 'six figure grid reference'.

Add a fourth digit, and we get a square 10m by 10m. This is your 'eight figure grid reference'.

Add a fifth digit, and we get a square 1m by 1m.

If you look at the border of an OS map, you will see numbers in the four corners. These are the purely numeric grid references of the four corners of the map, measured in metres from the false origin of the OSGB36 grid (that's the bottom left corner of the alpha grid square SV). The larger font numbers in these values are km. The smaller font numbers are metres (or vice versa; can't remember which, and don't have a map to hand at the moment).

I developed a 'vernier romer' nearly 20 years ago that allows you to accurately measure an 8-figure GR from a 1:25k map. Given that 10m on the ground corresponds to 0.4mm on a 1:25k map, there's little point trying to read a map beyond 8 figures. I have a vague recollection that the OS have an rms error specification of 7m on 1:25k mapping.

To get beyond 8 figures, you'll need a GNSS receiver, and, even with EGNOS providing some correction for ionospheric and tropospheric effects, you won't get much better than 3m accuracy with a commercial receiver. And that can easily be fooled by reflections off mountains. You need to be aware of that, and not rely blindly on the GNSS receiver; always think about whether your local environment might be causing shadowing of the direct path to the sky, and reflections of large slabby faces.

Grid references are so last decade - all you need is 3 random words  

Get.F*cking.Lost

^ This.

Use a grid reference but use it sensibly in combination with logical directions.

A 6 fig reference qualifying 'North bank' is better than an 8 fig (or 3 word) reference that might put you 2m away on the S bank of a fast 2m wide river. 

I agree, I don't own a GPS for that reason. It's why I'm a bit baffled with 8 figures and roamer 

You will get numbers to more precision, but that doesn’t mean they are more accurate.  

Most phone GNSS display apps are useless at giving decent estimates of accuracy.

Yes indeed, there are times and places for 6 figure refs and 8 figure refs. 6 figures are easier to relay over the radio; 8s useful in featureful areas.

A handheld GPS will tell you what accuracy it thinks it is able to provide from the signals it is receiving. However, checking them against accurately measured survey stations (+/- 20mm or 0.020 metres in Easting and Northing) the GPS states say 3 metre accuracy, but comparing to the survey stations typically 5 metres out E-W and 15 metres out N-S.

I have noticed this discrepancy on several different construction sites, it seems quite consistent, my theory of explanation is that the satellites are in a polar orbit so less accurate timings and hence position calculation are possible in that axis. Not looked into it  in any detail, any thoughts?

When I’ve done noise and error analysis on GNSS signals, it’s not had a long term (>2 hours) directional preference in terms of 2D displacement, just the well recognised poorer accuracy in altitude.

Their orbits aren’t polar but various different non-equatorial planes, so that’s out as a theory.  Fun aside fact - you can actually get more information out by measuring the Doppler shift in the signal which gives you better information in the orbital plane.  About 18 years ago (at least) this was used to get sub-meter accuracy in some tests.

If you see a consistent offset it may be in the grid units the GPS and ground reference points are using.  There’s quite a rabbit hole to go down on different grid systems and calculating origins and offsets from origins.  If you’re comparing OSGB try switching to lat/lon.

if you can log position on your unit, try leaving on location for 4-5 hours, then subtract the mean position and scatter plot the residuals - is it circular in shape or elliptical? 

Mobile apps seem poor at this.  I last used the iOS nav API about 5 years ago and it passed very limited information to the application on nav accuracy. I haven’t used a newer handheld device than my now vintage Garmin 12.  Critically the signals themselves don’t contain enough information to allow a receiver to produce a good estimate of accuracy under all conditions.  In particular receiver local effects like reflections and heavy foliage are unknown; the degree to which their effect is detectable in the position calculation is not fixed.  

When I was practising 8 figure grid nav to join an MRT I went up onto Kinder at night with another aspirant, we picked precise  but random features on the map - bends in contour lines, kinks in reentrants, spot heights or the centre of a flat area etc and navved to them. When we were both happy we calibrated ourselves with a GPS. Then chose another spot not less than 200m away. After 10 or 12 it is amazing how accurate you can get. Then repeat a few nights later. The fundamentals are the same as basic nav but I recall that things like slope aspect and angle aquire more significance. 

One more tip, if you take a bearing off a particularly bright star, just watch it for a few minutes.  We were following a bearing on a good star, and were a little dismayed when it turned left to begin its descent to Manchester airport.

Handy also getting your head around UTM and Lat / Long. 

Not everyone navigates with OSGB

And remember with GPS to select correct system ie OSGB36, WGS84 for that system

UTM is essentially similar to OSGB, except that it breaks the entire Earth's surface into a number of zones, rather than just one. Since the zones overlap, and the projection parameters are different for each zone, it's important to be clear which zone you are using; the coordinates for any one point in the two zones are quite different.

It's really not that baffling.

A 1:25k romer division is 4mm. You really ought to be able to estimate the position within such a division to at least 1mm, or 25m. With a bit of practice, you will be able to estimate to give an extra digit, or 10m, or 0.4mm on the division. I can estimate to 0.1mm with a 1mm graduated ruler.

It really ought to be Romer, since it's named after the person who invented it:

https://en.wikipedia.org/wiki/Romer

To make 8-figure properly usable, you need mapping at 1:25000 or better. That's not always practical unless you are dealing with honeypot rescue areas where it's all happening in a relatively small area. On 1:50000, 8-figure is a bit of a stab in the dark. Add real dark, maybe some rain and wind, and it's a waste of time on paper map.

Once you go to digital mapping, like Viewranger for instance (free maps for UK MRT!), the 8-figure can be your normal precision and you probably have 1:25000 zoomable to make it practical. 

If you are failing to grasp what 8-figure really means in practice then start by thinking about what a 2-figure reference means (10km x 10km), 4-figure (1km x 1km), through to 10-figure (1m x 1m). 

MRT rescue co-ordinators often have to deal with unusual location references using a range of positioning systems. In my area, police sometimes come up with 12-digit numeric full grid references. This defines a 1m square with the position measured in metres from the OS Grid origin (same as a 10-figure reference but without the two-letter prefix). Useless for those on the ground until they covert it to an ordinary 6-figure reference. 

There is 'Education' stuff on the ordnance survey website. Like here: 
https://www.ordnancesurvey.co.uk/resources/maps-and-geographic-resources/th...

Someone referred to using a 10-figure reference to get an accurate 8-figure. This is the case if you use an electronic application that does rounding of grid references instead of truncating. Lat-Long can be rounded but this does not work properly for grid references. The combat that problem, people use one precision greater and truncate from there. Taking a 10-figure and removing the 5th and 10th digits leaves a correct 8-figure reference.

Here are some mobile apps for OS grid references offline.
GRID REFERENCE by Arthur Embleton, for Android
GPS OS GRID REFERENCE by Tomtrails, for Apple
OS LOCATE by Ordnance Survey, for Android or Apple.
 

Delving into the Apple iOS docs, accuracy of the location fix isn’t well reported - and it doesn’t tell you if the fix derived from GNSS or WiFi location.

https://developer.apple.com/documentation/corelocation/cllocationaccuracy

The system reports the accuracy as “Best”, “10 meters”, “100 meters”, “1 km” and “10 km”.  

It does not specify what “accurate to” means - guaranteed max error?  Variance of the error?  Radius of circle of probability?

Someone's been watching QI. 

Anyone care to explain the dislikes? Because I said you need to be aware of the potential error causes for GNSS?

I worked for a number of years developing GNSS, including the 'golden reference' simulator for ESA' s Galileo. I'm not an anti-GNSS luddite, but I am aware of the potential problems of natural and urban canyon effects.

Lat/Long is all explained here:

https://xkcd.com/2170/

CB

Not from me, but the post does go on a bit. 

Perhaps you have deeply offended a trig point bagger with this line - “To get beyond 8 figures, you'll need a GNSS receiver”.

Beyond people just believing digits on a digital readout, my other issue with 8-figure grid references is the accuracy of maps at that scale.  OS maps can be a little whimsical in terms of fine detail sometimes...  A lot of interesting work going on now in making deeply sub-meter resolution 3D maps for autonomous vehicle positioning.

Ah, the good old effect of attention spans reducing...

I see; providing comprehensive answers is a bad thing now, is it?

I've started using 8 figure references, but then our local map is at 1:15k so it's actually worthwhile, especially in our landscape of tiny little lanes.

Well, no, but the OP already understands 6 figure grid, so a great chunk of what you wrote is superfluous.

Not really. You should write to your audience.

I thought the OP's replies suggested their understanding of grid references wasn't that clear, since they didn't seem to understand how you add an extra digit. So I went back to first principles to explain how grid references are constructed, in the hope of helping the OP to understand what increasing number of digits mean.

Regarding increasing numbers of digits, I am at what I refer to as work atm.

The architects often give me 18 or even 28 figure grid references.

I can assure this audience that their understanding of this subject can be quite frustrating , don't get me started

I was checking out the Taylor burn exit from the lairig ghru for the missus the other year . I got a bit confused trying to position myself exactly the map I had with me about halfway up .

Checking the various maps , several os and the Harveys, we have at home , I found out the source of the confusion .

All the contours for that section of hillside were different on the maps.

Hopefully one was correct .

That was certainly one of my concerns and is why I referred to 2-figure through to 10-figure and then gave the OS link. There is a sense of the OP having just a little window on the grid reference idea through which he only sees the 6-figure picture. It's quite hard to open that up to the big picture from the end of an ADSL line.

Seriously? 18 digits correspond to 10 nm (which is in the range of a bog standard electron microscope), 28 to something smaller than an atom even if not yet close to the Planck length. And I thought the XKCD link I posted above was joking!

CB

18 digits is 9 each for kilometre, metre and millimetre in both axes. So to the nearest millimetre if they are in the correct OS letter block.   Up to 999 km.  You also need to drop the first digit off the nine to get it looking like an OS grid ref.  

Now try looking in the Garmin handbook for where it explains that a 10 figure grid ref display is actually in metres, I could not find it.

Get involved with various  map projections and scale factors like wintertree and Capn P and it really gets confusing.

I'm guessing that the 28 digits are a purely numeric GR. Which means 14 digits for E and N. Assuming this is to cover the UK from the false origin, which requires 800 by 1400km, we need four digits for the km. Leaves us with 10 digits sub km, and so 7 digits for sub metre. i.e. a resolution of 100nm.

I think even SketchUp allows you to specify dimensions down to silly resolution, so I'm guessing their drafting tool offers something similar, and they haven't selected a more sensible resolution for their drawings (1mm sounds a bit more reasonable than 100nm...).

Ok, I see. I was thinking of 18 per axis.

CB

I understand the principles of it, just trying to work out how to get the 4th and 8th digit when the map/roamer isn't graduated as such 

You have to estimate by eye. As I said, you can easily estimate to 1mm in the 4mm graduated romer division. With practice, you can estimate to read the extra digit in each axis.

My Vernier romer removes the need to estimate, and allows you to measure the fourth digit using a graduated scale.

Smashing, thank you

Alex

When you're out in the field with a 1:25000 in hand really the best you can usually do is to the nearest 25m (1mm), I do have a ruler with 0.5mm graduations so in theory I can do 12.5m steps.

A trick I sometimes use is to use the 1:50000 graduations on my compass on a 1:25000 map to give to the nearest 50m and then go halfway inbetween to get the 25m. So for instance I could quote SK187851 for a trig point on a hill using the 1:25000 scale rule, but get SK18675 85100 with the 1:50000 scale rule and estimating the closest 25m.

For most of what goes on out there that's more than enough, and in reality someone always has some form of gps and or tracker (eg traccar on their phone).

Thank you very much