1. Introduction
Let us agree that a map projection is a mapping of a curved surface, for example the Earth's sphere or ellipsoid, into a plane. Changes that occur in such mapping are called distortions. We can differentiate distortions of length, area, and angles. If there is no distortion at any point of a line/curve we say that it is a zero distortion line or a standard line. If the same is true of a parallel, we say that it is a standard parallel. At first glance, these are familiar, generally accepted definitions, but this paper will show that a major problem exists and provides a correct, mathematically based approach.
First of all, let us remember that developing a cylindrical or conical surface into a plane is isometric mapping. In other words, all distances should remain the same after the developing procedure. Obviously, there is no reason to use a developable surface in a definition of a map projection if developing will cause additional distortions. According toDeetz and Adams (1969, pages 10-11):
There are some surfaces, however, that can be spread out in a plane without any stretching or tearing. Such surfaces are called developable surfaces and those like the sphere are called nondevelopable. The cone and the cylinder are the two well-known surfaces that are developable. Any curve drawn on the surface will have exactly the same length after development that it had before.
In the literature on map projections, the common interpretation is that the Earth's sphere or ellipsoid is first mapped onto an auxiliary developable surface, which is then developed in a plane and thus becomes a map (see, for example,Deetz and Adams 1969,Richardus and Adler 1972,Slocum et al. 2009,Clarke 2015). This approach is generally unsatisfactory. This is because the use of developable surfaces in the definition of map projections applies to only a small number of projections. If a projection is called cylindrical, it does not mean that it is a mapping onto a cylinder. Such an understanding is very common but wrong. Correct thinking is: a projection is called cylindrical if a map produced in such a projection can be bend in a cylindrical surface. The same approach holds for conical projections (Close and Clarke 1911). In addition to cylindrical and conical projections, there are many other types, such as azimuthal, pseudocylindrical, pseudoconical, conditional, and so on, which cannot be interpreted by mapping onto a cylinder or conical surface. Some attempts have been made in that direction. For example, a pseudocylindrical projection is interpreted as mapping onto an oval surface (oval, ovoid) (Clarke 2015), without noticing that such a surface is undevelopable. It does not make sense to map a sphere, which is an undevelopable surface, onto another undevelopable surface. Any surface can be transformed by animating it into any other surface. This simply confuses the essence of map projections.
More than hundred years ago, the Encyclopaedia Britannica (Close and Clarke 1911) said:
Conical projections are those in which the parallels are represented by concentric circles and the meridians by equally spaced radii. There is no necessary connexion between a conical projection and any touching or secant cone. The name conical is given to the group embraced by the above definition, because, as is obvious, a projection so drawn can be round to form a cone.
Lee (1944) remarked in The Nomenclature and Classification of Map Projections:
The projections are termed cylindric or conic because they can be regarded as developed on a cylinder or cone, as the case may be, but it is as well to dispense with picturing cylinders and cones, since they have given rise to much misunderstanding. Particularly is this so with regards to the conic projections with two standard parallels: they may be regarded as developed on cones, but they are cones which bear no simple relationship to the sphere.
The most famous cartographers, for instance Mercator and Lambert, did not use any intermediate or developable surfaces in the projections known today as the Mercator projection and Lambert conformal conical projection. The introduction of developable surfaces in the theory of map projections is artificial and unnecessary, in general. One of the contributions of this paper is the author's effort to show that the theory of map projections should be approached mathematically, not dogmatically. We advocate thinking critically about broadly accepted, but sometimes wrong, ideas and statements.
Figure 1. Illustration of standard parallels as secant parallels. Source:Deetz and Adams 1969, page 80. This is a misguided approach accompanied by the caption: "Diagram illustrating the intersection of a cone and sphere along two standard parallels". In other words, for Deetz and Adams (and many others) standard parallels and secant parallels are identical. / Slika 1. Ilustracija standardnih paralela kao presječnih paralela. Izvor:Deetz i Adams 1969, str. 80. To je pogrešno vođen pristup praćen tekstom ispod slike: "Diagram illustrating the intersection of a cone and sphere along two standard parallels". Drugim riječima, za Deetza i Adamsa (i mnoge druge) standardne paralele i presječne paralele su identične.
Figure 1, taken fromDeetz and Adams (1969), shows the intersection of a cone and sphere along two parallels. For Deetz and Adams (and many others), standard parallels and secant parallels are identical. This paper disagrees with such an approach. Our main goals are to propose definitions of three different types of parallels: secant, standard and equidistant, and to show that some broadly accepted facts about secant and standard parallels found in many references are incorrect and should be revised. This requires a critical approach by the reader to established customs in map projection teaching and research. In several earlier published papers (Lapaine 2015,2017a,2017b,2018,2019) it was shown that there are conical and azimuthal projections with three and more standard parallels. Obviously, this could not be explained by using the secant projections approach, and led to differentiating between standard and secant parallels. This paper goes further and introduces equidistant parallels in order to facilitate understanding and indicate the need to abandon the secant projections approach.
2. Terminology
Let us agree that the principal (linear) scale PS is the ratio of the length in the plane of projection and its origin on the surface (sphere, ellipsoid) to be projected/mapped. PS is usually indicated on maps because it determines the general degree of reduction of the length on the map. On most maps, it is usually simply called ‘scale' and is known as the map scale.
It is well known that scale changes from point to point, and at certain points usually depends on direction. This is the local scale. The local linear scale factor c is the ratio of the differential of the curve arc in the plane of projection and the differential of the corresponding curve arc on the ellipsoid or spherical surface (more details in section 5).
The local scale LS is the product of the principal scale PS and the local linear scale factor c:
LS = PS x c (1)
Since the local scale also depends on the direction, then instead of (1) it would be more correct to write
LS(α) = PS x c (α) (2)
where we denote the observed direction as α.
Length distortion is the difference between 1 and the local linear scale factor. Distortion is expressed by a number, so if c(α) = 1 for each α, then instead of the expression ‘no distortion' it would be better to say ‘distortion is equal to zero'.
This definition is found in the Multilingual Cartographic Dictionary (Borčić et al. 1977) and the Geodetic-Geoinformatic Dictionary (Frančula and Lapaine 2008):
line without distortion = Line on the map along which there is no distortion of lengths, area and angles or lines along the principal scale is preserved at all points in all directions. For perspective projections, these are the lines by which the auxiliary surface or the projection plane touches (touching parallel, touching meridian) or intersects (secant parallel, secant meridian) the sphere or ellipsoid. Remarks: (1) For the line without distortion, the term standard line is used, e.g. standard parallel (2) The point where the projection plane is tangent (touches) the surface of the ball or ellipsoid is called the touch point. See: projection, azimuthal. English: line, standard. German: Berührungslinie.
This definition actually contains two: a line without distortion and a line along which the principal scale has been preserved. Furthermore, introducing perspective projections and, in this respect, the tangent and secant parallel or meridian in the definition of a line without distortion is not good because the secant parallel does not need to be simultaneously parallel without distortion. Also, there is disagreement between languages about naming a line without distortion. For instance, in English it is called a standard line, and in German, the touch line (Berührungslinie).
In Enzyklopädischer Wörterbuch Kartographie in 25 Sprachen (Neumann 1997) we read this on page 76:
37.1 Berührungslinie = Linie auf der Kugel oder auf dem Ellipsoid [ex: Breitenkreis, Meridian etc.] in der eine für die kartographische Abbildung benutzte Hilfsab-bildungsfläche diese (a) berührt [ex: (1) Berührungsbreitenkreis, (2) Berührungs-meridian] oder (b) schneidet [ex: (3) Schnittbreitenkreis, (4) Schnittmeridian].
NB: Berührt die Hilfsabbildugsfläche einen Punkt, so spricht man vom (5) Berührungs-punkt.
E: standard line. A line on a Map Projection along which the Principal Scale obtains. (1) tangent parallel; (2) tangent meridian; (3) secant parallel; (4) secant meridian.
The terminology in the other 23 languages follows.
In the above definition in German, Berührungslinie is literally translated as touching line, and the other terms are touching parallel, touching meridian, secant parallel and secant meridian. The auxiliary surface (Hilfsabbildugsfläche) has a decisive role, while distortion is not mentioned at all!
In the English definition in same dictionary, a standard line is "a line in map projection along which the scale is equal to the principal scale". The auxiliary surface is not mentioned explicitly, but the terms tangent parallel, tangent meridian, secant parallel and secant meridian occur. In French, it is the "contact line or intersection with auxiliary projection surface", and in Russian the "zero distortion line, i.e. a line on the map where the principal scale at each point is preserved". The same understanding appears in the Portuguese definition.
In ESRI's Web Dictionary (ESRI 2017) we find this definition: "Standard line = [map projections] A line on a sphere or spheroid that has no length compression or expansion after being projected; usually a standard parallel or central meridian". Thus, only what happens along the line is important, not around it, although it is known that the linear scale factor or scale at each point generally depends on the direction (Tissot indicatrix).
Snyder and Voxland (1989) say this of perspective conical projection. " Scale. True along one or two chosen standard parallels, which may be on the same side of or both sides of the Equator... Distortion. Free of distortion only along the one or two standard parallels." These statements are incorrect because a simple perspective conical projection cannot have two standard parallels.
From the above, we can conclude that there is terminological confusion. The standard line and secant line are not uniformly defined terms. Confusion appears to arise from misunderstandings and results in the acceptance of unfounded assertions. We will attempt to resolve this confusion in this paper.
The same idea can be applied to the transverse and oblique aspects of both cylindrical and conical projections.
In order to be mathematically correct in all the sections that follow, we refer to published papers byLapaine 2015,2017a,2017b,2018,2019.
3. Points, Lines and Areas with Zero Distortions
Let us say that at some point the distortion is zero if the local linear scale factor is equal to 1, i.e. if it is
c(α)=1, for each α ∈ [0,2π], (3)
where c(α) is defined in (4)
where E, F and G are coefficients in the first differential form of the map projection. We use a sphere intentionally to facilitate understanding and to shorten derivations.
Note there is a difference between zero and none. Zero is a number, while none is not. Due to the fact that distortion is a number, it would be preferably to use "zero distortion" instead of "no distortion".
The requirement (3) is obviously equivalent to the condition
cmin = cmax = 1 (5)
that tell us it is a Tissot indicatrix transformed into the unit circle. Let us emphasize that in (3) the term "for each α ∈ [0,2π]" is important, and it is not enough that the written property is valid for only one α. In other words, in order for an ellipse to be a circle, it is necessary for all of its radiuses to be mutually equal or, in the same way, for the semi-axes to be of the same length. It is not enough for one semi-axis of the ellipse to be equal to 1 so that the ellipse is the unit circle. This should be clear to everyone, although it can be inferred from the introductory chapter that it is not.
In a special case, when F = 0, the expressions (5) can be written like this
h = k = 1, (6)
where h and k are local linear scale factors along meridian, and along parallel, respectively.
Let us introduce for the sake of abbreviation:
a = cmax,b = cmin (7)
Let us first note that a > 0 and b > 0 is always true.Table 1 shows all possible cases when a=1, b=1 or a=b=1. This is very simple but new in the theory of map projections:
(1,1)
If at some point a = 1, we can say that this point is locally equidistant in the direction of maximum local linear scale factor.
(1,2)
If at some point b = 1, we can say that this point is locally equidistant in the direction of minimal local linear scale factor.
(1,3)
If at some point a = b = 1, we can say that this is a point with zero distortion, a zero-distortion point or a standard point.
(2,1)
If at all points of a line a = 1, then it is not generally a line with zero distortion, but we can say that this line is equidistant in the direction of maximum local linear scale factor.
(2,2)
If at all points of a line b = 1, then it is not generally a line with zero distortion, but we can say that this line is equidistant in the direction of minimal local linear scale factor.
(2,3)
If at all points of a line a = b = 1, then it is a line with zero distortion, a zero-distortion line or a standard line.
(3,1)
If at all points of an area a = 1, then we can say that this area is equidistant in the direction of maximum local linear scale factor.
(3,2)
If at all points of an area b = 1, then we can say that this area is equidistant in the direction of minimal local linear scale factor.
(3,3)
The expression a = b = 1 cannot be true at all points of a two-dimensional area on the map, as this would mean map projection without distortion. Leonhard Euler first proved that this was not possible (Euler 1777).
Table 1. All possibilities of standard points and lines vs. equidistant points and lines in map projections. If a = b = 1, we have a standard point or line. If a = 1 or b = 1, we have an equidistant point, line or area. The case (3,3) is not possible.
In a special case, when F = 0, then we can use h and k instead of a and b.
If at all points of a meridian we have h = 1, then it is not generally a zero-distortion meridian, but a meridian along which, or in the direction of which, the distortion is zero. We can say that the meridian is mapped equidistantly.
If at all points of a mapped area h = 1, then we say that the map projection is equidistant along meridians. For example, normal aspect of the Postel projection is azimuthal projection equidistant along all meridians.
If at all points of a parallel we have k = 1, then it is not generally a zero-distortion parallel, but a parallel along which, or in the direction of which, the distortion is zero. We can say that the parallel is mapped equidistantly.
If at all points of a mapped area k = 1, then we say that the map projection is equidistant along parallels. For example, normal aspect of the orthographic projection is azimuthal projection equidistant along parallels.
If at all points of a meridian we have h = k = 1, then it is a zero-distortion meridian, or the distortion of that meridian is zero in all directions, i.e. this is a standard meridian.
If at all points of a parallel we have h = k = 1, then it is a zero-distortion parallel, or the distortion of that parallel is zero in all directions, i.e. this is a standard parallel.
4. Equidistant, Standard and Secant Parallels
4.1 Equidistant, standard and secant parallels in normal aspect cylindrical projectionsNormal aspect cylindrical projections of a sphere are mappings defined by the formula
x = Rn(λ-λ)0, y = Rf(φ), (8)
where
, λ ∊ [-π ,π], constants are n>0, R>0 and λ0 ∊ [-π ,π], function y = Rf(φ) is piecewise continuous, differentiable, monotone increasing and odd. As in any map projection, x and y are coordinates of a point in a rectangular (mathematical, right oriented) plane coordinate system. As anyone can see, this is about mapping into the plane, not onto the surface of a cylinder. For such mapping,
and the first differential form is
The local linear scale factor squared for a mapping a sphere by using normal aspect cylindrical projection (8) is
Local linear scale factors along a meridian and a parallel, respectively are
For normal aspect cylindrical projections, along standard parallels with latitudes
φ±φ1 it should be h(φ1) = k(φ1) = 1, (13)
or considering (12) and (9)
and
From the formula (15) and the properties of trigonometric function cosine we can conclude that if the latitude of the standard parallel is φ 1, then it should be -φ 1, too. The number of standard parallels in normal aspect cylindrical projections we can conclude based on the relations (14) and (15).
If 0<n<1 then two standard parallels exist (φ1 and -φ1), if it is also
If n = 1 then only one standard parallel (φ1 = 0) can exist, the Equator, providing that it is
If n>1 then there is no φ1 for which (15) holds, i.e. there is no standard parallel at all.
Let φ and -φ be the latitudes of two parallels mapped by normal aspect cylindrical projection (8). The vertical distance between these two parallels on the sphere is 2 Rsinφ, and the distance between their images in the plane of projection (seeFig 2) is 2y( φ). It is clear that 2 y( φ) = 2Rsinφ, i.e. y( φ) = Rsin φ is not true in general. It is true for the Lambert equal-area projection only. Thus, bending a map made in the normal aspect cylindrical projection onto a cylindrical surface and then putting it in a particular position with the sphere generally makes no sense. In other words, map projection and folding or unfolding of the cylindrical surface are two completely different actions. Since secant parallels do not have any particular property in terms of distortion distribution, we conclude that the use of cylindrical surfaces in the theory of map projections generally makes no sense and can lead to wrong conclusions about the identities of secant and standard parallels (Fig 3). The use of cylindrical surfaces makes sense in perspective projections on a cylinder only, but such projections are rarely applied.
Figure 2. Illustration showing that the distance between two parallels symmetrical to the Equator and the distance between their images in the plane of projection are different in general. It follows that equidistant or standard parallels can not be secant parallels because bending is an isometry, i.e. squeezing or stretching is not allowed. / Slika 2. Ilustracija koja pokazuje da udaljenost između dviju paralela simetričnih s obzirom na ekvator i udaljenost između njihovih slika u ravnini projekcije općenito nisu jednake. Odatle slijedi da ekvidistantne ili standarnde paralele ne mogu biti istodobno presječne paralele jer je savijanje izometrija, tj. stiskanje ili razvlačenje nije dopušteno.
Figure 3. The usual erroneous illustration: standard parallels are at the same time secant parallel in normal aspect cylindrical projection (Mercator 2009). We have proved that this is impossible. The illustration is an illusion. / Slika 3. Uobičajena pogrešna ilustracija: standardne paralele su istodobno presječne paralele u uspravnom aspektu cilindrične projekcije (Mercator 2009). Dokazali smo da je to nemoguće. Ta ilustracija je iluzija.
To summarize about cylindrical projections given by (8):
Two parallels (φ1 and -φ1) are equidistant parallels if
n = cosφ 1 (16)
Two parallels (φ1 and -φ1) are standard parallels if holds (16) and
Two parallels (φ1 and -φ1) are secant parallels if the distance between these two parallels on the sphere equals the distance between their images in the plane of projection:
f (φ1) = sinφ1. (18)
Two examples on cylindrical projections follow.
4.1.1. Example 1Let the normal aspect cylindrical projection be defined as
λ ∊ [-π ,π], where the constants are R>0 and λ0 ∊ [-π ,π].
Since f(φ) = φ and
we can conclude that all meridians are mapped equidistantly (h = 1) in this projection. From the condition
, which is easy to derive from (15) and (19), we see that the parallels that correspond to latitudes
are also mapped equidistantly
Since all the meridians in this projection are mapped equidistantly, the two equidistantly mapped parallels are also standard parallels
The distance between two parallels that corresponds to latitudes φ and -φ on a sphere is 2 Rsin φ (seeFig. 2). If
then the distance between these two parallels on the sphere is
The distance of their images in the plane of projection is 2y(φ) = 2R(φ), which for
amounts
Thus, by bending the map into the cylindrical surface and placing the cylinder so that its axis fits in with the straight line connecting the North and South Poles, and putting the image of the Equator in the equatorial plane, it is impossible for the secant parallels coincide with the equidistant (in this case also standard) parallels. Thus, we have shown that secant and standard parallels in cylindrical projections generally cannot be identical.
4.1.2. Example 2The Behrmann projection equations reads (Lapaine 2018)
For that projection
so we have
It is immediately apparent that this projection is equal-area because it is hk = 1 for any φ.
Snyder and Voxland (1989) write that the Behrmann Projection is true along latitude 30° and that it is a projection on a cylinder secant at ±30° (True along latitudes ±30°, projection onto a cylinder secant at ±30° ). The standard parallels of the Behrmann projection correspond to latitudes ±30°. The distance between these parallels on the sphere with radius R is obviously R. In the plane of Behrmann's projection the distance between their images is
which is obviously different from R. This means that a map cannot be made in Behrmann's projection by developing into a plane the Behrmann cylindrical surface that cuts the sphere along the parallels of ±30° latitudes. Conversely, a map made in Behrman's projection cannot bend into a cylindrical surface that will intersect the sphere along the parallels of ±30° latitudes. Identifying the equidistant or standard parallel and the parallel in which the cylindrical surface intersects the Earth's sphere is obviously not a valid procedure (Lapaine 2018).
4.2. Equidistant, standard and secant parallels in normal aspect conical projectionsA conic projection is mapping defined by the formulae
ρ = Rf(φ), θ = n(λ-λ0) (23)
where
λ ∊ [-π ,π], the constants are n>0, R>0 and λ0 ∊ [-π ,π],
and the function ρ = Rf(φ) is continuous, with positive values and monotone decreasing, or monotone increasing, and ρ and θ are coordinates of a point in the polar coordinate system in the plane. As can be seen, this is about mapping into the plane, not onto a conical surface. For such mapping we have
and the first differential form
The square of the local linear scale factor of normal aspect conic projection (23) is
From (26) we can read the local linear scale factor along the meridians and parallels, respectively
The minus sign in the formula for h was chosen because functions ρ and f are monotone decreasing by hypothesis. If these functions are monotone increasing, then φ should be replaced by -φ in the derivation that follows.
In normal aspect conic projections, F = 0 and (27) holds.
If for any
it is
then all meridians are mapped equidistantly. If for some
it is
then parallels corresponding to these latitudes are mapped equidistantly.
Along a standard parallel φ = φ 1 it should be true that
h(φ 1) = k(φ 1) = 1, (28)
that is,
(29)
and
We cannot make any conclusions about the number of standard parallels in normal aspect conical projections as we could with cylindrical ones. This is because for the given n there are infinitely many continuous functions ρ = Rf(φ), which are differentiable with positive values and monotone decreasing for
, and which satisfy conditions (29) and (30) for several different values of the latitudes
Let us note two parallels on the sphere with radius R, and let φ' and φ'' be the corresponding latitudes 0 ≤ φ'<φ'' ≤ 90°. The shortest distance d between these two parallels (Fig. 4) equals
The distance between their images on the map will be
However, the equality
is not true in general. Thus, it is not applicable to parallels mapped equidistantly and for standard parallels. This means that bending a map produced in the normal aspect conical projection into a conical surface and placing it in a particular position related to the sphere generally makes no sense. In other words, map projection and developing the conical surface into a plane are two quite different actions. Since secant parallels do not have any particular property in terms of distortion distribution, we conclude that the use of conical surfaces in the theory of map projections generally makes no sense and can lead to wrong conclusions about the identities of secant and standard parallels.
Figure 4 The distance d between two parallels given by their latitudes φ' and φ”. / Slika 4 Udaljenost d između dviju paralela zadanih njihovim geografskim širinama φ' i φ''.
To summarize about conical projections given by (23) and two parallels (0 ≤ φ' < φ" ≤ π/2) :
Two parallels (φ' and φ'') are equidistant parallels if
nf(φ') = cos φ' and nf(φ'') = cos φ''. (34)
Two parallels (φ' and φ'') are standard parallels if holds (34) and
df/dφ (φ') = 1 and df/dφ (φ'') = 1 (35)
Two parallels (φ' and φ'') are secant parallels if the distance between these two parallels on the sphere equals the distance between their images in the plane of projection:
f(φ')-f(φ") = 2sin sin(φ"-φ')/2 (36)
Three examples on conical projections follow.
4.2.1. Example 3
Let the normal aspect conical projection be defined as:
where
λ ∊ [-π ,π], R>0 and λ0 ∊ [-π ,π]. Since
and
we conclude that all meridians are mapped equidistantly.
The condition (34) implies that two parallels corresponding to latitudes 0° and 60° are mapped equidistantly
Since all meridians in this projection are mapped equidistantly (h(φ) = 1 for all φ), the two equidistantly mapped parallels are standard parallels.
The distance d between two parallels with latitudes φ' and φ'' on the sphere is given by (36). If φ' = 0° and φ'' = 60°, then the distance between them on the sphere is 2Rsin30° = R. The distance between their images in the plane of projection is
, and for φ' = 0° and φ'' = 60° this is
which is different from R. Thus, by bending the map into the conical surface and placing the cone so that its axis coincides with the straight line connecting the North and South Poles, it is not possible to obtain secant parallels identical with the standard parallels. Thus, we have shown that secant and standard parallels in conical projections generally cannot be identical.
4.2.2. Example 4In their well-known book on map projections, Richardus and Adler (1972, page 94) interpret Lambert's conformal conical projection with two standard parallels and illustrate the derivation by depicting a cone that intersects the ellipsoid into two parallels (Fig. 5). Richardus and Adler wanted to show that these parallels were secant and standard parallels, but this cannot be true if we accept the definitions of secant and standard parallels recommended in this paper. Let us prove this using a simple example of a sphere of radius 1 and Lambert's conformal conical projection with two standard parallels corresponding to the latitudes φ' = 0° and φ'' = 60°.
Figure 5. Illustration of standard parallels as parallels in which the conical surface intersects the Earth's ellipsoid. Source:Richardus and Adler (1972, page 94) . This is generally a misguided approach and is accompanied by the inaccurate statement, "This is the Lambert conical conformal projection with two standard parallels. The cone intersects the ellipsoid at these parallel circles". Figure without red cross was taken fromRichardus and Adler (1972, page 94) , and it is an illusion. Red cross was used to stress that something is wrong with the figure. After looking at the figure and reading the text, it is expected that the reader will be able to distinguish between fact and illusion. / Slika 5. Ilustracija standardnih paralela kao paralela u kojima konusna ploha siječe Zemljin elipsoid. Izvor:Richardus i Adler (1972, str. 94) . To je općenito pogrešan pristup koji je popraćen netočnom tvrdnjom: "To je Lambertova konusna konformna projekcija s dvije standardne paralele. Konus siječe elipsoid u tim kružnim paralelama". Slika bez crvenih crta preuzeta je iz knjigeRichardusa i Adlera (1972, str. 94) , i to je jedna obmana. Precrtano crvenim je upotrijebljeno da bi se istaknulo da na toj slici nešto nije kako treba. Nakon gledanja slike i čitanja teksta, čitatelj bi trebao razlikovati činjenicu od iluzije.
First of all, there is no intermediate conical surface in the Lambert conformal conic projection. This is evident from Lambert's original work or the English translation (Lambert 1772).
If a map in the Lambert conic conformal projection has two standard parallels (right-hand part ofFig. 5), then it is not possible to bend it into a conical surface which will cut the earth's sphere along these two parallels (left-hand part ofFig. 5). Though this may surprise some people, it will be proved in this example.
In order for the conical projection to be conformal, it must be true that
h = k, (38)
where h and k are defined in (27). From there, because R =1, it follows
and then
from where, after integration, we get
that is,
where C is a constant of integration.
Let φ' = 0° and φ'' = 60° be two standard parallels. The conditions for standard parallels
and
form a system of two equations with two unknowns, n and C, which can be transformed into
and
from where it follows that
and
If we now substitute values φ' = 0° and φ'' = 60° we will get
The distance between the images of standard parallels with latitudes φ' = 0° and φ'' = 60° in the plane of Lambert conformal conical projection equals
Let us bend the map into a conical surface and place the cone so that its axis coincides with the straight line connecting the North and South Poles on the sphere. It is not possible to obtain secant parallels identical with the standard parallels, because the shortest distance between standard parallels on the sphere measured along a straight line connecting the two parallels is equal to 1 (Fig. 6). Bending (and developing) is isometry, that is, transformation that preserves distances.
If we bend a map made in the normal aspect conical projection into a conical surface and then try putting it in a particular position with the sphere, standard parallels in the projection will not coincide with standard parallels on the sphere. In other words, map projection and folding or unfolding of the conical surface are two completely different actions.
We have proved that secant parallels are not standard parallels, in general.
Figure 4 The distancFigure 6. In conical conformal projection of a unit sphere the distance between two standard parallels latitudes φ 1=0° and φ 2=60° equals 1 on the sphere, and less then 1 in the plane of projection.e d between two parallels given by their latitudes φ' and φ”. / Slika 6. U Lambertovoj konusnoj konformnoj projekciji jedinične sfere udaljenost između dviju standardnih paralela kojima odgovaraju geografske širine φ1= 0° and φ2= 60° jednaka je 1 na sferi, a manja je od 1 u ravnini projekcije
4.2.3. Example 5Snyder (1993), in his famous book Flattening the Earth, says this on page 123, "Like the central cylindrical projection described earlier, a conic projection may be developed by geometrically projecting the globe from its centre onto a cone either tangent at one parallel or secant at two. It is so elementary that there seems to be no formal origin of it ...". "For two standard parallels φ 1 and φ 2, the polar coordinates are as follows:
, θ = λsin(φ0),
where
... For the tangent form with one standard parallel, φ1 may be equated to φ1. If the standard parallels favour the northern hemisphere, only part of the southern hemisphere can normally be shown". Snyder was apparently convinced that there was a gnomonic or central perspective projection on a cone with two standard parallels. He did not check the truth of this and accepted it as a fact, though it is erroneous. Here is the evidence!
The local linear scale factor along the meridian in a normal aspect conical projection is calculated according to (27)
For a function ρ = ρ(φ) defined by (49) one can easily calculate
and then
from where
which is different from 1, unless φ1 = φ2. We have shown that a simple gnomonic perspective projection on a conical surface cannot have two standard parallels. In other words, if the conical surface intersects the sphere at two parallels, these two parallels cannot be standard parallels in the gnomonic perspective projection.