A nice result on sunshine computation

I am both pleased and perplexed by a result which raises more questions.

There are 4 traces on this plot.

This is the Chilbolton data shown a few days ago but now I have worked out more of the insolation computation. If you look at the type CNR4 pyranometer specification this is within accuracy. Water vapour value from a specialist instrument is shown as a guide to humidity change, not used in the computation.

A good result, a tribute to the authors of Solpos and those creating the algorithms used.

I applied what was learnt to the Koorin data expecting Northern Territories Australia from 40 years ago with older equipment to be in considerable error. Surprisingly not so…

Less than 4% error at peak insolation. An instrument offset of 7 W/sqm was present here too and removed first. No temperature or air pressure profile data is available, which might improve the match slightly.

Then there is the matter of the solar constant, fixed for Solpos at 1367 W/sqm but I can alter it. Was it the same 1974 and 2013? Recomputing for Koorin using 1362 does improve the match slightly, ie. proof this is of significance.

The Kt function

Solpos references: –

RENORMALIZATION OF KT (“PRIME”)
Perez, R., P. Ineichen, Seals, R., & Zelenka, A. 1990. Making full use of the clearness index for parameterizing hourly insolation conditions.
Solar Energy 45 (2), 111-114.

PDF scan of paper (4 pages, 3.2MB)
Whole reference page at Albany Uni. here

This partially solved some of the riddle on the meaning of output parameters prime and unprime where normalization is the context. I guessed that this had not been applied to the extra-terrestrial figure. I applied this as etr / prime and to my surprise corrected the insolation shape.

Scaling the serpent

With the shape corrected there comes a surprise, a linear scaling factor is sufficient to produce a remarkable match to measured data under particularly clear sky conditions. This seems almost free of effects from atmospheric changes during the day as humidity burns out etc.

I used a hand produced 0.789 for best least squares fit on Chilbolton data. This means that an additional 21% of solar insolation is lost.

Chilbolton conditions

Here is additional information. More could be accessed. Of the data I have this is the most perfect day. A slight day wind blows up. under anticyclonic conditions.

 

Comment

Just two results is grossly inadequate for certainty but this does look promising as a starting point.

Being able to dead reckon on exact time and location is remarkable.

I expect all this is well known and my struggling would cause amusement to those who deal with this daily.

 

---

 

Notes.

An A/D converter or ground loop offset was present, 4W sqm, subtracted out, the Koorin case shows a 7W offset, dark is dark, zero.

I also had problems with exact timing a 3+ minute time offset. This turned out to be a side effect of the data decimation using my own filter software, also applied to the time data. Correctly this centres and hence a time offset appeared. CNR4 data has 86k points per day which has to be matched to the web version of instrument data digitised and plotted at a very different rate. There is a small time delay in the pyranometer anyway.

Correcting from approximate to exact Google Earth Chilboton instrument location, WGS84 referenced also improved results.

Simple things of mice and men. This work is horribly complicated.

Post by Tim