Documentation

Some Possible Situations

In this imaginary coastal region, you take a visual bearing on the radio tower of 270°. At the same time, Gilligan's Lighthouse bears 0°. If you plot a 90°-270° line through the radio tower and a 0°-180° line through the lighthouse on your Mercator chart, the point at which the lines cross is a fix. Since you have used only two lines, however, its quality is questionable.

But wait; your port lookout says he took a bearing on Cape Jones of 300°. If that line exactly crosses the point of intersection of the first two lines, you will have a perfect fix.

Whoops. What happened? Is your lookout in error? Possibly, but perhaps one or both of your bearings was slightly in error. This happens all the time. Which point, 1, 2, or 3, is correct? As far as you know, they are all equally valid.

In practice, the little triangle is plotted, and the fix position is taken as either the center of the triangle or the vertex closest to a danger (like shoal water). If the triangle is large, the quality is reported as poor, or even as no fix. If a fourth line of bearing is available, it can be plotted to try to resolve the ambiguity. When all three lines appear to cross at exactly the same point, the quality is reported as excellent or perfect.

Notice that three lines resulted in three intersection points. Four lines would return six intersection points. This is a case of combinatorial counting. Each intersection corresponds to choosing two lines to intersect from among n lines.

The next time you traverse these straits, it is a very foggy morning. You can't see any landmarks, but luckily, your navigational radar is operating. Each of these landmarks has a good radar signature, so you're not worried. You get a range from the radio tower of 14 nautical miles and a range from the lighthouse of 15 nautical miles.

Now what? You took ranges from only two objects, and yet you have two possible positions. This ambiguity arises from the fact that circles can intersect twice.

Luckily, your radar watch reports that he has Cape Jones at 18 nautical miles. This should resolve everything.

You were lucky this time. The third range resolved the ambiguity and gave you an excellent fix. Three intersections practically coincide. Sometimes the ambiguity is resolved, but the fix is still poor because the three closest intersections form a sort of circular triangle.

Sometimes the third range only adds to the confusion, either by bisecting the original two choices, or by failing to intersect one or both of the other arcs at all. In general, when n arcs are used, 2x(n-choose-2) possible intersections result. In this example, it is easy to tell which ones are right.

Bearing lines and arcs can be combined. If instead of reporting a third range, your radar watch had reported a bearing from the radar tower of 20°, the ambiguity could also have been resolved. Note, however, that in practice, lines of bearing for navigational fixing should only be taken visually, except in desperation. A radar's beam width can be a degree or more, leading to uncertainty.

As you begin to wonder whether this manual plotting process could be automated, your first officer shows up on the bridge with a laptop and Mapping Toolbox™ software.

Using navfix

The navfix function can be used to determine the points of intersection among any number of lines and arcs. Be warned, however, that due to the combinatorial nature of this process, the computation time grows rapidly with the number of objects. To illustrate this function, assign positions to the landmarks. Point A, Cape Jones, is at (latA,lonA). Point B, the radio tower, is at (latB,lonB). Point C, Gilligan's Lighthouse, is at (latC,lonC).

For the bearing-lines-only example, the syntax is:

[latfix,lonfix] = navfix([latA latB latC],[lonA lonB lonC],...
                         [300 270 0])

This defines the three points and their bearings as taken from the ship. The outputs would look something like this, with actual numbers, of course:

latfix =

	latfix1       NaN          % A intersecting B
	latfix2       NaN          % A intersecting C
	latfix3       NaN          % B intersecting C

lonfix =

	lonfix1       NaN          % A intersecting B
	lonfix2       NaN          % A intersecting C
	lonfix3       NaN          % B intersecting C

Notice that these are two-column matrices. The second column consists of NaNs because it is used only for the two-intersection ambiguity associated with arcs.

For the range-arcs-only example, the syntax is

[latfix,lonfix] = navfix([latA latB latC],[lonA lonB lonC],...
                         [16 14 15],[0 0 0])

This defines the three points and their ranges as taken from the ship. The final argument indicates that the three cases are all ranges.

The outputs have the following form:

latfix =

	latfix11  latfix12          % A intersecting B
	latfix21  latfix22          % A intersecting C
	latfix31  latfix32          % B intersecting C

lonfix =

	lonfix11  lonfix12          % A intersecting B
	lonfix21  lonfix22          % A intersecting C
	lonfix31  lonfix32          % B intersecting C

Here, the second column is used, because each pair of arcs has two potential intersections.

For the bearings and ranges example, the syntax requires the final input to indicate which objects are lines of bearing (indicated with a 1) and which are range arcs (indicated with a 0):

[latfix,lonfix] = navfix([latB latB latC],[lonB lonB lonC],...
                         [20 14 15],[1 0 0])

The resulting output is mixed:

latfix =

	latfix11       NaN          % Line B intersecting Arc B
	latfix21  latfix22          % Line B intersecting Arc C
	latfix31  latfix32          % Arc B intersecting Arc C

lonfix =

	lonfix11       NaN          % Line B intersecting Arc B
	lonfix21  lonfix22          % Line B intersecting Arc C
	lonfix31  lonfix32          % Arc B intersecting Arc C

Only one intersection is returned for the line from B with the arc about B, since the line originates inside the circle and intersects it once. The same line intersects the other circle twice, and hence it returns two points. The two circles taken together also return two points.

Usually, you have an idea as to where you are before you take the fix. For example, you might have a dead reckoning position for the time of the fix (see below). If you provide navfix with this estimated position, it chooses from each pair of ambiguous intersections the point closest to the estimate. Here's what it might look like:

[latfix,lonfix] = navfix([latB latB latC],[lonB lonB lonC],...
                         [20 14 15],[1 0 0],drlat,drlon)

latfix =

	latfix11                    % the only point
	latfix21                    % the closer point
	latfix31                    % the closer point

lonfix =

	lonfix11                    % the only point
	lonfix21                    % the closer point
	lonfix31                    % the closer point

A Numerical Example of Using navfix