# Transect lines, parallel lines, offset lines

Blog Post created by Dan_Patterson on Jan 16, 2019

Lines

Different incarnations and names

Pretty easy to form the origin-destination pairs.

Start at a point.

Throw in horizontal and/or vertical offsets.

A dash of an azimuth/bearing.

A bit of Arcpy and.... A good way to spend some time, so you write it down because you will forget and reinvent it later.

Almost forgot...

There is always one student that thinks outside the box.

Hmmmm could be a bonus here... I wonder if any of mine can replicate the compass with 10 degree increments? In the attached code, I made these changes

``    rads = np.deg2rad(bearing)    dx = np.sin(rads) * dist    dy = np.cos(rads) * dist    #    n = len(bearing)    N = [N, n][n>1]  # either the number of lines or bearings``

And used this

``b = np.arange(0, 361, 22.5)a, data =transect_lines(N=1, orig=[some x, some y],                        dist=100, x_offset=0, y_offset=0,                        bearing=b, as_ndarray=True)``

You can't have it both ways in a manner of speaking.  By limiting N to number of bearings, you use numpy to generate the desired angles,.  There is no x or y offset since the origin is now fixed.

How to use the attached...

``"""  ---- use these as your inputs, with edits of course# ---- make the x, y coordinate tableSR = 2951  # a projected coordinate system preferablya, data =transect_lines(N=10, orig=[299000, 5000000], dist=100,                        x_offset=10, y_offset=0, bearing=-10, as_ndarray=True)p0 = r"C:\Your_path\Your.gdb\a_tbl"arcpy.da.NumPyArrayToTable(a, p0)# ---- now for the linesp1 = r"C:\Your_path\Your.gdb\some_lines"arcpy.XYToLine_management(p0, p1,                          'X_from', 'Y_from',                          'X_to', 'Y_to',                          spatial_reference=SR)"""``

PS

The python/numpy part is quite speedy, using variants of

%timeit transect_lines(N=10, orig=[0,0], dist=1, x_offset=0, y_offset=0, bearing=0, as_ndarray=True)

That is microseconds for the speed geeks.  I couldn't see a use case to test for larger arrays.
N    Time
10     36.0 µs ± 309 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
50     39.3 µs ± 3.4 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
100     42.9 µs ± 6.57 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
500     46.5 µs ± 502 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
1000     54.9 µs ± 1.39 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
I didn't bother to test the featureclass creation since I have no control over that.