Build Your Own Error Bars — Sunshine
How much sunshine reached the ground yesterday? An instrument can tell you — where one exists. Everywhere else, the number you'll be quoted comes from a model. Here you measure the gap between instrument and model yourself — then test whether the gap you measured in one place tells you anything about another.
Pick a site with a long pyranometer record. Put its measurements next to a reanalysis model's estimate for the same spot. Build the distribution of their disagreements, freeze a band around it, and carry that band to a second site to see how it holds up. There is a sibling lab that runs the same idea on temperature instead of sunshine — see Build Your Own Error Bars — Temperature (Lab 05).
- What a pyranometer measures, and where modelled radiation estimates come from.
- How to turn two overlapping series into a difference distribution and an empirical error band.
- How to test — rather than assume — whether an error band built in one place transfers to another.
What's being compared
For every site in this lab there are two daily solar-radiation series. One is measured: the total solar energy reaching a horizontal surface each day, recorded by a pyranometer and published by the national weather service that operates it. The other is a reanalysis estimate — a weather model's reconstruction of the same quantity at the same coordinates, from the ERA5 dataset.
Both series claim to describe the same sky over the same days. This lab never treats one as the answer key for the other; it only measures how much, and when, they disagree — and puts that measurement in your hands.
Choose a calibration site
This is where you will build your error band. The sites run from a remote mountain summit to a capital-city rooftop — built-up context is shown on each card so you can consider it as part of your choice rather than having it hidden. Each comparison starts when that site's instrument record does.
Sites are shown in a random order to avoid nudging your choice.
Select a site above to continue.
Look at the two series together
Both series are drawn with equal weight — which one gets which colour, and which is listed first, is randomized on each visit, so neither is presented as the baseline the other is checked against. Change the averaging to see how much the comparison depends on it: individual days, monthly means, or whole years. When you compare by day or by month, a second control picks which calendar month — so like is always compared with like, rather than mixing times of year that can disagree with the model by very different amounts.
The starting averaging level and time-of-year filter are randomized on each visit; both are yours to change at any time.
Loading…
Construct the difference
So far you've seen the two series side by side. The next step subtracts one from the other, period by period — measured minus modelled — and collects every one of those differences into a distribution you can put a band around.
Subtracts the model's estimate from the instrument's reading at your chosen averaging.
The difference, period by period
Each point is the instrument's reading minus the model's estimate for the same period. Points above zero are periods where the instrument recorded more solar energy than the model estimates; points below zero, less. The averaging control above still applies.
Your difference distribution
The same differences, tallied — for the month or year the controls above select. The shaded band marks the central share of them — you pick how central. A 50% band that misses half the points is behaving exactly as labelled; the label is the claim being tested later, so it stays on screen.
The starting coverage level is randomized on each visit; it's yours to change at any time.
Freezing locks the band so you can test it at another site.
Test your band somewhere else
Every time a model output stands in for a missing measurement, an assumption is being made: that the model's error where it can be checked resembles its error where it can't. Here you get to check it. Pick another site; your frozen band is drawn around the reanalysis series there, with the site's own measurements hidden until you choose to reveal them.
Sites are shown in a random order to avoid nudging your choice.
Select a site above to draw your band there.
Loading…
Pick a site above first.
Your scorecard
Each site you reveal adds a row: how many of its measurements your band caught, against the share the band held at your calibration site. Built-up land around each site is listed so you can look for any relation to built-up context yourself — in either direction.
| Site | BU in 1 km | Inside band | Share inside | Band claimed |
|---|