v Conformal Projections Based on Jacobian Elliptic Functions 42. Conformel projection of a circle within a square The transformation sdv2w = N2s. For the case n = 4, Schwarz's integral (3.1) cen be identified with Jacobian elliptic functions for modulus 1/W2; that is, it is expressed by sa(N2w, 1/N2) =N2e (42.1) by which the interior of the unit circle = is conformally repres- ented by the interior of the square w. The origir. is at the centre of the square and the axes are the diegonals. The side of the square is K = 1854075, and the semidiagonal is K/N2 = 1+311029. Conformal projections of the sphere or parts of the sphere within squares can be essily derived by this transformeticn. Adams used the mathematically ingenious but labcrious method, first described by Guyou, involving elliptic coordinates on the sphere and the use of tables of elliptic integrals. The elliptic coordinates do not form an isometric system, although isometric coordinates can be derived from them. Hence, Adams's reference to elliptic isometric coordinates is misleading. Computation of coordinates from closed formulae, From (42.1) in the equivalent form, en(K-N2w) = a@+iy, (42.2) we get sn°(K-W2w) = 1- (x? -y?) -tzay, } (42.3) an*(x -W2w) = $[14(e?-y?) + i2ay]. ite If we let wey sa, wtmy? =B, (42.4) then on(K-N2w) en(K -N2i5) = A, sn%(x ~N2w) sn? (x -V2i3) = 1- 28407, 1 ‘i (42.5) an(K -2w) an?(K -V2m5) = 3(1+ 28 +47), ~$en2(K-2w) en? (K-V2%) = $(4+ 2B-A2). Then en Wau = - en 2(K-N2u) = - en[(K-N2w) +(K-V2i)] Wasa ane 2 BNO B? (42.6) 1+ B- Ae 22u 7 ieee enli(koWow) = (kwon) 1+ 2B-A? 7 bd : (42.7) 2a l(1 4A?) ? = 4B? Tebles for firding u from cn 2V2u for modular angle 45°, based on the Smithsonian Buliptic Functions Tables, are given in Tables 10 to 14. On the circumference of the unit circle, o?+y? = 1, and if we let w+ty = cos@ +ising, the above expressions reduce to -cn2Vau = en2N2v = tan(45°+ 6), (42.8)68 taking whichever sign will give e tengent in the range -1 to +1. These results show also that tutus KNN2, (42.9) which are the equations of the four sides of the square. Computation of coordinates by series. A series for win powers of z is easily derived from Schwarz's integral (3.1) for n= 4, w= = 94 43:3 4 423 +, (42.40) 2.4.6 13 or we ze 40+ 007 223 #79 40-4 a +0005 951 533 40-041 66729 #0+005 013 #37 +0024 036 313 +0004 297 841 40-016 085 #17 +0003 738245 40°011 719 322 eee, +0+ 009 023 a?5 (42.11) which gives €-figure accuracy for |s| 0-82. When the computation is done by desk calculator, the closed formulse wil] be simpler than the use of meny terms of this series, and it is recommended only for [s| } 0-5. By reversing this series, or from s = snV2w/W2dnV2u, we get i Ht 4 244 ee ot gy 2. 7 10” * 420 75600” * 35 36000 ” ae or gs w +361 61 (w?9 x 1077) -1+0 (wSx 1071) 3C6.0 (m3 x 1078) +0° 833 333 (w9 x 1072) 0-705 128 (w)3 x 1073) +C+596 715 (wi? x 40-4) 0-504 965 (w2? x 10-5) + +0°427 32 (w25 x 10-6) (42.13) +2590 (w37 x 10-9) nel aai(wsnOsOn 185 (w4S x 10711 ) which is adequate for 6-figure accuracy over the whole area of the square. 43. Projection Guyow 7887, “Adams 7 f a hemisphere wit! a square (Peirce 1877 cog FoTmULG In terme of Jacobian functions. ‘Tre conformal pro- jection of a hemisphere within a square is given by sa(V2w, 12) = Wer, (43.1) where r is the stereogrephic projection with m = } at the origin. Computation of coordinates is easily done by the closed formulae (42.6) and (42.7) or the series (42.11) with r substituted for s.Sod SO PRY Fig. 28. Projection of a hemisphere within a square, pole at the centre (sphere within a larger square). Conformel projections of a hemisphere within a square were devised by Peirce 1877, Guyou 1887, and Adams 1925, none of whom used the definition (43.1). It seems not to have been generelly recognized that their projections are merely different aspects of the same projection, and all of them can be derived in the same way from different aspects of the stereographic, Scale. By differentiation of (43.1) we get @ _ an?W2w fl 2.5 -—_ 43.2 ar" cavao *V(1-r4) (43.2) so that the scale of the projection is given by m= B14 p? 4 q?)/W(1-r4) J. (43.3) At the origin, m = 3; at the midpoint of a side, m = 1/N2. There are singular points at the four corners of the square. Projection of a hemisphere within a square with the pole at the centre of the square. If we start from the stereographic pro- jection with the origin at the pole (16.3), we obtain the projec- tion of Peirce 1877, with the pole at the centre of the square. Peirce arranged the second hemisphere in four pieces, so that the Whole sphere was mapped within a larger square, as shown in Fig. 26. This five-piece or quincuncial arrangement led to the description "quincuncial" being inappropriately applied to the projection it- self, although other arrangements are possible and have been used.Fig. 29. Projection of a hemisphere within a square, poles at opposite vertices. Projection of a hemisphere within a square with the poles at opposite vertices. If we start from the stereographic projection with the origin on the equator and the poles on the real axis (15.2), we obtain the projection of Adams 1925, with the poles at opposite vertices of the square, as shown in Fig. 29. Projection of a henisphere within a equare with the poles at the midpcinte of opposite sides. This is the projection of Guyou 187, and is slso obteined from a stereogrephic projection with the origin on the equator but with the poles equidistent from the axes. It can therefore be computed by letting the stereographic coordinates (15.2) be turred through -45°, that is, they are multiplied by (1~+%)/¥2. When the coordinstes within a square have been computed, it is then convenient to turn them through +45°, or to multiply them by (1+%)/N2. The effect of both rota- tions is merely to change the signs of alternate terms in series (42.11). That is, usirg the stereograrhic coordinates (15.2) Without change, the Guyou projection is given by w= Pr - 0-175 + 0-041 667 r9- 0-024 038 r13 + +++ etc., (43.4) Where the axes of coordinates are now parallel to the sides of the square. The inverse series (43.13) similerly undergoes a chenge of signs of alternate terms, so that all the coefficients are positive. The vertices of the square are at latitudes +45°. The projection of Guyou, shown in Fig. 30, has often been described as the doubly-periodic projection, although it is only one example of a large number of doubly-periodic projections.* “Doubly-pericdic projections discussed in this study are the projections of the sphere or of a hemisphere within an equilateral triangle, a rhoubus, ® square, or a rectangle, and the projection of the sphere upon a regular tetrahedron,Pig. 30. Projection of a hemisphere within a square, poles at the midpoints of opposite sides (sphere within a rectangle). That it is, the transverse aspect of the Peirce projection may not be immediately apparent because Guyou placed two hemispheres side by side‘to map the sphere within a rectangle while Peirce divided the second hemisphere into four pieces so placed as to map the sphere withir a larger square. 44. Projections of the sphere within a square (Adams 1929 1336). “To map the sphere within a square, we use the Lagrange projection instead of the stereogrephic as the parent projection in equation (42.1). It should be noted, however, that the Lagrange projection, unlike the stereographic, has two sirgular points on the boundirg circle, and the character of the projection is changed as these singular points are placed in different positions on the Doundary of the square. In the first three examples below, the singular points of the Lagrenge projection are mapped at opposite vertices of the square; the only singular points are those at the four vertices, and these three projections are different aspects of the same projection. In the fourth example, the singular points of the Lagrange projection are mapped at the midpoints of opposite sides of the square; the resulting projection has six singular points, and is not the same projection as that in the first three examples. The singular points of the Lagrange projection could be placed in other positions on opposite sides of the square, result- ing in further projections with different properties. Projection of the phere within a square, with the poles at opposite vertices. In this case, we use the direct Lagrange pro- jection as the parent projection, so that the projection is defined by sa (N2w, 12) = 2 tanh gC. (44.1) Cocrdinates of the Lagrenge projection are computed from (17.2), and computation of the coordinates of the projection within a Square then follows the procedure of Sec. 42. This projection is illustrated in Fig. 31. By differentiation of (44.1) we have72 Fig. 31. Projection of the sphere within a square, poles at opposite vertices. enV2u i ey 3 on’W2w ee ay =e BAe = : 44.2 Gnivdy = F (1 tanh a0 a = ¥ Tae, we (44.2) whence aw/aC = Bonvaw, (44.3) from, which the scale can be found. The scale at the origin is meh, a Projection of the aphere within a square, with the potes gumnetrically placed on a diagonal. If the coordinates of the transverse Lagrange projection (20.5) are used in (44.1) instead of those of the direct Lagrenge, we get the projection illustrated ip Fis. 32) with the poles symmetrically placed on a diagonal of the Square. This is the projection of Adams 1929, obtained by him by a much more complicated method of computation. Projection of the sphere within a square, with a pole at the contre. By starting with the coordinates of a Lagrange projection73 Fig, 32. Projection of the sphere within a square, poles on a diagonal. i [ ! Fig. 33. Projection of the sphere within a square, pole at the centre.74 us Fig. 34. Projection of the sphere within a square, poles at the migpoints of opposite sides. with a pole at the centre (21.4), we can derive a projection of the sphere within a square, with one pole at the centre of the agers pid the other pole represented by two points at opposite ends of a diagonal. This projecticn is illustrated in Fig. 33. Projection of the ephere within a square, with the poles at the midpointe of opposite sides. It the direct Lagrange projection Je rotated through 45°, i.e. if the coordinates (17.2) are maltin plied by (1-%)/N2, before the Projection within the square is midpoints of opposite Sides of the Square. We can use the same Geytge as thet adopted for the Guyou projection (43.4), changing the signs of alternate terms in the serios (42.11), and getting cooréinates referred to axes parsllel to the sides of the square. This projection is illustrated in Fig. 34. It was derived by Adams 1936 by a much more complicated method of computation. The Worth pused for wall decoration in an airport building at Fort Worth, Texas.75 45. Projection of the sphere upon a cube (Lee 1970). Derivation of formula. The one-sixth part of the surface of the sphere which is represented by one square face of the cube is a spherical quadrilateral whose disgongls divige it into four spherical triangles, each with angles 47, in, $7. The stereo- graphic projection of each such spheriéei triehgle is conformally represented upon the infinite half-plane, with origin at the vertex with angle $7, by the function € in (23.9), that is, _ tev 308 (4 + 4)? (1 = 2N3n2- n4)3 i £ Hy aigre pays? 18" Cyewjrts nays (45-4) Representing the infinite half-plane upon the querter-plene, and then representing the whole plane within the unit circle by (31.1), we can next represent the circle conformally within a square by sd(vew, 12) = N2[1-N(1-E)]NE, (45.2) the axes of coordinates being the diagonals of the square. By Squaring both sides and componendo and dividendo, we get on V2w =N(1-€), (45.3) 80 that one square face of the conformal projection of the sphere upon a cube is given by _ (l= 2N3r? - 4 ie a oway - Ce er) (45.4) As on(-u) = cnu, there is an ambiguity of sign in this definition, but it is easily settled in practice since the origins and quad- rants can be made to correspond. Computation of coordinates from closed formulae. If we let on 2m = o+ty, then sn’¥2w = 1-(a?-y?)-i2zy, (45.5) anWew = $[1+ (x? -y*) + t2ay]. Also let wry =A, x?ay? =B. (45.6) Then 2a -N( 4a?) ?~ 4B? cn W2u = en (New +25) = awe, : : (45.7) one 4 1+ 2B-A on aWav = = . on (N2w aha (14h) a8 On the boundary of the spherical quadrilateral, 2* = y?, and -on 2N2u = on 2N2v = (1-2x?)/(1+4 2x7), (45.8) so that tutv = KN2, (45.9) which are the equations of the four sides of the square.Pig. 35. Projection of the sphere upon a cube, two faces with @ pole at the centre, four faces with centre on the equator.7 Computation of coordinates by series. To express the coordin- ates by a series in powers of r, it is simpler to return to (45.2) and obtain NIZE (14 gr? ~ 49/2 ~ (4 = avgr? - 4) 9/2 : NE 2.3784 rt) = 3/4 r(4- ort — tort? . 5ar%6 — 4a6r20--.-), (45.10) Note that there is a zero term in r®. From Schwarz's integral (3.1) we have ace wee toges (45.114) Substituting the value of s and its powers from (45.10) in (45.11), and also substituting Rt = 35rt, or B = 2279507 r, (45.12) we obtain z 9, 5075213 = = ceed ale 45.4 een B+ 7 sacba” * (45.13) or w= Eg +0° 000 053 R25 +0°025 926 R° +0+000 021 R29 40+ 004 630R9 +0- 000 009 R33 +0-001 240R}3 +0000 004 R37 +0+000 397K}? +0-000 0024! +0°000 141 R22 totes (45.44) This series is suitable for use within the region where |R| > 1, which is the greater part of the area of the square face. Both this series and its inverse series are too slowly convergent to be practically useful in the region near a vertex. Scate. From (45.3) we get sn2W2n = 1-N1-&, an2N2p = $(1+47-€), 45.15, sn? V2 dn2V20 = 3€, (45.15) and thus we get den V2w/aw = - V2sn V2v an Van = - NE 2n3h3n(4 4 r4) eae © Cre nar? = 14) 372” a" We also have @ (1 ~ N32 - p4y 3/4 6N3r(44 r4) a ar\ty avg ort) * 7” ay adapt et) AT = ge ( ) From these we getFig. 36. Projection of the sphere upon a cube, two faces with the equator a8 a diagonal, four faces with a pole at the midpoint of an edge.Fig. 37. Projection of the sphere upon a cube, every face with a pole at a vertex. aw 33/4 ar * (ate ey iy and therefore the scale of the projection is given by m= $3414 p? 4g?) [C1 tard a 84 (45-19) the scale at the centre of a square face is 4.394 = 4-139 754; at the midpoint of an edge it is W3/l2 = 1:224745. At the mid- point of an edge, the ,jgerggge in Scale over the scale at the origin is therefore 2°/73 7/4, or 745708.80 Nature of the projection. Three examples of the projection are shown in Figs. 35 to 37. In Fig. 35, two faces have a pole at the centre and four faces have a centre on the equator. On the common boundery of two dis— similar faces, tang = cos, where A is reckoned from the centrel meridian of the equatorial fece. ‘The vertices are therefore given by tang = +1/N2, or ¢ = 235° 15’ 51-803. The poler face is derived from the polar stereogrephic (16.3); the equatorial face is derived from the transverse stereographic 13 2} rotated through 45°, thet 1s, with coordinates multiplied by (1~ i)/w2. In Fig. 36, two faces have the equator as a diagonal and four faces have a pole at the midpoint of an edge. On the common bound- ary between two dissimilar faces, ten¢ = N2sinh~- cosh, where X is reckoned from the centrel meridian of the face conteining the pole. The vertices are therefore on the equator et longitudes given by tand = +1/N2, or A = +35° 15’ 51"-803, and on the meridian \ = +}7 at latitudes given by tang = +N2, or ¢ 254° 44° 08"-197. The graticules of Figs. 35 and 36 are also shown in assemblies in which some faces have been cut along the diagonals. Fig. 37 shows an example in which all six faces have a similar graticule, with a pole at a vertex. Each face is derived from the oblique stereogrsphic (14.3) with origin given by tang +t /N2, oF go = +35° 15’ 51*-803. Two edges represent the merigians A = 460°, and on the other two edges, 2N2 tang = V3sin\ — cos r. If one vertex is the north pole, the opposite vertex is on the centrel meridian at latitude given by sing = - 2, or ¢ = -19° 28' 16"-394; the other two vertices are at latitudes +19° 28' 167-394. ‘The equator passes through the midpoints of adjacent edges at longitudes #30°. In. Pig. 35, the equatorial face is obviously the transverse aspect of the polar face. If we imagine the sphere rolled through a right angle about an axis joining the centres of two opposite equatorial faces, so that a meridian takes up the position formerly occupied by the equator, the effect on the projection is merely to trensfer the greticule from one face to another or to rotate the greticule on one face. If the six faces together are considered aS one projection of the whole sphere, the graticule of the trans- verse aspect is identical with that of the direct aspect. The same remarks apply to Fig. 36, where the sphere can be rolled through a right angle so that the pole at the midpoint of an edge is trans- ferred to the midpoint of the opposite edge, and the diagonal representing the equator in one face becomes the other diagonal of the same face. The graticule of Fig. 37 does not have this property, for the sphere cannot be rolled through a right angle so as to transfer the pole from one vertex to another. It can, however, be rolled through & right angle so as to transfer the pole from a vertex to the mid- point of an edge, and there is a meridian in the graticule of Fig. 36, although not one of the meridians actuslly drawn, which follows the same course as the equator in Fig. 37. 46. Conformal representstions of a circle within a rectangle. The conformal representation of the interior of a circle by the interior of a rectangle can also be effected by means of Jacobian elliptic functions, the dimensions of the rectangle being depend ent on the choice of the modulus *. The representation necessar— ily involves singular points on the boundary, but these may be jocated in various positions. We shall consider two cases.81 Singular pointe at midpoints of opposite eidee. In the representation of the unit circle a given by Nkise ww = 8, (46.1) separation of the real and the imaginary parts gives ¥ictsn fun $ucn' gu an'gv 1~- sn? duan'* $v a (46.2) Nktdngusn' gu _ ‘ 4- en? $uan'? $v which show that = = 0 when u = 0, and y = O when v = 0, so that origins and axes correspond. We also have (4 ~ en? gu en!? $v) 4- sn? guan'? $v (46.3) wsyrs On the circumference of the unit circle, z*+y* = 1, and equation (46.3) is then satisfied by u = *K. When v = £2K', (46.2) gives z=0, y = tvetna gu. (46.4) 0 At w= 0, v = #2K', we therefore have = 0, Uy = The transformation (46.1) therefore represents the unit circle within a rectangle of height 2K and width 4K’. The circum- ference of the circle is represented by u = 2K, the upper and lower sides of the rectangle, and the imaginary axis of the circle is represented by u = 0 and also by v = #2K', the left and right sides of the rectangle. At u = 0, v = +k’, there are singular points where the imaginary axis of the circle is turned through a right angle, and the upper half of the boundary of the rectangle represents the same line as the lower half. Singular points at the vertices. Another conformal repres— entation of the unit circle # within a rectangle is given by Mit sc $(K- wv) (46.5) or, in the inverse form, _ tewetso$(K -w) (46.6) 44 NKTscd(K -w) If we use the abbreviations, s = sn¢(K-u), 8’ = sn’gv, etc., Separation of the real and the imaginary parts gives 1-%a'? ~K'(1-070'? ) © = 7Tyigi? skr(1~o%e"*) + aH e00'a"” B2d'2 +k'(1-0 ) (46.7) avktas' : u Teta'? aki(1-0%0'2) + 2NkIsco'd! When u = 0, then x = 0, and when v = 0, then y = 0, 80 that origins and axes correspond.82 When u = 4K, (46.7) become 1-krat® avkte! . Se 46.8 @ iT Rig? U*T oR a and when v = #2K', they become a?-kt Nita az aK’ (46.9) and in both cases, z*+y*= 4. The circumference of the circle is therefore represented by the boundaries of the rectangle, u = +K, and v = +2K', and there are singular points at the vertices, where a ick ane cae : a T+ 14k! 47. Projection of the sphere within a rectangle with a merid- ian as. boundery (Adems 1925) Definition of the projection. A conformal projection of the sphere within a rectangle, with a meridian as the boundary, and with axes parallel to the sides of the rectangle, is defined by Nk'80$(K-w) = exp(-4¢), (47.1) which is the transformation (46.5) with the substitution ae tanh }¢. Adams defined it differently, but his formula may be transformed into this form by changing the origin and axes and by doubling the dimensions. By separeting the real and the imaginary parts, we get (46.10) Vke'sn $(K ~u) en $(K - u) en'iv dn'$v tenia ca vance F exp (-34) cos gr, 47. Mkt an} (K -u) antgo Ce T= sn $(K—u) an’? gy ~ O*P (~d#) sin gn, and thence, by squaring and adding in the first case and by div- ision in the second, we get ie _ RiLt= on? 4(K-u) on! dv] exp (-$y) =tande = toon?i(h ou) ane? (47.3) 3(K- "4 tan $d = — dn $(K ~ u) entdo : sn §(K-u) en}(K —u) en’go an’ do These formulae are amenable to some further simplification, using some of the many formulae relating to Jacobian functions. Thus tengg - 12 tenge 1+tange ~ dosn?$(K~u) an’? gv- td ~ on? $(K~u) on’? $0] 1-sn?4(K-~u) an’? soy k'[1~ en? 4(K-u) en! 2g]83 Fig. 38. Projection of the sphere within e rectangle ‘bounded by @ meridian, for k = sin 45°. _ tak! an? $(K -u) - kt 4-k'sn'*$v Toh? an’g(K-u) +k! 1+k'ant?h0 > /feee. eee (47.4) lisanu 1-k'sn (K : ten $A = = 4-sn‘(K' - v) 4-an(K-u) 1+n!(K' -v). i @nu+k? 1-sn'(K' -v) - JE 1 +n’ (K? ~»). (47-5) t-tan?$¢ __anu- k!sn'(K' -v) "Ty tan?4¢ q-kanusn'(K'-v)” and thence 47-6 cosh 4-tan24\ anusn'(K’-v)-k! ote) op h = ee Se 44¢tan?$\ dnu- Kian! (Kt - v) Nature of the projection. The expressions (47.6) are the formulae for inverse computation of latitude and longitude from projection coordinates, and enable the nature of the projection to be studied. When u = 0, then ¢ equator. When v = 0, then = 0, 80 that the v-axis represents the 420, so that the u-axis represents the zero meridian. When u = 4K or when V = +2K', then A = tW. The sphere is therefore mapped within a rectangle of height 2K and width 4K’, the boundary of the rectangle being the meridian A= 4m, The poles are located at the midpoints of the upper and lower sides of the rectangle. At the vertices of the rectangle, tango =k.84 By differentiation of (47.1) we get dw/at = snd(K-w) sn}(K+w) = cnw/(dnw+k'). (47.7) From this we find that the scele at the origin is 1/(1+k'). We also find that |d»/a¢| is zero at the poles and infinite at the vertices of the rectangle, so that these are singular points of the projection. For a rectangle in which the length of the equator is twice the length of the central meridian, we take K =K', or k =k’ = sin 45°. This projection is illustreted in Fig. 38. Equation (47.1) can be written in the equivalent form, 14580 $wdn$w —_— = aC. 8 Gnu- hacky — P20 lees For k = 1, k' = 0, the elliptic functions become hyperbolic func- tions, and this expression becomes cosh}w+sinh}w = exp$¢, (47.9) or w= ¢, which defines the Mercator projection. We now have K' = 3a and K = @, so that the rectangle is of width 2m and of infinite height. Computation of coordinates. Adams described a method of computing the coordinates of points in the interior of his projec- tion within a rectangle by introducing a number of auxiliary quentities. A much simpler approach is possible by the use of equations (47.6) which, with obvious abbreviations, can be written d-k's' 4 eos $ = (47-10) whence cos $ cosh = ———, (47.44) From (47.11) we get k" + cosg cos r gt = ~~ SObE COB La, 7.42 8" Ty i cos geosn 4 (47.42) and from (47.12) in conjunction with either of equations (47.10), wcos¢ 1 +k" cos ¢ cos A ad-kle! = =8. (47.13) Solution of equations (47.12) and (47.13) finally gives dnu = $[V(B7 441A) +B], = -14) kien'(K'~v) = 3[V(B? 4 4%'A) -B]. eee On the equator, for u = 0, # = 0, the second of equations (47.2) gives Aah! -N (14h? — 2K te ( cosh) 2h'singr (47.15)85 On the central meridian, se3(K-u) = Vtango/Vk', (47.16) and on the bounderies, week, sntiv = Vtan$e/vk', (47.17) v = OK, an}(K-u) = kI/Ntan go. (47.18) 48. Projection of the sphere within a rectangle with singular ‘points on the equator (Lee 1965): Definition of the projection. A conformal projection of the sphere within a rectangle, without singular points at the vertices, is defined by Mist scdw = tanh FC. (48.1) which is transformation (46.1), # denoting the Lagrange projection. Separating the real and the imaginary parts, we get Nk'snjuecnfucn'$vdn'gv _ sinh $y ns 4 -sn-gudn'* gv * ‘Gosh $$ + cos $A . (48.2) kan gu sn'gv singh Seen eee erg eee T-sn?fudn'* tv ~ cosh4y+cos$nr whence we get k!(4 - on?guen'?$v) cosh $y ~ cos $h ene ee oo (48.3) ge +n? = Set anid cosh $y + cos th From this we also get 29 tan$¢ = tamhdy = . We goth? ' evk'sn gu cn suentgu an'gy mex (48-4) q-sn?juan'?gv+ k'(4-cn*gucn! $v) 2h Ceaay endh = as 2aktan gusn'bv (48.5) 7 —en® gu dn’? yo- k’(1- en? gucn'® gv) After some further manipulation, these become sn $uecn$u en'$vdan'gv Fer Ta hlen! oh Qn?duek? 1+hian! $v : radia 44sa'(K'-v) (48.6) Qnusk’ 1-kfsn'(K'-v) tandg = evki(1+k")*Fig. 39. Projection of the sphere within a rectangle, with singular points at the midpoints of the sides, for k = sin 45°. an tu sn'}u . (tak!) pee. sate tanga = 2Nk'(14+h') in? ju Then’? fo Then, from cos ¢ = eee, cosh = tee, (48.8) we derive ee (1-k')[anu eee ee : goo -v)) 48.9) (-k')Lanu + k'en'(K! - v)] - 261[1- anusn'(K'- v)] cos d= (4+%')[dnu- k'an'(K'-v)] which are the formulee for inverse computation of latitude and longitude from projection coordinates. Nature of the projection. From the original definition (48.1) we see that when u = 0, then ¢ = 0, and when py = 0, then A = 0, so that the v-axis represents the equator and the w-axis represents the zero meridian. From (48.9), when u = 4K, then \ = #0, and when v = +2K', then $= 0. The equator is therefore represented by the three straight lines, u = 0 and v = +2K', and at their intersections we have tand\= +Vk!. The sphere is mapped within a rectangle of height 2X and length 4K', with the poles located at the midpoints of the upper and lower sides, which represent the meridian \ = +7. The equator is represented by the v-axis and by the left and right sides of the rectangle.87 By differentiation of (48.1) we get ao _ an f~ an (K- gv) ac) OONK(A-kT) Prom this we find that |do/aC| is zero at u = +X, v = 0, and infinite at u=0,v +2K', 30 that the singular points are the midpoints of the four sides of the rectangle. The scale at the origin is 1/(2%K'). This projection is illustrated in Fig. 39 for the case k =k! = sin 45°. (48.10) Computation of coordinates. Equations (48.9) can be written, with obvious abbreviations, as (4K )(a- kis") 00s § = oe’ ToK Tet)) Oe’ (1 ode? (i-k')(a+k'8') + 2k" (1-de') (48.11) (1-k!)(a + ha") ~ 2k"(4-a8') oo (isk (a= kis") (1-k')(a+h's') - 2k'(1 -de') oe del) 8.12 whence cos Co8N = ar) y re’) + oe/(1-a8")” ee) and thence tecosdoosd (1-k')(a+h!8') = sawicce a eae T= cos ¢ cos 2k (1-de') os cos ¢ . (144h')(a-kla') To cosé cosh Ak?(1-Gs") © From these equations, we get eile eG RL) eom eres (48.44) ad+k's (1+k")(1+4 cos ¢ cos A) whence a _ (re mry(t4 cos gp cose) + 24-81) 8056 (48.45) rat ~ (dak! )(44 coag cosh) - 2(1-h') cos g equations (48.13) now Subst: i on in either of stitution of this expressi Seen produces quadratic equations in @ and h'e!- A = k2(1- cos g cosh), B= 2(4 =!) cos G- (14K')(1 + cos g c08 X)y (48.16) = 2(4- Kt) cos g+(44K')(1+ COB g cO8 Nr), the equations are a a } (48.17) OKre!? +Aklal +h to which the solutions are88 dnu k'sn!(K'-v) " [a - (a2 ~ 4%130)]/28, } ~ [A= (A? = e130) ]/20. (48.48) Simplified formlae for the coordinates on the axes and on the boundaries are u=0, sn'iy = tan gk", =0, segu = tanhzy/NE', (48.19) v= 2K', angu = vet/tanin. When u = K and \ = m, the second of equations (48.2) gives ry - Ut!) cosh $y -VE(14 8" cosh? by — 4k") sn’ seat ee oy eee aaa 49. Projections of a hemis, here within a rectangle. For a projection of thes Safe Gifiensions as those iver sees for the sphere within a rectangle, but covering only a hemisphere, we qrite 2¢ instead of ¢. The formulae already given will apply to the new projections with the substitution of \ for $A and of cos’4/(1"+ sin’ 4) for cos ¢. » (48.20) These projections are not considered of sufficient interest to be illustrated, but two examples are briefly treated below, Projection of a heminphere within a rectangle bounded by a meridian (Lee 1965). The projection of the sphere within a rect- angle defined by (47.1) becomes Vxtse$(K-w) = exp(-c¢) (49.4) when applied to a hemisphere. The v-axis represents the equetor, the u-axis represents the zero meridian, and the boundaries of the rectangle, u = and v = 2K’, represent the meridien \ = dm. For case K = 2K', op k! = (V2-1)/(W24+1), the hemisphere is Be ee within @ square, and the projection is identical with that of Guyou, described in See. 43 above and shown in Fig. 29, For k = 0, k' = 1, equation (49.1) becomes tand(da-w) = exp (-¢), or w = lente, (49.2) which defines the transverse Mercator projection. Projection of a henierhere within a rectangle bounded by the equator (Lee 1965). The projection of the sphere within a rect- angle defined by (48.1) becomes Nk'schw = tanh 3¢ (49.3) When applied to a hemisphere. The equator is represented by the dines, u = 0 and v = 2k’, and the meridian » ~ 3m is represented by the line w= XK. The equator is also represented by the line Yen woks 480 thet if we extend the projection to thie line, that is, if we add another rectangle, the hemisphere ic represented by a@ rectangle bounded by the equator, with the pole at the centre. if we take K = 2k’, the rectangle becomes a square, and the projection is that of Peirce (Fig. 58), rererres te a new origin and new axes. For k = 0, k’ = 1, equation (49.3) becomes tanjw = tanh $¢, which defines the transverse Mercator projection.89 50. Conformal projection of a rectangle within an ellipse. ‘A conformal represéntation of an ellipse within a circle was des- eribed by Schwarz 1869c, the formula implying a rectangle as an intermediate transformation. That is, a rectangle w, of height 2 gad width 4K', can be conformally represented within an ellipse # y a = sinh (mo/4K'). (50.1) By separating the real and the imaginary parts, we get inn cos — na gin (50.2) ie ae = cosh sin, : . 4K" aa 4K’ : : aK * 4K" from which . os =1, (50.3) sinh @(ru/4k") * cosh2(1u/AK" ) s0 that any ordinste of abscissa u is represented by an ellipse of which the semiaxes are sinh (7u/4X') and cosh (7u/4K'), and the foci are « = 0, y= t1. Thus, all the ordinates are represented by confocal ellipses. The boundaries u = +K are represented by an ellipse of semi- axes sinh (nK/4K') and cosh (aK/4K'). The vertices of the rect- angle, u = 4k, v = +2K', are represented by the extremities of the major axis of the ellipse. The extremities of the v-axis of the rectangle are represented by the foci. The left and right sides of the rectangle are represented py those portions of the major axis between the foci and the extremities; these are cuts in the representation, and the areas above and below them are not connected. For the mejor axis of the ellipse to be twice the minor axis, we must have sinh (nK/4K') = 1/N3, whence K/K' = 0-699 398, and we then find k = sin23¢-8958. To agree with the projection of Adams , and to make use of available tables of elliptic functiors, we have instead taken k = sin 25°, for which the semiaxes of the ellipse are 0°5908 and 1-1615. 1. Projection of the sphere within a cut ellipse (Adams 192 Adoms"desoribed a conformal projection of the sphere within an ellipse, which is defined by ae sim( 2, (= 2000 SEO) , (51-1) which is equivalent to the two transformations, Wk'seh(K-w) = exp (-$¢)> a = sinh (w/4K"). (51.2) the first of these is the projection of the sphere within a rect- angle with singular points on the bounding meridian (47-1), the second is a representation of this rectangle within an ellipse. By differentiation of the second of equations (51-2) we get ot cosn, (51.3) qe aK" 4x so that, with (47.7), we get aoa enw mo (51.4) ae _ 7. SE. cosh ac. 4K’ dnw+k' 4K"Fig. 40. Projection of the sphere within an ellipse, with singular points on the bounding meridian. Fig. 41, Graticule at 1° interval in the region near the extremity of the major axis in the projection of Fig. 40. From this we find that |as/ac| is zero at u = +K, v = 0, infinite at u = 2K, v = +2K', and zero at u = 0, v= +2K', so that there are singular points of the projection at the poles, the extremit- ies of the major axis, and the foci. ‘There are slits along the major axis from each extremity to the nearer focus- Adams mentioned only the singular points at the poles and was apparently unaware of the others. His table of coordinates wrongly gives the intersection of the equator and the bounding meridian a8 the extremity of the major axis instead of the focus. The projection, for k = sin 25°, is illustrated in Fig. 40, and the graticule in the region near the extremity of the major axis is shown in Fig. 41, where, of course, the width of the slit 18 exaggerated. 52. Projection of the sphere within an ellipse without cuts enhéim 1950, Lee 7 Another conformal projection of the sphere within an ellipse is defined by w tanh 3¢ 2 eee ol enn 4 # = sinh | 5 se? nae |, (52.1)92 *setod ayy 38 ATuo squtod avrnfure yatm ‘osdeTTe we UTURTH exeYds aN Jo LOTIOSPoOLg “zy “STE92 Which is equivalent to the two transformations, Nk'sc$w = tanh}, # = sinh (w/4K"). (52.2) The first of these is the projection of the sphere within a rect- angle with singular points onthe equator (48.1), and the secona is a representation of this rectangle within an ellipse. The singular points on the equator are cancelled in the second trans- formation, and the only singular points remaining are those at the poles. From (51.3) and (48.10) we now have az _ alan gw ~ an (K ~ 4w)} mo — eee =, 2. ag 8KNKI(1 =k!) CSN er Cea from which we find that |ds/a¢| is zero at u = +X, vp = 0, 80 that the poles are singular points of the projection. There are no singular points at the extremities of thé major axis or at the foci. In the latter case, |ds/ac| takes the form Ox, but it can be shown (by expansion in series, for exemple) that it is finite and non-zero. The scale at the origin is /8KWh', The projection, for k = sin25°, is illustrated in Pig. 42. Equation (52.1) was first used as the definition of a pro- Jection by Gougenheim 1950, who however believed that it was merely another formulation of Adams's projection (51.1). Lee 1965, in ignorance of Gougenheim's work, noticed the additional singurer Rotnts which Adams hed overlooked, and derived equation (52-1) to define a projection without these singular pointe. 53. Transverse Mercator brojections of the entire spheroid. The definition (12.3) can be applied to a projection of the spher- oid, using the spheroidal isometric latitude (53) instead of the spherical isometric latitude (4.4), and the resulting projection has every right to be called a transverse Mercator projection of fhe spheroid. It is not, however, the only projection of the spheroid entitled to this name. The transverse Mercator projection of the sphere possesses a number of properties, not ali or gison can be possessed by any one projection of the spheroid. Instead, there are a number of projections (Gdowski 1964 assorted that there is an infinity of such projections), each corresponding in one way on jonother to the transverse Mercator projection of the sphere, and all becoming identical with that projection when the ees srieity of the spheroid is put equal to zero. Of three such pro- dections so far investigated, one, known as the Gauss-Kriiger pro ection, preserves a constant scale along the central meridian; in the spheroidal projection given by (12035, the scale along the Central meridian increases from the equator towards the pole; in fhe Thompson projection the scale along the central moriaian decreases from the equator towards the pole. The projection of (12.3), like the transverse Mercator projection of! the sphere, represents the point, #= 0, \ =i, at infinity; the other tvo Projections represent the entire spheroid within a finite sree The coordinates of the Geuss-Kriiger form of the transverse Mereator projection of the spheroid have been expressed ae infinite Series in powers of the longitude reckoned from the central meriae ian (or in powers of the complex variable (), and these ane ade. guate for the widths of the belts over which the projection has been used in geodetic survey. If the projection 1a to include the entire spheroid, however, it is necessary to seek formulas in closed form.3 The coordinates are expressed in terms of the isometric lat- itude and longitude by the equation, w+iy = f(¥+tA). When d = 0, then y = 0 and z = 7(). If the scale is to be correct along the length of the initiel meridian, we must therefore have fly) =e, where M is the length of the arc of the meridian from the origin (on the equator) to the point being projected. But M cannot be expressed in terms of y in closed form by means of any known functions. It ia possible, however, to express both (a, y) and (W, A) in closed form in terms of a third set of isometric para~ meters (u, v) by means of Jacobian elliptic functions. This method of attack is due to Professor E. H. Thompson, who established the transformation formulse in 1945 but did not publish them and did not pursue the matter further. Subsequent investigation of the graticule was by Lee 1962. ‘54. The Thompson transverse Mercator projection of the entire spheroid (Thompson et the u-axis correspond to the init- jal meridian or y-axis, and on this axis put sing = sn(u, k)> (54.1) where the modulus k is equal to the eccentricity of the spheroid. It follows that cosg = cnu, 4-Ksin’g = an*u, ag = dnudu, (54.2) and thence, from (9.1), p = ak!?nd@u, v=andu. (54.3) The isometric latitude (9-4) can now be expressed as y = tenn ?(snu) - tanh”? (# snu)- (54.4) To express (#, 0) in terms of (u, v), the same functional relation must hold between the complex variables, so that we have y+th = tanh“?sn (us tv) - ktanh7}[k en (u+iv)]. (54.5) If we use (u/a, v/a) instead of (u, v), (54.5) a8 the inverse definition of a projection which we can call the Thompson projec- tion. When = 0, (4:5) reduces to (54.6) ition of tanh (p+) = sin(u+iv), which, from (4.5), can be expressed as (12.3), the defin: the transverse Mercator projection of the sphere. To separate the real and the imaginary parts of (54-5)> we deal with the two terms separately. ‘Thus, we can let tanh (a+i8) = sn(utiv)s (54.7) from which we derive sinh 2a snudn'v 4 Gosh 2as cos 2p 1-an*usn'*v (54.8) sin 2p e cnudnusniven'D | cosh 2a + cos 28 4- an’usn’*v whence4 gosh 2a~ cos 26 1- cn®ycn!? 24h = = 77 (54.9) ean cosh 20+c0s 29 1-dnuen™ > ‘Thus we derive 2snudn'v Tesnudn'?y’ +snwdn'*p (54.10) 2 cnudnusn'ven'y en*u en’? y~ dn?uen! 2p’ and finally tanha = snu dn'v, tanB = deuse'y. (54.11) Similarly, for the second term of (54.5) we can let tanh (a! +i") = kan(u+tv), (54.142) from which we derive sinh 2a! _ Esnuan'y : tosh 2a’ +cos 26" 41-dn2usn’2p eee (54.43) sin 2p" k onudnusn'v en'v - cosh 2a" + cos 26" 4 = an®wsn’™o aaa whence gr? amie. 208 2a! ~ cos 26" (54.14) cosh 2a’ + cos 28 Thus we derive tamh 20" = a = ee uaa tegen késn?u+ dn'?y (54.15) 2nt 2konudnusn'ven'y tan 2g" = ———7__, : 1g ~nt® © an®uen® y- kentu snp and finally tanha! = kenund'v, tang’ = keduse'y. (54.16) Using the results (54.11) and (54.16) in (54.5), we therefore have @ = tanh (snuan'v) - ® tanh (* snund'v), (54.17) » = tan (deuse'y) - ktan™( cau se'v). Writing (54.5) in the form € = tanh (snw)- % tanh2 (ksnw), (54.18) we get° Fig. 43. Gratioule of the Thompson transverse Mercator projection of the International (Hayford) Spheroid. aC any ene re dw enw ano (54.19) and therefore az enw dnw —r (54.20) ac" By separating the real and the imaginary parts of this equation, we obtain Qu cnu anu dn'y(cn'?v— k?sn2u sn'?v) ay On 2 (4 ~ an’u sn! v)? 54.24 av aus snusn'ven'y (dn2u dn? v +k en2u) ( ) ay aN kre (4 anu sn” v)? , Which is therefore ity in the equator. All four derivatives are zero at u = 0, v = a singular point where there is a discontinu: The longitude at this point is \ = $a(1-k)- The graticule of one quadrant of a hemispheroid on the Thompson projection is shown in Fig. 43, using a 10° interval in latitude ana’ longitude except in the outermost quadrilateral Where a 1° interval is used. The projection of the whole spheroid is shown in Fig. 44. Both figures have been computed using the eccentricity of the International (Hayford) Spheroid. On the initial meridian, u is given by (54.1); the coordinates (u, v) in other cases were computed by a method of successive approximations. Seale and convergence can be obtained from (54.21). The scale is given by (1+ sr@wan' v) 2/2 (antu+ an!?y- 4)ue aus + (54.22) veos ¢ Kt2(4- dn*usn'*v) On the initial meridian, from (54.2) and (54.3), a/vcos # = dew, and (54.22) reduces to are ye ee (54.23) which ranges from 1 at the pole to 1/z’? at the equator. On that part of the equator for which u = 0, an'v/k'? on!*v, (54. 24) m and on the meridian ) = $7, where u = K, m = (a/v cos $) sa'v ca'v. (54.25)“proasyds (progkey) TwuoTywureQUI oy Jo WoT}oePord 10ye0ZeH estonsuET, UOSduONL “by “Ta =e GES \| | | ! ' \ | | | | 97 The Gauss-Kriger transverse Mercator projection of the entire spheroid (Thompson 1945) The ng definitioy suggested by Thomas Wray, is equivalent to the definition used by Thompson. From (54.2) and (54.3), an element of arc of the merid~ dan can be expressed as aM = pad = ak" nd@u. (55.1) The length of a meridional arc is therefore given by rm u = ax [ nd*udu =aedu, (55.2) fo the constant of integration being zero since M = 0 when u = 0. On the initial meridian of the Gauss-Kriiger projection, a =M, and therefore the projection coordinates (x, y) can be expressed in terms of the coordinates (u, v) of the Tnompson projection (54.5) as aed(u+iv), (55-3) whence, by separating the real and the imaginary parts, we obtain the projection coordinates as k@snucnudnu z Se enu - ne" a dn?u + dn’ @y- 1 2 (55-4) y k?? sn'v en'vdn'v Sear ae OU cea rea a an*u+dn'*v Por large values of v, the divisor in the last terms of (55.4) ‘ecomes small, and extra decimals must be used to give adequate accuracy in the quotient. The graticule of the Gauss-Kruger projection, for one quad- rant of a hemispheroid, as computed from (54.17) and (55.4), is shown in Fig. 45. The x-axis represents the initial meridian > = 0, and on this meridian M. (55-5) @ =aedu The line « = aE represents the meridian \ = 7, and on this meridian (where u = K) y = a(v-en'v+se'v an'p). (55-6) The positive y-axis represents part of the equator, extending from X= 0 to \=4n(1-k), the latter value being at y = a(K'-5'). At this point, the equator changes smoothly from a straight line to a curve. On the y-axis, we have y = a(v-ed'v)- (55-7) © spheroid is shown in Pig, 46, again using the eccentricity of the International (Hayford) Spher- oid. It can be seen that the entire spheroid is represented within a finite area without singular points. Writing (55.3) in the form az 2aedw, (55-8) The projection of the entir345678510 Fig. 45. Graticule and scale in the Gauss-Kruger transverse Mercator projection of the International (Hayford) Spheroid. we have dsz/dw = ak'?@nd@w, (55.9) so that, with (54.20), we get as/a¢ = a caw (55.10) Separation of the real and the imaginary parts gives oe _ dy _acnudnuan'v ey 8A odn*usran'?y af (55-11) ay _ ae ak snusn'ven'v =e. ay aN nu + dnt?y 4 The convergence is now given by kisnusn’ven'y dn (Kk ~u) ~t = = -sn! "(Kt-v), 212 any = chu dnwan'y Be Clu) enn ean mata and the scale coefficient is given by m a petvanto vcos ¢ . od in“u + an" v ~ 4. (55-13) On the initial meridian, a/v cos ¢ = deu, and (55.13) reduces to m= 1. On that part of the equator for which u = 0, m = nd'v, (55-14) and on the meridian \ = 4m, where u = K, a = (a/v cos $) se'v. (55-15) Fig. 45 shows also the pattern of isomegeths.+prozeyds (piogiey) TeuoTZeULe,UT oy Jo UOFOEford TOZuOOR ooeACUETy ToFNTY-FETOD “Oy “FTE100 56. Comparison and nomenclature of transverse Mercator pro- jections of the spheroid. The three varieties of transverse Mercator projection described above can conveniently be compared, in the vicinity of the initial meridian, by means of series. For the projection defined by (12.3), where ¥ is the, spher— oidal isometric latitude (9.4), the definition s =a lam *¢ leads at once to See eae eae (56.1) a 6 "24 7” 5040 With s/a used instead of s, all the formulae of Sec. 12 apply to this projection, except that for scale coefficient. From (9.7) and (12.9) we get a 1 “‘yeos ¢ (cosh Y-sin Aj? (56.2) On the initial meridian, this becomes m = a/v cos $coshy, which varies fron a minimyy value of 1 at the equator to a maximum of t+k) yi =k (4 +n) (-k at the pole, the latter value being 1-003 371 for the Hayford spheroid. For the Thompson projection, if ¢ in (54.18) is expanded in powers of w and the series then reversed, we get v 44K? ° 5+ 20h? + 5K oy a” Toes GT G0 NBS 64 + 627K? + 627K* 4 61Ke e see 6. iaemisomo(i= tae (56.3) As shown by (54.23), the scale on the initial meridian ranges from 1/k'? at the equator to 1 at the pole, the scale at the equator being 1-006 768 for the Hayford spheroid. For the Gauss-Kriiger projection, with m = 1 along the initial meridian, we can expand # = a@edw in powers of , and substitute (56.3), witha = 1, in this series, to get 8 1 5- Ke 61+ 267 414 = sO eect cs St 26K? eM So legen 201 Keys § Soaa(t ways’ + 7+ (56 4) which can be shown to be equivalent to the usual Taylor series for this projection. it can be seen that, for k = 0, all three projections become identical with the transverse Mercator projection of the sphere. A profusion of names has been applied to the first and third of these projections. In English-language publications, the name “transverse Mercator" is generally understood to refer to the projection given by (56.4), the constancy of scale along the initial meridian being regarded as the criterion for application of the name. A mathematician would be more inelined to apply it to the projection given by $26.1), since this is derinea by the game mathematical function (12.3) as establishes the transverse Mercator projection of the sphere. The projection with const: originally due to Lambert 1 and indicated how it could ant-scale initial meridian was 772, whd established it for the sphere be modified for the spheroid. It was101 next derived by Gauss c.1822, as one example of his investigations in conformal representation, and was introduced by him into the survey of Hannover. It is sometimes known as the Gauss conformal projection, a description which is not sufficient to identify it, and sometimes as Gauss's Hannover projection. Gauss left but few details, and a more detailed analysis, aiming to reproduce Gauss's method, was published by Schreiber 1866. Formulae adapted to logarithmic computation were developed by Kriiger 1912, whence came the name, Gauss-Kriiger projection, now used in many European countries. Particular truncations of the defining series have received other names, such as Gauss-Boaga as used in Italy. The titles of the papers by Gauss, Schreiber, and Kriiger, with an indication of their content, are given in Jorjan and Eggert's Handbuoh der Vermessungemunde (there is an English trans— lation by the U.S. Army Map Service 1962), and from these it can be seen that all three wrote about both projections. The name Gauss-Kriiger is therefore attached to the projection given by (56.4) merely by usage, and not with complete historical justifi- cation. The projection given by (56.1) was due to Gauss 1843, who derived it by first projecting the spheroid conformally upon a sphere so that all the meridians correspond, and then making a transverse Mercator projection of this sphere. It was described by Schreiber 1697, and is sometimes known as Schreiber's double projection. Tt was also described by Kriiger 1914. The name Gauss-Laborde is used to describe it in France. It was derived in a more general form by Hotine 1947. The names, Gauss-Schreiber and Gauss-Kriiger, a6 used by Lee 1962 do not accord with European usage.