Refactor crossings and lat/lon conversion functions

src/analysis/forecast_twl.py

@@ -3,11 +3,11 @@ from multiprocessing import Pool
 
 import click
 import numpy as np
-import numpy.ma as ma
 import pandas as pd
 from scipy import stats
 
 from analysis import runup_models
+from utils import crossings
 from logs import setup_logging
 
 logger = setup_logging()
@@ -270,35 +270,6 @@ def slope_from_profile(profile_x, profile_z, top_elevation, btm_elevation, metho
         return -slope
 
 
-def crossings(profile_x, profile_z, constant_z):
-    """
-    Finds the x coordinate of a z elevation for a beach profile. Much faster than using shapely to calculate
-    intersections since we are only interested in intersections of a constant value. Will return multiple
-    intersections if found. Used in calculating beach slope.
-    Adapted from https://stackoverflow.com/a/34745789
-    :param profile_x: List of x coordinates for the beach profile section
-    :param profile_z: List of z coordinates for the beach profile section
-    :param constant_z: Float of the elevation to find corresponding x coordinates
-    :return: List of x coordinates which correspond to the constant_z
-    """
-
-    # Remove nans to suppress warning messages
-    valid = ~ma.masked_invalid(profile_z).mask
-    profile_z = np.array(profile_z)[valid]
-    profile_x = np.array(profile_x)[valid]
-
-    # Normalize the 'signal' to zero.
-    # Use np.subtract rather than a list comprehension for performance reasons
-    z = np.subtract(profile_z, constant_z)
-
-    # Find all indices right before any crossing.
-    # TODO Sometimes this can give a runtime warning https://stackoverflow.com/a/36489085
-    indicies = np.where(z[:-1] * z[1:] < 0)[0]
-
-    # Use linear interpolation to find intersample crossings.
-    return [profile_x[i] - (profile_x[i] - profile_x[i + 1]) / (z[i] - z[i + 1]) * (z[i]) for i in indicies]
-
-
 @click.command()
 @click.option("--waves-csv", required=True, help="")
 @click.option("--tides-csv", required=True, help="")