fix linke turbidity time interpolation

[mikofski]

As indicated in in clearsky.lookup_linke_turbidity on L212-224

This is approximate and could be made more accurate.

>>> import numpy as np
>>> np.linspace(-15, 380, num=14)
array([ -15.        ,   15.38461538,   45.76923077,   76.15384615,
        106.53846154,  136.92307692,  167.30769231,  197.69230769,
        228.07692308,  258.46153846,  288.84615385,  319.23076923,
        349.61538462,  380.        ])

Another method would be to use the calendar month days:

>>> import calendar
>>> import numpy as np
>>> mdays = calendar.mdays[1:]
>>> np.concatenate([[-calendar.mdays[1] / 2.0],
                    np.cumsum(mdays) - np.array(mdays) / 2.,
                    [365 + calendar.mdays[-1] / 2.0]])
array([ -15.5,   15.5,   45. ,   74.5,  105. ,  135.5,  166. ,  196.5,
        227.5,  258. ,  288.5,  319. ,  349.5,  380.5])

This still doesn't account for leap years, but we can use calendar.isleap(<year>):

import calendar
import numpy as np


def calendar_month_middles(year):
    mdays = np.array(calendar.mdays[1:])  # copy
    ydays = 365
    if calendar.isleap(year):
        mdays[1] = mdays[1] + 1
        ydays = 366
    return np.concatenate([[-calendar.mdays[-1] / 2.0],
                           np.cumsum(mdays) - np.array(mdays) / 2.,
                           [ydays + calendar.mdays[1] / 2.0]])

now you can see the difference in leap year

>>> calendar_month_middles(2016)  # a leap year
array([ -15.5,   15.5,   45.5,   75.5,  106. ,  136.5,  167. ,  197.5,
        228.5,  259. ,  289.5,  320. ,  350.5,  381.5])
>>> calendar_month_middles(2015)  # NOT a leap year
array([ -15.5,   15.5,   45. ,   74.5,  105. ,  135.5,  166. ,  196.5,
        227.5,  258. ,  288.5,  319. ,  349.5,  380.5])

Note: it's important to copy mdays because it's a borrowed reference, so if you change it, the change is permanent.

>>> calendar.mdays
[0, 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31]
>>> calendar.mdays[3] = 100
>>> calendar.mdays
[0, 31, 28, 100, 30, 31, 30, 31, 31, 30, 31, 30, 31]
calendar_month_middles(2015)  # yikes!
array([ -15.5,   15.5,   45. ,  109. ,  174. ,  204.5,  235. ,  265.5,
        296.5,  327. ,  357.5,  388. ,  418.5,  380.5])

[wholmgren]

I'm guessing that the biggest practical impact of this will be that people see an example of the calendar module, but sounds good to me!

[mikofski]

Do you want me to add this to pull request #264 ?

Also I forgot to change 365 to 366 in calendar_month_middles()

[wholmgren]

Sure. It might make sense to do it all at once if it's going to change the
unit test expected values.
On Wed, Nov 16, 2016 at 8:50 AM Mark Mikofski notifications@github.com
wrote:

Do you want me to add this to pull request #264
#264 ?

Also I forgot to change 365 to 366 in calendar_month_middles()

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
#265 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AELiRyE-vtSDIc7XqppD0TMh1p_NAwetks5q-yYtgaJpZM4KzLt3
.