wviewweather.com

bodemory · Joined: 06 Dec 2005 Posts: 43

Hi,

UV and SR graphs are plotted according daily/weekly temporal averages.
It would be more interesting to plot, for every day/month the maxima data only, that appear in the monthly/yearly view. Averaging over time (including night...) is not accurate and unfortunately makes the graphs useless.

Thanks,
Brice

mteel · Joined: 30 Jun 2005 Posts: 435 Location: Collinsville, TX

Well I (unfortunately) don't agree they are useless.

bodemory · Joined: 06 Dec 2005 Posts: 43

I'd be very interested to know how do you get information from
those graphs then...
Solar radiation is based on sky transparency, latitude and epoch of the year. For a given latitude, along the year or the month, you would see a variation proportional to the sun height above horizon squared.
If you average over 24 hours including night, you put an additional variating quantity that is length of day, leading to a quantity that does not allow
inter day/month comparison.

The best proof I can give is the shape of your yearly UV/SR data...

I don't agree with your statement, but I do respect it Wink

Brice
Geneva Observatory, Switzerland.

mteel · Joined: 30 Jun 2005 Posts: 435 Location: Collinsville, TX

I have looked at my radiation graph for year - it shows a distinct pattern which tells me the total radiation for each day is decreasing uniformly to some point during our winter, and apparently will increase uniformly to some point in our summer. Is that true for all locations around the world? I don't believe so, thus the graph provides meaningful data.

I also like simple to understand charts. All are sample average based except for rain and ET since they must be considered as cumulative values.

I took your earlier comment to be directed at all graphs and the usefulness of hour (for monthly charts) and day (for yearly charts) averages. If we are restricting the conversation to radiation and UV, then I am less indignant Wink

bodemory · Joined: 06 Dec 2005 Posts: 43

For the maxima, you're right, it would be better for the global shape of the dataset but this is not the aim here.
So let's move to my main concern :

Let me explain it differently :
Suppose you're gathering SR data over 24h. You'll have (assuming you're averaging this way):
sum(SR0+SR1+...+SR23)/24, right ?
A better way to do (more meaningful from my point of view) would be :
sum(SR0+SR1+...+SR23)/int(sunrise-sunset)
and then estimate monthly and yearly averages this way.
This way you're ensuring that only meaningful data are taken in account,
by meaningful I mean only hours during the sun is up.

Today, you underestimate the mean by dividing your sum by day+night hours.

The problem is that int(sunrise-sunset) is not constant over the year, so
by calculating the averages this way, they are biased. This is especially true for the yearly average (depending on the latitude though!)

Brice

mteel · Joined: 30 Jun 2005 Posts: 435 Location: Collinsville, TX

First, that is not what you were first proposing.

This is, simply stated, a request to change the averaging period for solar-based daily measurements to be daylight hours only. N'est pas?

That (if that does summarize your new request) is a reasonable request, although I still do not see the overriding value in it. Do you look at a yearly graph to determine the absolute magnitude of a reading for a given day? Boy I don't. It provides a way to observe trends throughout the year. And as such, whether the daily solar radiation average is for daylight hours only or all hours is insignificant. Think of the average as a way to scale the TOTAL radiation for a day. Total doesn't care about daylight hours. Using the average instead of the cumulative is just a way to scale it to a more meaningful range.