Diffuse Irradiance (Part 3)

A project which has been going on for an embarrassingly long time requires the ability to take a weather forecast and turn it into a forecast of possible wind and solar energy.  The concept is that some software downloads a weather forecast (in this case a TAF) and estimates the amount of solar energy that a PV Panel might produce.  If overcast conditions are likely to prevail for the next 24 hours, it charges it's batteries using off-peak electricity.  This requires a simple cloud sky model which in turn requires a simple clear sky model.  The clear sky model base based on Sun-Earth geometry and gives an estimate of the solar irradiance from a sky with no clouds, the cloud sky model then reduces the clear sky irradiance according to the nature of the cloud cover.  At this stage, work on the cloud sky model  consists of grinding out the correlations.

After some experimenting, it was decided to use a correlation found in "Applied Solar Energy" by A.P. and M.P. Meinel which was published by Addison Wesley in 1976.  Despite being based on observations made in the Mojave Desert in the late 60s this formula provides reasonable estimates of Direct Normal Irradiance (DNI)  for several climate types.  It is a simple model, requiring only input being Air Mass.  What I could not find was an equivalent for formula for clear sky Direct Horizontal Irradiance (DHI).  I had gained a little experience of making solar irradiance measurements whilst cloud watching in the back yard.  This formed the basis for a crude prototype shaded radiometer, but having collected data for just over a year, it is still in its crude form, this is partly due to laziness, but mainly the risk of introducing inconsistencies into the results.

Observations are made on clear days, there is more data for moorings than afternoons.  Even though an English day might start with a clear sky, clouds start to form mid-morning.  I still need to collect some more readings when the air mass is in the range 1.5 to 3.0.  The graph below shows the results to date of spending several sunny mornings in the park or on the beach:

An important part of the observing routine is making notes on the state of the sky.  Diffuse irradiance increases significantly when conditions are hazy or cumulus cloud is forming/dispersing, observations made under these conditions were discarded from the correlation, they are the green series on the graph.  Excel's Solver was used to produce this correlation:

This formula should be treated with caution, if only because it has not been reviewed by anyone other than the author.  Also, the calibration of the radiometer needs to be reviewed.  In addition, the Sun-Earth geometry code has been revised an recently ported to Python and is still being tested.  Thus some form of revision is likely.

This the shaded radiometer in use:

I would have liked to made some observations whilst the dust from the Saharan Desert was airborne over the UK around April-03, at a guess this would have resulted in above average levels of diffuse irradiance.