Estimating Sunshine Hours using the Davis Weather Station

 Our Davis Weather Station (Vantage Pro 2 Plus) is equipped with both solar radiation and UV sensors. In principle, this should allow estimation of the number of sunshine hours over a specified period (e.g. per month, per year) using an empirical model derived from local data. Meteorological weather stations generally report sunshine hours as one of a suite of climatological data series.

I contacted my supplier, Prodata Weather Stations, to see if the manufacturer (Davis) offered this option via their software (WeatherLink). Apparently, they do but only using their Windows version of WeatherLink which I do not have. Furthermore, the Davis algorithm is relatively crude with a tendency to overestimate. Third-party software for Davis Weather Stations (e.g. CumulusMX) uses a more sophisticated method for estimating sunshine hours and maybe an option.

The standard instrumentation for measuring sunshine hours is the Campbell Stokes recorder, a real piece of art invented in the Victorian age.

The Ross-on-Wye weather station changed from the Campbell Stokes recorder to an automatic Kipp & Zonen sensor in 2019. For my purposes, the changeover came at the right time; I could just use the most recent K&Z results from Ross-on-Wye and not have to worry about converting CS sunshine hours to K&Z sunshine hours. The Davis solar radiation sensor works on the same principle as the K&Z sensor, as far as I can tell (both are pyranometers). Sunshine hours determined by the Campbell Stokes recorder and the Kipp & Zonen sensor are broadly comparable though the relationship is not simple.

The Davis radiation sensor measures the sun's power in W/㎡ and covers the visible and near infrared parts of the electromagnetic spectrum (400 to 1100 nm). The WeatherLink software records the 

daily average solar radiation which I have summed to give monthly solar radiation data for Hereford. These data are then plotted against the Ross-on-Wye sunshine hours (Kipp & Zonen) provided by the Met Office.

The correlation coefficient (0.862) indicates a good fit for a simple linear model to estimate sunshine hours in Hereford:

Monthly Sunshine Hours (Hereford) = 0.03*Total Monthly Solar Radiation (W/㎡) + 25

Interestingly, if UK sunshine hours (rather than Ross-on-Wye) are used then the fit (correlation) is better although the model (equation) is very similar.

The weather station also has a UV sensor monitoring the 280 to 360 nm region of the electromagnetic spectrum.  This covers both the UVA (400-320 nm) and UVB (320-280 nm) parts of the spectrum; some UVA is needed to produce Vitamin D while UVB is harmful (skin cancer). UV intensity is reported as daily average UV Index in the WeatherLink software - each day has a value of 0, 1 or 2. Daily values were summed to give monthly totals and plotted against the monthly sunshine hours for Ross-on-Wye.

Again, an excellent fit though there is a 'threshold' issue with winter months that record zero UV Index values but have non-zero sunshine hours.

In summary, I have a reasonable and simple linear regression model for estimating monthly sunshine hours using the daily solar radiation values recorded by the Davis Weather Station. In the future, I'll look at non-linear and multiple linear regression techniques to see if a better model is possible.