The Shortest Distance Between 2 Points On The Earth

Calculus Level 4

Assuming that the earth is a perfect sphere with radius 6378 kilometers, what is the expected straight line distance through the earth (in km) between 2 points that are chosen uniformly on the surface of the earth?


The answer is 8504.00.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

2 solutions

Pi Han Goh
Mar 20, 2014

Fix a point to find the average distance of the other points from it. Take the point as A A and centre O O . Consider another point B B on the surface. Let A O B = θ \angle AOB = \theta . A B = 2 sin ( θ 2 ) \overline{AB} = 2 \sin \left ( \frac {\theta}{2} \right) with band radius sin θ \sin \theta . So the average distance is

6378 1 4 π 0 π ( 2 π sin θ ) 2 sin ( θ 2 ) d θ = 6378 0 π 2 ( sin 2 θ 2 ) cos ( θ 2 ) d θ = 6378 4 0 1 y 3 d y , let y = sin θ 2 = 6378 4 3 = 8504 \begin{aligned} 6378 \cdot \frac {1}{4\pi} \displaystyle \int_0^{\pi} (2 \pi \sin \theta ) \cdot 2 \sin \left ( \frac {\theta}{2} \right ) \mathrm{d} \theta & = & \displaystyle 6378 \cdot \int_0^{\pi} 2 \left ( \sin^2 \frac {\theta}{2} \right ) \cos \left ( \frac {\theta}{2} \right ) \mathrm{d} \theta \\ & = & \displaystyle 6378 \cdot 4 \int_0^1 y^3 \mathrm{d} y, \text{ let } y = \sin \frac {\theta}{2} \\ & = & 6378 \cdot \frac {4}{3} = \boxed{8504} \\ \end{aligned}

In short, the average distance between two points of a sphere with radius r r is 4 r 3 \frac {4r}{3}

Alternatively, consider a unit sphere which satisfy the equation x 2 + y 2 + z 2 = 1 x^2+y^2+z^2=1 . Without loss of generality, fix one point to be ( 1 , 0 , 0 ) (1,0,0) while the other is ( X , Y , Z ) (X,Y,Z) . the distance is ( 1 X ) 2 + Y 2 + Z 2 = 2 2 X \sqrt{ (1-X)^2+Y^2+Z^2} = \sqrt{2-2X} , so the average is 1 2 1 1 2 2 X d X = 1 4 0 4 w d w = 4 3 \displaystyle \frac {1}{2} \cdot \int_{-1}^1 \sqrt{2-2X} \mathrm{d}X = \frac {1}{4} \int_0^4 \sqrt{w} \space \mathrm{d} w = \frac {4}{3} . In this case, multiply it by 6378 6378 and you should get the same answer.

For further reading, you may refer to this

The 2nd half of this proof depends on the fact that, given one point on a sphere, another chosen at random on its surface is found at any latitude with equal probability. This stems from the fact that the area of a spherical segment is a function of the radius of the sphere and the height of the segment, regardless of what latitude it is. In trying to pick points randomly on a sphere, it's important to ensure that the distribution of the points occurs uniformly along the latitudes from pole to pole.

Because of this fact, it's okay to go ahead with the integration as you've done it, instead of having to weight it as might be normally expected in other cases.

Michael Mendrin - 7 years, 2 months ago

Log in to reply

Woah. I didn't see this comment (haha, took me over 500 days to reply). Thanks. Does your comment answers Mattapalli Ram's concern?

Pi Han Goh - 5 years, 8 months ago

Correction. It should be 1 2 1 1 2 2 X d X . \frac{1}{2}\int_{-1}^1\sqrt{2-2X}\;dX. Overall, indeed a nice solution! (y)

Tunk-Fey Ariawan - 7 years, 2 months ago

Log in to reply

FIXED! THANKS

Pi Han Goh - 7 years, 2 months ago

Could you elaborate on the second method? I mean in view of the fact that x-values are not all equally likely.

Peter Byers - 7 years, 2 months ago

Log in to reply

You're choosing a fixed point ( 1 , 0 , 0 ) (1,0,0) , then consider a random point ( X , Y , Z ) (X,Y,Z) , find the expected value.

Pi Han Goh - 7 years, 2 months ago

Log in to reply

Peter Byers is making a good point. Suppose this didn't involve a sphere, but some other solid of revolution. Then your 2nd method likely wouldn't have worked at all, since random points on its surface wouldn't likely be evenly distributed along the axis of the solid. The sphere happens to have this property that points distributed randomly on its surface are also uniformly distributed along its axis.

Michael Mendrin - 7 years, 2 months ago

Hello Pi Han Goh. I am actually confused with your solution. As per my understanding, selecting a point randomly from a surface of sphere (or infact a circle) and if you consider the marginal distribution of X coordinate, it may not follow a uniform distribution. Which violates the assumption in your proof. If we consider the same approach with respect to the angle AOB, and assume it follows a U(0,2 pi) distribution, the answer is 4 R/pi. Kindly let me know if I am wrong.

Mattapalli Ram - 7 years, 2 months ago
Tunk-Fey Ariawan
Mar 25, 2014

Let A A and B B be the two points that are uniformly and randomly chosen on the surface of the earth. Let d d be the straight line distance between A A and B B . See the picture below.

Circle Intersection of Earth Circle Intersection of Earth

Using Δ ABO \Delta\text{ABO} , D \text{D} can easily be determined. d = 2 R sin ( ϕ 2 ) . d=2R\sin\left(\frac{\phi}{2}\right). In this case, there are three random variables, namely: D , Φ , \text{D},\Phi, and Θ \Theta . D \text{D} is the straight line distance between A A and B B , Φ \Phi is the polar or zenith angle from the positive z z -axis with 0 ϕ π 0\le\phi\le\pi , and Θ \Theta is the azimuthal angle in the x y xy -plane from the x x -axis with 0 θ < 2 π 0\le\theta<2\pi .

Spherical Coordinates Spherical Coordinates

Assuming that Φ \Phi and Θ \Theta are independent, the joint density function (pdf) of Φ \Phi and Θ \Theta turns out to be f Φ , Θ ( ϕ , θ ) = f Φ ( ϕ ) f Θ ( θ ) f_{\Phi,\Theta}(\phi,\theta)=f_{\Phi}(\phi)\cdot f_{\Theta}(\theta) where f Φ ( ϕ ) = 1 2 sin ϕ f_{\Phi}(\phi)=\frac{1}{2}\sin\phi and f Θ ( θ ) = 1 2 π . f_{\Theta}(\theta)=\frac{1}{2\pi}. Thus, the average distance or expectation value of D \text{D} is E [ D ] = θ = 0 2 π ϕ = 0 π d f Φ , Θ ( ϕ , θ ) d ϕ d θ = θ = 0 2 π ϕ = 0 π 2 R sin ( ϕ 2 ) f Φ ( ϕ ) f Θ ( θ ) d ϕ d θ = 2 R θ = 0 2 π ϕ = 0 π sin ( ϕ 2 ) 1 2 sin ϕ 1 2 π d ϕ d θ = R 2 π θ = 0 2 π d θ ϕ = 0 π sin ( ϕ 2 ) sin ϕ d ϕ = R 2 π 2 π ϕ = 0 π sin ( ϕ 2 ) 2 sin ( ϕ 2 ) cos ( ϕ 2 ) d ϕ = 2 R ϕ = 0 π sin 2 ( ϕ 2 ) cos ( ϕ 2 ) d ϕ = 2 R ϕ = 0 π sin 2 ( ϕ 2 ) 2 d ( sin ( ϕ 2 ) ) = 4 R 1 3 sin 3 ( ϕ 2 ) ϕ = 0 π = 4 3 R = 4 3 6378 = 8504 km \begin{aligned} \text{E}[\text{D}]&=\int_{\theta=0}^{2\pi}\int_{\phi=0}^{\pi}d\cdot f_{\Phi,\Theta}(\phi,\theta)\;d\phi\,d\theta\\ &=\int_{\theta=0}^{2\pi}\int_{\phi=0}^{\pi}2R\sin\left(\frac{\phi}{2}\right)\cdot f_{\Phi}(\phi)f_{\Theta}(\theta)\;d\phi\,d\theta\\ &=2R\int_{\theta=0}^{2\pi}\int_{\phi=0}^{\pi}\sin\left(\frac{\phi}{2}\right)\cdot \frac{1}{2}\sin\phi\cdot\frac{1}{2\pi}\;d\phi\,d\theta\\ &=\frac{R}{2\pi}\int_{\theta=0}^{2\pi}\;d\theta\int_{\phi=0}^{\pi}\sin\left(\frac{\phi}{2}\right)\sin\phi\;d\phi\\ &=\frac{R}{2\pi}\cdot2\pi\int_{\phi=0}^{\pi}\sin\left(\frac{\phi}{2}\right)\cdot2\sin\left(\frac{\phi}{2}\right)\cos\left(\frac{\phi}{2}\right)\;d\phi\\ &=2R\int_{\phi=0}^{\pi}\sin^2\left(\frac{\phi}{2}\right)\cos\left(\frac{\phi}{2}\right)\;d\phi\\ &=2R\int_{\phi=0}^{\pi}\sin^2\left(\frac{\phi}{2}\right)\cdot2\;d\left(\sin\left(\frac{\phi}{2}\right)\right)\\ &=4R\cdot\left.\frac{1}{3}\sin^3\left(\frac{\phi}{2}\right)\right|_{\phi=0}^\pi\\ &=\frac{4}{3}R\\ &=\frac{4}{3}\cdot6378\\ &=\boxed{\color{#3D99F6}{8504\text{ km}}} \end{aligned}


# Q . E . D . # \text{\# }\mathbb{Q.E.D.}\text{ \#}

How did you come up with the density function ?

Vedananda Pandeya - 7 years, 2 months ago

Log in to reply

In spherical coordinate f Φ ( ϕ ) sin ϕ f_{\Phi}(\phi)\sim\sin\phi and f Θ ( θ ) k where k is constant . f_{\Theta}(\theta)\sim k\quad\text{where } k\text{ is constant}. Thus, to make correct weighting I must first compute the factor of normality (normalizing), which has to be chosen such that the probabilities add up to 1 1 . Therefore 0 π f Φ ( ϕ ) d ϕ = 1 0 π c sin ϕ d ϕ = 1 c = 1 2 \begin{aligned} \int_0^\pi f_{\Phi}(\phi)\;d\phi&=1\\ \int_0^\pi c\sin\phi\;d\phi&=1\\ c&=\frac{1}{2} \end{aligned} and 0 2 π f Θ ( θ ) d θ = 1 0 2 π k d θ = 1 k = 1 2 π . \begin{aligned} \int_0^{2\pi} f_{\Theta}(\theta)\;d\theta&=1\\ \int_0^{2\pi} k\;d\theta&=1\\ k&=\frac{1}{2\pi}. \end{aligned}

Tunk-Fey Ariawan - 7 years, 2 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...