Author Topic: Can we estimate the number of sunshine hours using the solar radiation data?  (Read 2990 times)

0 Members and 1 Guest are viewing this topic.

Offline dean.martin

  • Senior Member
  • **
  • Posts: 77
Hi Weather Experts,

Can we estimate the number of sunshine hours using the solar radiation data say from a weather station?

I am thinking we can by assigning a minimum solar radiation, say 300 W/m^2 for example ,for the sunshine duration to be counted or recorded. I want to compare the estimated Sunshine Hours from this method to an actual Sunshine hours from a Campbell Stokes recorder to confirm but I don't have one.

Offline mcrossley

  • Forecaster
  • *****
  • Posts: 1132
    • Wilmslow Astro
If you have the "usual" amateur solar sensor (as Davis etc) you cannot use an absolute value, you have to compare the measured value against the predicted value. Pick some threshold - say 75 or 80% of predicted - and if the measured value exceeds this, count it as "Sun is shining". You will also need some minimum value to cut-off readings at the start/end of the day where the values are too small to be reliable - say 30-50 W/m2

Your figure of 300 W/m2 is oft misquoted, that is the recognised threshold for direct solar radiation. Sensors like the Davis measure global solar irradiation - very different.
Mark

Offline dean.martin

  • Senior Member
  • **
  • Posts: 77
If you have the "usual" amateur solar sensor (as Davis etc) you cannot use an absolute value, you have to compare the measured value against the predicted value. Pick some threshold - say 75 or 80% of predicted - and if the measured value exceeds this, count it as "Sun is shining". You will also need some minimum value to cut-off readings at the start/end of the day where the values are too small to be reliable - say 30-50 W/m2

Your figure of 300 W/m2 is oft misquoted, that is the recognised threshold for direct solar radiation. Sensors like the Davis measure global solar irradiation - very different.

What do you mean by predicted value? The Weather Station I am using is Davis Vantage Pro 2 plus.

Offline mcrossley

  • Forecaster
  • *****
  • Posts: 1132
    • Wilmslow Astro
You have to calculate the predicted Solar irradiation value using one of the documented methods for the lat/long & date/time of each solar reading and then compare them.

The "solrad" spreadsheet implements a few different prediction algorithms: https://rredc.nrel.gov/solar/codesandalgorithms/links.html

Most weather programs will do this for you and provide a sunshine hour figure based on the comparison.
Mark

Offline PaulMy

  • Forecaster
  • *****
  • Posts: 5509
    • KomokaWeather
I don't have the Campbell-Stokes but do have the Blake-Larsen http://www.sunrecorder.net/the-history/

The attached screen shot compares the Davis VP2 Plus as calculated by Cumulus (using default settings so no tweaking done) vs the B-L.
I have compared the log hours from Weatherlink, CumulusMX and Blake-Larsen over the past 15 days:
WL CumMX B-L

11.0    8.8    9.15
  9.5    3.8    5.30
  2.8    0.0    0.00
  5.5    0.1    0.00
  7.7    0.0    0.10
10.8    5.2    7.54
  1.2    0.6    0.00
  9.2    7.1    7.84
  0.7    0.1    0.00
10.8    9.4  10.14
  9.3    2.7    2.94
  7.2    0.6    1.38
10.7    2.7    2.94
  9.0    2.8    3.22
10.0    5.0    5.63

The Weatherlink seems to be way off, and that could be because I run Weatherlink only periodically, every few days, and the extraction from the logger could throw the calculation off.

CumulusMX Follows the B-L reasonably close but shows sunshine hours when there were none.
The B-L is quite precise as can see the sunshine off or on in a matter of seconds as the sky changes.

Our local airport is one of the location where Environment Canada no longer records bright sunshine as they are not replacing aging equipment (Campbell-Stokes).


Enjoy,
Paul

Online broadstairs

  • Forecaster
  • *****
  • Posts: 853
Weather Display will do sunshine hours for you using the Davis solar sensor. It does this using the method mentioned earlier. I use it and it seems reasonably accurate considering the deficiencies of the Davis unit and the calculation used. It is unlikely to be as accurate as measuring direct radiation but it is the best we can do.

Stuart
Ecowitt GW1003 with ultrasonic wind gauge, lightning sensor and PM2.5 sensor with Personal Weather Tablet as a console.

Offline dean.martin

  • Senior Member
  • **
  • Posts: 77
Thank you for all your answers. I just notice the Weatherlink can also calculate or estimate the "Bright Sunshine Hours". It's under the "Reports" tab of the weatherlink software (I am using WeatherLink 6.0.5). And found that it is just using similar method I mentioned and asked. It is solely based on a threshold value of solar radiation with a default value of 100 W/m2 (can be changed base on the user preference). Is this safe to use?

Offline PaulMy

  • Forecaster
  • *****
  • Posts: 5509
    • KomokaWeather
My post above gives the comparison of the Davis VP2 Plus with Weatherlink (from the report you noted) and an actual sunshine sensor.


Quite a large variance, and showing sunshine when there wasn't any, but as I mentioned that could be because I don't run WL full time so the WL report would be based on the logger download data at 10 minute interval whenever I run WL.




Enjoy,
Paul

Offline dean.martin

  • Senior Member
  • **
  • Posts: 77
My post above gives the comparison of the Davis VP2 Plus with Weatherlink (from the report you noted) and an actual sunshine sensor.


Quite a large variance, and showing sunshine when there wasn't any, but as I mentioned that could be because I don't run WL full time so the WL report would be based on the logger download data at 10 minute interval whenever I run WL.




Enjoy,
Paul


What Solar Radiation Threshold Value did you use?

Offline Mattk

  • Forecaster
  • *****
  • Posts: 2135
Have done this over quite a number of years to determine # sunshine days which is a little different to sunshine hours and it involves adopting W/m2 value in conjunction with the total number records and some analysis relative to factors, rainfall and what makes sunshine days a little more tricky is the different day lengths and solar intensity which is where the magic factor is computed for each record. When compared year by year over quite a few years the end result trends quite well with what might be expected.

At first when asked to come up with this it sounded quite simple but it was not as simple as initially thought but in the end gave some quite consistent and quantitative results but the analysis code is a lot more complicated than what a spread sheet can handle.       

Offline PaulMy

  • Forecaster
  • *****
  • Posts: 5509
    • KomokaWeather
I did not set anything in Weatherlink, just the default.


Enjoy,
Paul

Offline Mattk

  • Forecaster
  • *****
  • Posts: 2135
Having 10 minute archives is fairly typical which probably achieves the majority/bulk of the result over a day, 1 minute archive would not improve the end result that much apart from maybe fine tuning the beginning and ending periods which would attune to using 98% of the additional data to achieve maybe a 2% improvement?

Offline mcrossley

  • Forecaster
  • *****
  • Posts: 1132
    • Wilmslow Astro
I just notice the Weatherlink can also calculate or estimate the "Bright Sunshine Hours". It's under the "Reports" tab of the weatherlink software (I am using WeatherLink 6.0.5). And found that it is just using similar method I mentioned and asked. It is solely based on a threshold value of solar radiation with a default value of 100 W/m2 (can be changed base on the user preference)
No it doesn't. It may have a default minimum threshold of 100, but that isn't how it will be calculating the value. I don't know how it does calculate it, but will be using a comparison method like I outlined.

Oh dear, how wrong can you be! Any confidence I had in Davis is shaken.  ](*,)
« Last Edit: May 01, 2019, 04:13:41 AM by mcrossley »
Mark

Online broadstairs

  • Forecaster
  • *****
  • Posts: 853
I just notice the Weatherlink can also calculate or estimate the "Bright Sunshine Hours". It's under the "Reports" tab of the weatherlink software (I am using WeatherLink 6.0.5). And found that it is just using similar method I mentioned and asked. It is solely based on a threshold value of solar radiation with a default value of 100 W/m2 (can be changed base on the user preference)
No it doesn't. It may have a default minimum threshold of 100, but that isn't how it will be calculating the value. I don't know how it does calculate it, but will be using a comparison method like I outlined.
Well said Mark. There is so much rubbish talked about this 100w/m. I wrote a php script to do the comparison some years ago and Brian had a copy to assist him in getting Weather Display working better in this area. It is quite complicated but does an OK job, but will never be as accurate as either a Campbell Stokes or a sensor which measures direct solar radiation.

Stuart
Ecowitt GW1003 with ultrasonic wind gauge, lightning sensor and PM2.5 sensor with Personal Weather Tablet as a console.

Offline Mattk

  • Forecaster
  • *****
  • Posts: 2135
Well I hate to burst anybody's bubble on this BUT since WeatherLink 5.7.1 Hours of Bright Sunshine in Davis terms is simply the amount of time the Solar Radiation Threshold (W/m2) is above a user defined integer (Solar Rad.) value, nothing  more or nothing less. 0 (zero) W/m2 will give the exact time difference between the beginning of the first non zero value of the day and the end of the last non zero value of the day. There is nothing magic about this it is very simplistic       

Offline mcrossley

  • Forecaster
  • *****
  • Posts: 1132
    • Wilmslow Astro
Well I hate to burst anybody's bubble on this BUT since WeatherLink 5.7.1 Hours of Bright Sunshine in Davis terms is simply the amount of time the Solar Radiation Threshold (W/m2) is above a user defined integer (Solar Rad.) value, nothing  more or nothing less. 0 (zero) W/m2 will give the exact time difference between the beginning of the first non zero value of the day and the end of the last non zero value of the day. There is nothing magic about this it is very simplistic     
Well my flabber is well and truly gasted! So it is completely useless then.
Mark

Offline PaulMy

  • Forecaster
  • *****
  • Posts: 5509
    • KomokaWeather
Quote
Thank you for all your answers. I just notice the Weatherlink can also calculate or estimate the "Bright Sunshine Hours". It's under the "Reports" tab of the weatherlink software (I am using WeatherLink 6.0.5). And found that it is just using similar method I mentioned and asked. It is solely based on a threshold value of solar radiation with a default value of 100 W/m2 (can be changed base on the user preference). Is this safe to use?
Yesterday's comparison:
For the first time in the nearly 8 years of the B-L sunrecorder it recorded a total of 1 second of sunshine yesterday.
The log shows SUN 08:20:53  08:20:54 ( time-on  time-off)
There is a second line SUN 08:20:56  08:20:56 which did not get registered as sunshine.
Quote
LAT  42.95394
LON  81.43507
TZ  -4
DST True
SRS 06:22:48 20:24:08
TOT 00:00:04 00:00:00
SUN 08:20:53 08:20:54
SUN 08:20:56 08:20:56
SUN 08:20:58 08:20:58
NRC 20:24:08 20:24:08

I don't actually recall seeing that 1 second of sunshine though  :o

Weatherlink 6.0.2 shows 6.8 hrs for yesterday, so not in the same world... (Threshhold is at at the default 100 W/m2)
CumulusMX calculated 1.4 hrs so much better than WL.


Enjoy,
Paul 

Offline mcrossley

  • Forecaster
  • *****
  • Posts: 1132
    • Wilmslow Astro
Sounds like you need to tune your Cumulus parameters Paul ;)
Mark

Offline PaulMy

  • Forecaster
  • *****
  • Posts: 5509
    • KomokaWeather
Yes, thanks Mark for nudging...  I'll relook at https://cumulus.hosiene.co.uk/viewtopic.php?f=27&t=17207#p132755 and begin some changes.

To help in answering the OP question I'll also make some changes in the Weatherlink settings and see if that would bring it more realistic compared to the B-L.

Enjoy,
Paul

Offline PaulMy

  • Forecaster
  • *****
  • Posts: 5509
    • KomokaWeather
Quote
To help in answering the OP question I'll also make some changes in the Weatherlink settings and see if that would bring it more realistic compared to the B-L.
I just noticed that the sunshine hours in Weatherlink are not logged but calculated when selecting the sunshine report.  So I can change the threshold to whatever and see the recalculated results for the last 15 days.   I presume this threshold setting in the Reports page is the only place in WL where it can be set?


My original default setting in WL was 100 W/m2.  By increasing it the sunshine hours are reduced.  It took to a setting of 250 W/M2 to get WL to be at zero on the days that the B-L was at zero.  A setting of 300 W/m2 or higher was needed to bring the WL sunshine hours to somewhat realistic with the B-L.

Enjoy,
Paul

Offline mcrossley

  • Forecaster
  • *****
  • Posts: 1132
    • Wilmslow Astro

My original default setting in WL was 100 W/m2.  By increasing it the sunshine hours are reduced.  It took to a setting of 250 W/M2 to get WL to be at zero on the days that the B-L was at zero.  A setting of 300 W/m2 or higher was needed to bring the WL sunshine hours to somewhat realistic with the B-L.

As I said Paul a fixed threshold is useless. In fact it is worse than useless, it implies that it supplies some meaningful data when it doesn't.
Mark

Offline Mattk

  • Forecaster
  • *****
  • Posts: 2135
WL sunshine hours works very regular based solely on the W/m2 threshold, 0 W/m2 gives exactly the time difference that one can determine based on the archive intervals, use the max value (or greater) for the day and presto zero hours, average the W/m2 values for a day and presto exactly half the total hours for the day.

In a relative sense it has some meaning as using whatever W/m2 value is comparable between days. Never been sure why it only reports 15 days? Maybe a predecessor to subscription  :lol: :roll:     
« Last Edit: April 30, 2019, 04:50:33 PM by Mattk »