Why data retrieved with get_pvgis_hourly doesn't match with the ones from PV Performance Tool?

Question

I am interested into retrieving the value of the yearly in plane irradiation via code, given a database of parameters to be taken from the function get_pvgis_hourly, from this parser\getter, as follows:

 get_pvgis_hourly(45.858217, 12.267183, angle=7, aspect=-44, outputformat='csv',
              usehorizon=True, userhorizon=None, raddatabase="PVGIS-SARAH",
              startyear=None, endyear=None, pvcalculation=False,
              peakpower=1, pvtechchoice='crystSi',
              mountingplace='free', loss=14, trackingtype=0,
              optimal_inclination=False, optimalangles=False,
              components=False, url=URL, map_variables=True, timeout=30)[0].sum()

and the output is the following:

 poa_global         17993672.40
 solar_elevation     1489417.07
 temp_air            1417261.89
 wind_speed           213468.18
 Int                    2386.00
 dtype: float64

But if I use the samen data in PVGIS - PV Performance Tool like this

I obtain different data:

Any hint would be apreciated.

Answer 1

I suggest getting into the habit of examining the underlying data prior to summarizing it. Oftentimes you will find that the assumptions you had about the data don't hold. Printing and plotting the data are good places to start.

data, inputs, meta =  get_pvgis_hourly(45.858217, 12.267183, angle=7, aspect=-44, outputformat='csv',
                                       usehorizon=True, userhorizon=None, raddatabase="PVGIS-SARAH",
                                       startyear=None, endyear=None, pvcalculation=False,
                                       peakpower=1, pvtechchoice='crystSi',
                                       mountingplace='free', loss=14, trackingtype=0,
                                       optimal_inclination=False, optimalangles=False,
                                       components=False, url=URL, map_variables=True, timeout=30)

print(data['poa_global'])

time
2005-01-01 00:10:00+00:00    0.0
2005-01-01 01:10:00+00:00    0.0
2005-01-01 02:10:00+00:00    0.0
2005-01-01 03:10:00+00:00    0.0
2005-01-01 04:10:00+00:00    0.0

2016-12-31 19:10:00+00:00    0.0
2016-12-31 20:10:00+00:00    0.0
2016-12-31 21:10:00+00:00    0.0
2016-12-31 22:10:00+00:00    0.0
2016-12-31 23:10:00+00:00    0.0
Name: poa_global, Length: 105192, dtype: float64

data['poa_global'].plot()

This shows that data holds hourly values spanning 12 years, so when you calculate the sum of the entire thing, it is total insolation over 12 years. The PVGIS website divides by 12 to get average annual insolation. Finally, notice that there is a units difference (kWh/m^2 vs Wh/m^2), so dividing by 1000 gets things lining up:

In [25]: data['poa_global'].sum() / 12 / 1000
Out[25]: 1499.4727