Finding a (Buffon’s) needle in a hierarchical triangular haystack
Ashton Shortridge
NCGIA, Department of Geography
University of California
Santa Barbara, California
Email: ashton@geog.ucsb.edu
Link to full paper
This paper develops a method for determining the probability for a line to intersect
an equilateral triangular tile. Such tiles comprise global data structures like
the quaternary triangular mesh (QTM), and such intersections are fundamental spatial
operations. Knowledge of the probability of intersection has implications for
the analysis of data stored in global data structures.
There are a number of ways in which the surface earth can be decomposed into a
set of finite elements. One method, termed the quaternary triangular mesh, begins
by projecting the earth on to an octahedron. Each facet of the octahedron is an
equilateral triangle; level 0 in the hierarchy is the set of eight facets. Each
triangle can be subdivided into four equilateral triangles by connecting the center
points of each side; these triangles comprise level 2. Any triangle within this
system can be recursively decomposed to increasingly higher (finer) level levels
in the hierarchy.
The quaternary triangular mesh has several advantages as a method of structuring
global spatial information. Triangles at higher levels nest congruently within
coarser, lower level triangles. Data structure density may vary with data density,
since the number of levels can be extended in data-rich locations. The level of
a triangle is implicit in its index number, making it straightforward to determine
the precision of the data. Efficient spatial operators have been developed for
the QTM, making it a useful structure for very large data sets.
Intersection is a fundamental operation for any spatial data structure. Usually
intersection is calculated deterministically, since feature locations are typically
known prior to performing the operation. However, there is utility in knowing
the probability of intersection without reference to feature-specific locations.
Buffon's Needle-type solutions from the field of geometric probability provide
a framework for deriving probabilities for a number of common GIS intersections,
including line-line, line-square, and line-rectangle. Knowledge of intersection
probabilities may be valuable for applications involving global QTMs, including
estimating complexity and time required to process a distance operation and identifying
optimal resolutions in the hierarchy for specific operations.
This paper reports on the extension of Buffon’s Needle to the equilateral triangle
case. More formally, consider the earth’s surface modeled as a quaternary triangular
mesh at some level n. This surface consists entirely of a repeatable pattern of
nested equilateral triangular tiles, such that all locations fall in one and only
one tile. A needle -- a line segment of length l -- is dropped randomly onto the
surface. What is the probability that the needle intersects a tile boundary? The
solution for this problem is presented and its utility is discussed.