As mentioned, it’s when the direction passes through North, or zero degrees, that our simple method falls down. Converting degrees into radians also will not solve the problem (trust me I tried). Instead, we need to make an assumption on the directional behaviour.
If the direction has changed from 350 to 5 that’s a change of 15 degrees if, and only if, it goes clockwise. I.e. if the wind direction changes from 350, up to 351, 352 ….. 359, 0, 1 … 5 degrees. There is also the possibility that the wind direction could have changed from 350, down to 349, 348, …. 6, 5 degrees, or counter-clockwise. The two options are shown below.
Angle direction can change either counter-clockwise (red) or clockwise (blue). Image by Author — Not to scale.Whether both situations are actually possible will depend on type of data and granularity of data. For example, is it possible for a vessel heading change of 190 degrees in three seconds? Can extreme weather cause extreme wind direction change?
For the proposed solution to work, we must assume that it is not possible for a direction change of over 180 degrees in the given period. I.e. if the data is timeseries of one minute granularity, from minute 1 to minute 2 a genuine change of direction by over 180 degree is impossible.
If this does not work for your data set, exercise extreme caution! Consider any other data points you have that can prove or disprove this theory and, if possible, source higher frequency data so that this assumption may become valid.
This isn’t a perfect pythonic solution but for my use case it does the job. The snippet below uses the pandas libray, for which the documentation can be found here.
# find diff between min and max wind directiondf['delta'] = df.max_winddirection - df.min_winddirection# check for if difference minimun < max - 180, if so, use different equation to get degree change
df.loc[df.min_winddirection < df.max_winddirection - 180, 'delta'] = 360 - df.max_winddirection + df.min_winddirection
Take the 350 to 5 degree example.
- The code above first allocates all rows the basic delta equation “maximum angle— minimum angle”. This would apply 350 as delta for this row.
- The code checks to see if the criteria for these ‘through north’ cases is satisfied. This would check if: “minimum angle< maximum angle— 180”. In this case 5 is less than (350–180) so it is TRUE.
- The code applies a new equation for delta when TRUE as “360 — maximum angle + minimum angle”. Resulting in 15 degrees.
Where the ‘through north’ criteria is not satisfied, the basic maximum minus minimum formula will be applied with no issues.