How Accurate Is Aerial Survey? *

H. G. DA WE, Manager, Survey and Production Hunting Aerosurveys Ltd., London, England

HE speed with which modern develop­ since differences in scale are automatically T ment programmes can be executed today accounted for. The relationship between de­ calls for a corresponding speed in the prelimi­ tail, say, on top of a cliff and at the bottom nary stages of survey and planning work. To is recorded wi th the same precision as if they carry out accurate surveys using ground were on one horizontal plane. Furthermore, methods alone is a costly and lengthy proce­ in the plotting instrument it is not necessary dure involving large amounts of men and to follow a circui tous path to go from one equipment. The answer to this need for point to another; where access to a site is dif­ speed in the preparation of maps and plans is fietdt or dangerous, across electric railway the use of aerial survey techniques. lines, for example, the major part of the sur­ Many people tend to regard aerial survey vey may be completed in complete safety. as insufficiently accurate, particularly where Broadly speaking, surface levels on clear large-scale plans are called for. The fact is, ground can be measured from aerial photo­ however, that the precision of modern photo­ graphs to an accuracy of between 1/5,000th grammetric equipment makes it possible to­ and 1/8,000th of the height from which they day to set up aerial photographs in stereo­ were taken. This means that photographs scopic plotting machines with such exactitude taken from an flying almost a mile that they may be observed and their details above the ground will yield measurements in recorded entirely free from any form of dis­ the plotting machine accurate to within one tortion. The article on the opposi te pagel foot. From lower flying heights the accura­ gives an indication of the precision of the cies will be proportionately increased, until modern survey -a precision which is the stage is reached where the photographic fully maintained throughout every stage of and photogrammetric aspects are no longer the photogrammetric process. the limiting factors. In many cases has con­ At this point, although plotting instru­ siderable advantages oyer ground survey ments are capable of finer measurements, methods, since it can provide a far truer rep­ they can be used only to record what the op­ resentation of the actual shape of the ground. erator can actually see. For example, in con­ The smallest detail can be closely followed in ditions that permit measurements to a the three-dimensional photographic "model" ± 2-3" accuracy, the subject may be a stub­ and contours can be observed at every point bly field or an ill-kept front garden and ob­ along their path. The accuracy of these con­ servations must of necessity be made on the tours can normally be guaranteed to within top of these features. In practice, an allow­ one-half of the interval; thus a specification ance may be made but under such conditions calling for one foot contours implies (if it does it is not possible to guarantee all heights to not actually state) that the tolerance is plus the accuracy of ± 2-3". Only on clear hard or minus six inches. In other words, no con­ surfaces can such an accuracy apply. tour deviates from its true path by more than A further point concerns the scale at which half the interval. Ground surveyed contours, the height data is presented. In a case where on the other hand, must be interpolations of a heights are required along a given line indi­ grid of spot levels, so that they cannot folio\\" cated on, say, an Ordnance Survey 1/2,500 the exact shape of the terrain. plan, the ground surveyor can record heights Variations in elevation, even violent ones, to 1" accuracy-precision which the air sur­ present no problem to the aerial surveyor, veyor could not hope to achieve. However, the leveller's stave could easily be moved by 1 "Camera Calibration" by R. Hall, Research an amount not plottable at that scale to show Officer, Hunting Aerosurveys Limited. differences in readings of as much as three to * This paper was included ill Aerial Survey Review, 1959 No. 21, published by Hunting Aerosurveys, Ltd. -1-, Albemarle St., London \V.l, England. For the privilege of reprinting in PHOTOGRAMMETRIC ENGINEERING, thanks are extended and credit is given to the author and to the publication. 759 760 PHOTOGRAMMETRIC ENGINEERING six inches. In fact, then, a specification call­ and based on a framework of sound ground ing for -1-" accuracy at that scale becomes control, there is no reason why aerial survey meaningless. should not attain any normal standard of ac­ Photogrammetry can always provide detail curacy required. Each stage of the survey is to within the plottable limits of the publica­ carried out by extremely precise instruments tion scale and complexity in detail-shape can and supplementary field checks provide any be dealt with as easily as regularly laid out detail which may be in doubt. \iVhere con­ features. In large-scale surveys exhaustive tours are required there is no doubt that it is field examination follows the photogrammet­ superior to ground methods and in addition ric plotting. Very often details which are re­ it has the considerable advantages of speed quired to be shown on large-scale plans are and economy of manpower. obscured in the vertical view from the air and The three basic advantages can be summa­ in some cases are too small to register on the rised as: photographs. 1/480 railway plans, for exam­ SPEED-Once the photography 1:S acquired ple, include such details as signals, mile and plotting can proceed rapidly, utilising a num­ gradient posts, telegraph poles, lamp posts, ber of machines if necessary. manhole covers, inspection pits, watercocks ACCURACY-Maps prepared from aerial and notice boards, all of which are plotted to photographs can be as accurate as those pre­ an accuracy of three inches. vVhere this type pared entirely by ground methods, and su­ of detail is obscured it can be readily supplied perior when true contour positioning is essen­ on the ground by simple offset measurements. tial. Normally, however, for 1/500 plans 80 to 85 ECONOMY-The need for large teams of sur­ per cent of the detail is plottable by photo­ veyors is eliminated. Costs are comparable grammetric methods. with ground surveys and can on occasion be I t is well known that aerial survey de­ less. pends in the first instance upon a network of ground surveyed control points so that over­ "How ACCURATE Is AERIAL SURVEY?" 2 all rigidity of the survey is ensured by the AN EXPLANATION precision of the field control rather than by This article was originally published in the any photogrammetric process. In the case of Aerial Survey Review earlier this year. Its pur­ a strip road survey, for example, the accuracy pose was to emphasise the accuracy of mod­ of the total distance from end to end is a re­ ern photogrammetric techniques and it was flection of the accuracy of the basic field in no way intended as a reflection on the work. It has been found from experience that classical methods of . To avoid any detail plotted from aerial photographs is in possible misunderstanding we would like to general at least as good as can be represen ted amplify some of the author's remarks: by field measurement and in many cases (1) Mapping from air photographs is still much better for scales up to 1/500 or 40' to dependent (and is likely to remain so for a one inch. very long time to come) on the control frame­ It might be mentioned that there have been work supplied by the field surveyor on the several occasions when the plotting instru­ ground. ments have detected errors in ground control (2) For mapping at scales of say 1/5,000 for, within fairly close limits, the instruments and smaller there can be very few cases in­ will not accept faulty data and soon locate deed where modern photogrammetric meth­ any anomalies. The photogrammetric proc­ ods have not entirely supplanted those of the ess is, therefore, a check in itself carried ou t conventional "topographer." One exception in the terms of the scale specified. to this rule is the use of ground levelling Perhaps the main difference between rather than photogrammetric contouring for ground and air survey techniques-and one areas of "flat" ground such as in an irrigation

that on occasion has led to misunderstanding 2 In a letter of June 15, Lord Pentland of Hunt­ -is that the latter is normally designed right ing Technical Services, Ltd. stated that a ground from the start to fit a given specification. The survey organization had complained that Mr. Dawe's paper was unfair to ground survey tech­ aerial surveyor must take into account the niques and said that they felt very strongly about purpose for which the map is required and the this. Lord Pentland stated that it was not Mr. accuracy specified, so arranging each stage of Dawe's intention to make any direct comparison the survey-the flight altitude, type of cam­ with ground techniques. However in view of the complaint it was requested that the herein-given era and plotting process-to ensure the best explanation follow the reprint of the original paper. possible results. PHOTOGRAMMETRIC ENGINEERING is very pleased Planned for presentation at a specific scale, to take the suggested action-ED. HOW ACCURATE lS AERlAL SURVEY? 761 survey where the amount of levelling required areas where the cost of is to control the photogrammetric survey is al­ disproportionately high to the size of the job. most equal to that required to provide a grid The undisputed field of the ground surveyor of spot levels dense enough to permit the in­ must lie however in practically all work which terpolation of contours by eye. involves setting out, including the running of (3) Again ground survey is essential for lines of precise engineering levels. small-scale mapping of dense areas of forest (6) As regards the reference to t inch ac­ or bush where trails, drainage, and even small curacy becoming "meaningless" this does villages may often be screened entirely from not, of course, refer to the many cases where the aerial view and can only be supplied by levels are referred to precise ground marks, running a traverse on the ground. but to level information read off a plan and (4) As the scale of the plan becomes large of which the accuracy is dependent on the so the work of the ground surveyor becomes scale of the plan. For instance, the nearest increasingly important. At very large scales, one can re-establish a given point on the say in excess of 30 ft. to the inch, it is usually ground from measurements taken on a advisable to do the whole survey on the 1/2,500 plan is ±3-4 feet. Within this circle ground, for the sake of both accuracy and of location on uneven or sloping ground it is economy. possible for the level to vary by an amount (5) Ground survey is also preferred for the far in excess of the precision with which it is preparation of large-scale plans of very small surveyed and its value marked on the plan.

Considerations tor the Design of a Projection Plotter*

J. A. EDEN, Directorate of Overseas Surveys, Kingston Bye-Pass, Tolworth, Surbiton, Surrey, England

HE advantage of the projection plotter with regard to model clarity and definition, T compared with the more complicated and obviates the necessity for spectacles. binocular instruments is its extreme simplic- (2) The instrument should he nominally ity. distortion free and utilise full size 9/fX9/f dia- Having examined a number of projection positives. Optimum projection should be be- plotters of different makes, I am convinced tween X 5 and X 6. The working table should that a cheap projection plotter could be built be large enough to accommonate the whole of for plotting individual models with better a X 6 projection. Allowing margins at front performance and facilities than any existing and back to accommodate the tracing table projection plotters and with a working ac- supports when working close to the model curacy as good, if not better than any existing edge, the table width would need to be 62 first-order plotter, except of course, the in- inches. This is not too large to be convenient- strument would not measure machine coordi- Iy worked from both sines. nates or be convenient for air triangulation. (3) The projectors should be mounted so The design should feature the following that the diapositives are held in an approxi- points: mately vertical plane, and prisms should be (1) Rotating shutter viewing should be provided in front of each lens to reflect the employed. Compared with anaglyph viewing projections down vertically onto the horizon- this system presents enormous advantages tal working table. The advantage of this * This paper was included in The Photogrammetric Record (Vol. llI. o. 13. April 1959) published by the Photogrammetric Society. London, England. Permission to reprint in PHOTOGRAMMETRIC ENGI­ NEERING was requested provided credit be given to the JOURNAL, the Society and the author. This per­ mission was given. This publication extends thanks and gives credit as above stated-ED.