sun.rise.set() accuracy and time zone handling

[clnsmth]

The accuracy of LakeMetabolizer::sun.rise.set() is relatively low with respect to suncalc::getSunlightTimes() (demonstrated below). Additionally, sun.rise.set() does not return values in the input tzone whereas getSunlightTimes() does. Do these discrepancies warrant changes to sun.rise.set()?

One solution is to nest sun.rise.set() in a clause controlled by a new method argument, which defaults to the current implementation but allows a switch to the alternative getSunlightTimes() when called upon. An additional ‘lon’ (longitude) argument would be added to meet getSunlightTimes() requirements. This solution would maintain backwards compatibility while offering improved accuracy and handling time zones if suncalc is installed (i.e. a suggested library, not a required one). I can submit these proposed changes in a pull request if you’d like.

Demo (script is located here):

Create test data (noaa = NOAA, lm = LakeMetabolizer, sc = suncalc)

> d <- create_test_data()
 [100%] Downloaded 750 bytes...
 [100%] Downloaded 827 bytes...
 [100%] Downloaded 742 bytes...
 [100%] Downloaded 824 bytes...
> head(d)
        date   dst           noaa_rise             lm_rise             sc_rise            noaa_set              lm_set              sc_set
1 2011-01-01 FALSE 2011-01-01 07:18:00 2011-01-01 07:29:44 2011-01-01 07:19:26 2011-01-01 16:30:00 2011-01-01 16:30:15 2011-01-01 16:30:56
2 2011-01-02 FALSE 2011-01-02 07:18:00 2011-01-02 07:29:22 2011-01-02 07:19:30 2011-01-02 16:31:00 2011-01-02 16:30:37 2011-01-02 16:31:46
3 2011-01-03 FALSE 2011-01-03 07:18:00 2011-01-03 07:28:59 2011-01-03 07:19:32 2011-01-03 16:32:00 2011-01-03 16:31:00 2011-01-03 16:32:39
4 2011-01-04 FALSE 2011-01-04 07:18:00 2011-01-04 07:28:33 2011-01-04 07:19:32 2011-01-04 16:33:00 2011-01-04 16:31:26 2011-01-04 16:33:33
5 2011-01-05 FALSE 2011-01-05 07:18:00 2011-01-05 07:28:05 2011-01-05 07:19:29 2011-01-05 16:34:00 2011-01-05 16:31:54 2011-01-05 16:34:28
6 2011-01-06 FALSE 2011-01-06 07:18:00 2011-01-06 07:27:35 2011-01-06 07:19:24 2011-01-06 16:35:00 2011-01-06 16:32:24 2011-01-06 16:35:26

LakeMetabolizer during standard time

> compare_to_noaa(
+   data = d, 
+   source = "LakeMetabolizer", 
+   daylight.savings = FALSE)
Sunrise difference:
  min: -30.63
  max: -0.22
  avg: -12.31
Sunset difference:
  min: -22.37
  max: 11.51
  avg: -1.58

LakeMetabolizer during daylight savings

> compare_to_noaa(
+   data = d, 
+   source = "LakeMetabolizer", 
+   daylight.savings = TRUE)
Sunrise difference:
  min: 29.59
  max: 55.42
  avg: 44.58
Sunset difference:
  min: 36.93
  max: 66.17
  avg: 53.75

suncalc during standard time

> compare_to_noaa(
+   data = d, 
+   source = "suncalc", 
+   daylight.savings = FALSE)
Sunrise difference:
  min: -1.98
  max: -0.67
  avg: -1.4
Sunset difference:
  min: -2.25
  max: -0.22
  avg: -1.12

suncalc during daylight savings

> compare_to_noaa(
+   data = d, 
+   source = "suncalc", 
+   daylight.savings = TRUE)
Sunrise difference:
  min: -2
  max: -0.55
  avg: -1.25
Sunset difference:
  min: -2.77
  max: -0.05
  avg: -1.42

Plot sun rise

Plot sun set

[jzwart]

Thanks @clnsmth , that seems like a good solution, and I'm pretty sure the current sun.rise.set() functions were handed down functions for earlier metabolism models. Having backwards capabilities will be important as changing the sunrise / sunset times will likely influence metab estimates. Where did you get the NOAA sunrise/sunset data?

[rBatt]

Are current SR SS estimates within an hour of truth for a given location?

On Tue, Sep 29, 2020 at 2:46 PM Jake Zwart notifications@github.com wrote:

Thanks @clnsmth https://github.com/clnsmth , that seems like a good solution, and I'm pretty sure the current sun.rise.set() functions were handed down functions for earlier metabolism models. Having backwards capabilities will be important as changing the sunrise / sunset times will likely influence metab estimates. Where did you get the NOAA sunrise/sunset data?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/GLEON/LakeMetabolizer/issues/138#issuecomment-700909619, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABK6W72NAYK7FA2NYNWUGRDSIITQDANCNFSM4Q67H6RQ .

-- Ryan D. Batt, Ph.D. Aquatic Ecologist turned Data Scientist/ Engineer Pronouns: he/ him

[jordansread]

Hi @rBatt ! 👋

[clnsmth]

Thanks @clnsmth , that seems like a good solution, and I'm pretty sure the current sun.rise.set() functions were handed down functions for earlier metabolism models. Having backwards capabilities will be important as changing the sunrise / sunset times will likely influence metab estimates. Where did you get the NOAA sunrise/sunset data?

https://www.esrl.noaa.gov/gmd/grad/solcalc/