You are here

A Political Redistricting Tool for the Rest of Us - The Population Density Function

Author(s): 
Evan Kleiner (Whitman College) and Albert Schueller (Whitman College)

<h3>The Population Density Function</h3>

  <p> All of the data collected by the United States during its decennial
  census is freely available from the <a
   href="http://www.census.gov" target="blank">Census Bureau</a>.  The data that we used
  throughout this paper comes from the 2010 data set.  Our redistricting
  approach is driven primarily by the population density function, \(\rho\), of
  a particular state.

  <p> The Census Bureau provides population data down to the resolution of
  a single census tract.  The Bureau also provides the geographic shapes of
  each census tract as they were during the collection of the data.  If we
  let \(T\) be a census tract, \(A(T)\) be the area of the census tract (as
  projected upon the GCS North American 1983 [arcGIS]), and \(p(T)\) be the
  population of the tract, we define the population density function as
  \[\rho(\phi,\theta) = p(T)/A(T),\qquad  (\phi,\theta)\in T\] The density
  function, \(\rho\), is a piecewise constant function defined within the
  boundaries of a single state.  For definiteness, we assume that \(\phi\in
  [0,\pi]\) is latitude (with \(0\) at the North Pole) and \(\theta \in
  (-\pi,\pi]\) is longitude (with \(0\) at the Prime Meridian and \(\theta\)
  increasing to the east).  Though not a sphere, we will assume a spherical
  approximation of the earth with radius \(R=3958.755\) miles [<a
   href="/node/220897#moritz">Moritz</a>].

 
  <p> To facilitate numerical algorithms, we further discretize \(\rho\)
  using a uniform grid in the latitude/longitude domain as in this figure.  For simplicity, we assume that the
  state that we are redistricting is bounded by a spherical patch (some
  rectangle in the \((\phi,\theta)\)-plane) that is
  completely contained in both the northern hemisphere and in the western
  hemisphere.  Hence, the latitudes spanning the state are contained in the
  interval \([\phi_{\min},\phi_{\max}] \subset [0,\pi/2]\) and the longitudes
  are contained in the interval \([\theta_{\min},\theta_{\max}] \subset
  (-\pi, 0]\).  I.e.  \(\phi_{\min}\) is the northern most latitude of the state
  and \(\theta_{\min}\) is the western most longitude of the state.

  <p> We <a name="discretization">discretize</a> the spherical patch as
  \begin{eqnarray*}
  \phi_i & = & \phi_{\min} +
  \frac{\phi_{\max}-\phi_{\min}}{M},~i=0,1,\ldots M \\
  \theta_j & = & \theta_{\min} +
  \frac{\theta_{\max}-\theta_{\min}}{N},~j=0,1,\ldots N.  \end{eqnarray*}

  <p style="border-style:inset;background-color:#F5F6CE;">
 
   <p align="center">A uniform latitude/longitude grid on the surface of a
    sphere.</p>
   
  </p>
  We then generate a discrete density function on the patch \(\rho_{ij} =
  \rho(x_i,y_j).\)


  <p> In order to reconstitute the population of a particular region, it is
  necessary to know the area of of each grid square in the discretization.
  Recall that each grid square of latitude and longitude represents a small
  patch on the surface of the earth.  Furthermore, though the grid squares
  are uniform in the latitude and longitude domain, the actual surface area
  represented by each square depends upon its latitude again see this figure and this figure.  We approximate the area of the patch
  with upper-left hand corner at \((x_i, y_j)\) by assuming that the earth is
  a sphere and using a spherical surface integral.

  <p style="border-style:inset;background-color:#F5F6CE;"></p>
 
   <p align="center"><strong>A flat projection of a uniform latitude longitude grid.</strong></p>  
   
   <p>For
    later reference, the
    colored dots indicate a Moore-type neighborhood set.  The red dot is
    the center of the neighborhood.  The yellow dots are the neighbors of
    the center dot.  Every dot is the center of an associated Moore
    neighborhood.</p>
   

  <p> First we convert \((x_i,y_j)\) to spherical coordinates.
  \begin{eqnarray*}
  \phi_i & = & (90^\circ - x_i)\cdot \frac{\pi}{180^\circ} \\
  \theta_j & = & y_j \cdot \frac{\pi}{180^\circ}.
  \end{eqnarray*}
  Then,
  \begin{eqnarray*}
  A(x_i,y_j) & = & \int_{\theta_j}^{\theta_{j+1}}
\int_{\phi_i}^{\phi_{i+1}} R^2 \sin\phi \, d\phi d\theta\\
& = & R^2(\theta_{j+1} - \theta_j) (\cos(\phi_{i+1}) - \cos(\phi_i)).
  \end{eqnarray*}
  It follows immediately that the population of a particular grid square is
  well-approximated by \(\rho_{ij} A(x_i,y_j)\).
 

Evan Kleiner (Whitman College) and Albert Schueller (Whitman College), "A Political Redistricting Tool for the Rest of Us - The Population Density Function," Convergence (October 2013)