Applied Field Engineering · The traverse
Walking a loop, and proving you walked it
score 0

Walking a loop, and proving you walked it

A traverse is a chain of measured courses round a parcel. Each course carries a length and a direction, and the whole method rests on turning that pair into two components that can simply be added up.

Directions are booked as azimuths: α\alphaalpha — measured clockwise from north, from 0° to 360°. Its reverse is the back azimuth, αb=α±180\alpha_b = \alpha \pm 180^{\circ}, wrapped back into that range. Same line, looked along from the other end — and it is what the instrument is set to when the crew sights back to orient at a new station.

Each course is then split. The latitude is the north–south component: Lat=Lcosα\text{Lat} = L\cos\alpha. The departure is the east–west component: Dep=Lsinα\text{Dep} = L\sin\alpha. LL is the horizontal course length in metres and α\alpha its azimuth. Cosine for latitude, sine for departure — and the reason is that azimuths are measured from NORTH, so a course pointing due north is all latitude, and cos0=1\cos 0 = 1. Swap the two and every plot closes beautifully in the wrong place. The signs are not decoration either: they carry south and west.

Round a closed loop the latitudes must sum to zero and so must the departures. They never quite do. The residuals ΣLat\Sigma\text{Lat} and ΣDep\Sigma\text{Dep} stand at right angles, so the gap between where the crew finished and the pin they started from is the hypotenuse: Ec=(ΣLat)2+(ΣDep)2E_c = \sqrt{(\Sigma\text{Lat})^{2} + (\Sigma\text{Dep})^{2}}, the linear closure error in metres.

Then the number that actually gets reported. N=PEcN = \dfrac{P}{E_c}, where PP is the total perimeter walked and NN the precision denominator, written 1:N. A 0.20 m miss round a 1,200 m loop is 1:6,000 — good boundary work. The same 0.20 m round a 200 m loop is 1:1,000, and would be rejected. That is the whole argument for a ratio: error only means something next to the distance it accumulated over.

And read the ratio the right way round. NN is a denominator. One part in eight thousand is finer than one part in three thousand, so a bigger N is a better traverse. It reads backwards to the eye, and mistaking it is how a rejected loop gets signed off.