| Conditions | 4 |
| Total Lines | 148 |
| Code Lines | 85 |
| Lines | 0 |
| Ratio | 0 % |
| Changes | 0 | ||
Small methods make your code easier to understand, in particular if combined with a good name. Besides, if your method is small, finding a good name is usually much easier.
For example, if you find yourself adding comments to a method's body, this is usually a good sign to extract the commented part to a new method, and use the comment as a starting point when coming up with a good name for this new method.
Commonly applied refactorings include:
If many parameters/temporary variables are present:
| 1 | import numpy as np |
||
| 6 | def inverse(lats1: 'list', lons1: 'list', lats2: 'list', lons2: 'list') -> 'dict': |
||
| 7 | """ |
||
| 8 | Inverse geodesic using Vincenty approach. |
||
| 9 | |||
| 10 | Calculate the great circle distance between two points |
||
| 11 | on the earth (specified in decimal degrees) |
||
| 12 | |||
| 13 | All args must be of equal length. |
||
| 14 | |||
| 15 | Ref: |
||
| 16 | https://www.movable-type.co.uk/scripts/latlong-vincenty.html |
||
| 17 | """ |
||
| 18 | |||
| 19 | eps = np.finfo(float).eps |
||
| 20 | |||
| 21 | lon1, lon2 = map(wrap180deg, [np.array(lons1), np.array(lons2)]) |
||
| 22 | lat1, lat2 = map(wrap90deg, [np.array(lats1), np.array(lats2)]) |
||
| 23 | lon1, lat1, lon2, lat2 = map(np.radians, [lon1, lat1, lon2, lat2]) |
||
| 24 | |||
| 25 | # WGS84 |
||
| 26 | a = 6378137.0 |
||
| 27 | b = 6356752.314245 |
||
| 28 | f = 1.0 / 298.257223563 |
||
| 29 | |||
| 30 | # L = difference in longitude |
||
| 31 | L = lon2 - lon1 |
||
| 32 | |||
| 33 | # U = reduced latitude, defined by tan U = (1-f)·tanφ. |
||
| 34 | tanU1 = (1 - f) * np.tan(lat1) |
||
| 35 | cosU1 = 1 / np.sqrt((1 + tanU1 * tanU1)) |
||
| 36 | sinU1 = tanU1 * cosU1 |
||
| 37 | tanU2 = (1 - f) * np.tan(lat2) |
||
| 38 | cosU2 = 1 / np.sqrt((1 + tanU2 * tanU2)) |
||
| 39 | sinU2 = tanU2 * cosU2 |
||
| 40 | |||
| 41 | # checks for antipodal points |
||
| 42 | antipodal = (np.abs(L) > np.pi / 2) | (np.abs(lat2 - lat1) > np.pi / 2) |
||
| 43 | |||
| 44 | # delta = difference in longitude on an auxiliary sphere |
||
| 45 | delta = L |
||
| 46 | |||
| 47 | # sigma = angular distance P₁ P₂ on the sphere |
||
| 48 | sigma = np.zeros(lat1.shape) |
||
| 49 | sigma[antipodal] = np.pi |
||
| 50 | |||
| 51 | cossigma = np.ones(lat1.shape) |
||
| 52 | cossigma[antipodal] = -1 |
||
| 53 | |||
| 54 | # sigmam = angular distance on the sphere from the equator |
||
| 55 | # to the midpoint of the line |
||
| 56 | # azi = azimuth of the geodesic at the equator |
||
| 57 | deltaprime = np.zeros(lat1.shape) |
||
| 58 | iterations = 0 |
||
| 59 | |||
| 60 | # init before loop to allow mask indexing |
||
| 61 | sinSqsigma = np.zeros(lat1.shape) |
||
| 62 | sinsigma = np.zeros(lat1.shape) |
||
| 63 | cossigma = np.zeros(lat1.shape) |
||
| 64 | sinazi = np.zeros(lat1.shape) |
||
| 65 | cosSqazi = np.ones(lat1.shape) |
||
| 66 | cos2sigmam = np.ones(lat1.shape) |
||
| 67 | C = np.ones(lat1.shape) |
||
| 68 | |||
| 69 | # init mask |
||
| 70 | m = np.ones(lat1.shape, dtype=bool) |
||
| 71 | |||
| 72 | while (np.abs(delta[m] - deltaprime[m]) > 1e-12).any(): |
||
| 73 | sindelta = np.sin(delta) |
||
| 74 | cosdelta = np.cos(delta) |
||
| 75 | sinSqsigma = (cosU2 * sindelta) * (cosU2 * sindelta) + ( |
||
| 76 | cosU1 * sinU2 - sinU1 * cosU2 * cosdelta |
||
| 77 | ) * (cosU1 * sinU2 - sinU1 * cosU2 * cosdelta) |
||
| 78 | |||
| 79 | # co-incident/antipodal points mask - exclude from the rest of the loop |
||
| 80 | m = (np.abs(sinSqsigma) > eps) & (sinSqsigma != np.nan) |
||
| 81 | |||
| 82 | sinsigma[m] = np.sqrt(sinSqsigma[m]) |
||
| 83 | cossigma[m] = sinU1[m] * sinU2[m] + cosU1[m] * cosU2[m] * cosdelta[m] |
||
| 84 | sigma[m] = np.arctan2(sinsigma[m], cossigma[m]) |
||
| 85 | sinazi[m] = cosU1[m] * cosU2[m] * sindelta[m] / sinsigma[m] |
||
| 86 | cosSqazi[m] = 1 - sinazi[m] * sinazi[m] |
||
| 87 | |||
| 88 | # on equatorial line cos²azi = 0 |
||
| 89 | cos2sigmam[m] = cossigma[m] - 2 * sinU1[m] * sinU2[m] / cosSqazi[m] |
||
| 90 | cos2sigmam[cosSqazi == 0] = 0 |
||
| 91 | |||
| 92 | C[m] = f / 16 * cosSqazi[m] * (4 + f * (4 - 3 * cosSqazi[m])) |
||
| 93 | |||
| 94 | deltaprime = delta |
||
| 95 | delta = L + (1 - C) * f * sinazi * ( |
||
| 96 | sigma |
||
| 97 | + C |
||
| 98 | * sinsigma |
||
| 99 | * (cos2sigmam + C * cossigma * (-1 + 2 * cos2sigmam * cos2sigmam)) |
||
| 100 | ) |
||
| 101 | |||
| 102 | # Exceptions |
||
| 103 | iterationCheck = np.abs(delta) |
||
| 104 | iterationCheck[antipodal] = iterationCheck[antipodal] - np.pi |
||
| 105 | if (iterationCheck > np.pi).any(): |
||
| 106 | raise Exception('delta > np.pi') |
||
| 107 | |||
| 108 | iterations += 1 |
||
| 109 | if iterations >= 20: |
||
| 110 | raise Exception('Vincenty formula failed to converge') |
||
| 111 | |||
| 112 | uSq = cosSqazi * (a * a - b * b) / (b * b) |
||
| 113 | A = 1 + uSq / 16384 * (4096 + uSq * (-768 + uSq * (320 - 175 * uSq))) |
||
| 114 | B = uSq / 1024 * (256 + uSq * (-128 + uSq * (74 - 47 * uSq))) |
||
| 115 | dsigma = ( |
||
| 116 | B |
||
| 117 | * sinsigma |
||
| 118 | * ( |
||
| 119 | cos2sigmam |
||
| 120 | + B |
||
| 121 | / 4 |
||
| 122 | * ( |
||
| 123 | cossigma * (-1 + 2 * cos2sigmam * cos2sigmam) |
||
| 124 | - B |
||
| 125 | / 6 |
||
| 126 | * cos2sigmam |
||
| 127 | * (-3 + 4 * sinsigma * sinsigma) |
||
| 128 | * (-3 + 4 * cos2sigmam * cos2sigmam) |
||
| 129 | ) |
||
| 130 | ) |
||
| 131 | ) |
||
| 132 | # length of the geodesic |
||
| 133 | s12 = b * A * (sigma - dsigma) |
||
| 134 | |||
| 135 | # note special handling of exactly antipodal points where sin²sigma = 0 |
||
| 136 | # (due to discontinuity atan2(0, 0) = 0 but atan2(eps, 0) = np.pi/2 / 90°) |
||
| 137 | # in which case bearing is always meridional, due north (or due south!) |
||
| 138 | |||
| 139 | azi1 = np.arctan2(cosU2 * sindelta, cosU1 * sinU2 - sinU1 * cosU2 * cosdelta) |
||
| 140 | azi1[np.abs(sinSqsigma) < eps] = 0 |
||
| 141 | |||
| 142 | azi2 = np.arctan2(cosU1 * sindelta, -sinU1 * cosU2 + cosU1 * sinU2 * cosdelta) |
||
| 143 | azi2[np.abs(sinSqsigma) < eps] = np.pi |
||
| 144 | |||
| 145 | azi1 = wrap360deg(azi1 * 180 / np.pi) |
||
| 146 | azi2 = wrap360deg(azi2 * 180 / np.pi) |
||
| 147 | |||
| 148 | # if distance is too small |
||
| 149 | mask = s12 < eps |
||
| 150 | azi1[mask] = None |
||
| 151 | azi2[mask] = None |
||
| 152 | |||
| 153 | return {'s12': s12, 'azi1': azi1, 'azi2': azi2, 'iterations': iterations} |
||
| 154 | |||
| 257 |