Smart ways of monitoring solar power

Solar power is a major source of renewable energy. In recent years the installation of solar panels has boomed, while prices have fallen sharply. To implement energy transition policy effectively, the government needs reliable production estimates for the total amount of solar power generated in the Netherlands. The same applies to parties in the energy sector tasked with balancing the supply of – and demand for – electricity. This article gives details of a study being carried out by Statistics Netherlands (CBS) to facilitate more accurate and detailed estimates of solar power production. This study focuses on two elements in particular: establishing the location of each individual solar panel and determining how much energy these panels are effectively generating.

Energy transition at national and regional level

Energy transition is high on the political agenda. At the end of February 2018, the Minister of Economic Affairs and Climate Policy set out the government’s approach to climate policy in a letter to the House of Representatives. The government’s goal is to cut greenhouse gas emissions by 49% (relative to 1990 emission levels) by 2030. The regions (local authorities, private citizens, businesses) have an important part to play in achieving the climate goals. CBS is studying new sources and methods that could be used to more effectively monitor the Netherlands’ energy transition. In this way, CBS wants to help make the energy transition measurable, both at regional and at national level.

Current estimate based only on annual national figures

CBS’ current statistical analysis of solar power is based on a survey of approximately 350 companies that import and supply solar panels. Calculations of total solar power production are based on an estimate of installed capacity (installed panels) and on a fixed production value (875 kWh/kWp) per unit of installed capacity. This method involves an estimated uncertainty of 20 percent and can only provide yearly national figures. However, the energy transition is generating a demand for information at regional level and on shorter time scales. To this end, CBS is developing new methods for monitoring the amount of solar power being generated in greater detail.

Measuring solar power at regional level

It is vital to get a complete picture of the total number of solar panels that have been installed, together with details of their location. Only then will it be possible to establish information resources at regional level. CBS has a range of existing administrative sources that, in combination with new sources, can generate detailed information on the locations and capacities of solar panels. A new source for CBS is the Production Installation Registry (PIR), which contains details of all solar panels in the Netherlands. We received this source from Energie Data Services Nederland (EDSN) and have combined it with existing data from the tax authorities about households reclaiming the VAT they paid on their solar panels. In addition, we were already making use of data on households’ annual electricity consumption and the Addresses and Buildings (BAG) registry.

Growth in the number of domestic solar installations

We can use data from the PIR to illustrate the growth in the number of solar installations. The figures below illustrate the spectacular growth in solar panels since 2010, both in absolute numbers and at regional level. Note: this is restricted to small-scale consumption. Accordingly, it does not reflect the overall increase in the number of installations in the Netherlands, nor indeed the overall growth in the amount of solar power generated throughout the country.

plaatje van Nederland over zonnepanalen

Number of newly installed solar panels
 Newly installed solar panels
2010 j261
2010 f116
2010 m182
2010 a223
2010 m197
2010 j364
2010 j310
2010 a235
2010 s293
2010 o400
2010 n299
2010 d288
2011 j400
2011 f410
2011 m521
2011 a693
2011 m735
2011 j868
2011 j870
2011 a541
2011 s731
2011 o577
2011 n695
2011 d669
2012 j723
2012 f720
2012 m1264
2012 a1434
2012 m1923
2012 j2268
2012 j3845
2012 a4614
2012 s6334
2012 o6672
2012 n5645
2012 d4296
2013 j3986
2013 f4398
2013 m6019
2013 a7014
2013 m7743
2013 j8585
2013 j8144
2013 a6388
2013 s6113
2013 o4457
2013 n4322
2013 d4334
2014 j4215
2014 f4524
2014 m5617
2014 a5777
2014 m6385
2014 j6302
2014 j7481
2014 a5819
2014 s5713
2014 o6014
2014 n5569
2014 d5604
2015 j5286
2015 f5289
2015 m7739
2015 a8791
2015 m9053
2015 j10137
2015 j8733
2015 a6493
2015 s7340
2015 o7625
2015 n7213
2015 d6282
2016 j6111
2016 f6073
2016 m7027
2016 a7520
2016 m7423
2016 j8498
2016 j8614
2016 a5825
2016 s6820
2016 o7096
2016 n7310
2016 d5785
Source: CBS, Production Installation Register (PIR)

Distribution of solar panels between different types of homes

The link between new and existing administrative sources can also be used to determine the distribution of solar panels between different types of homes. Solar panels are significantly more common in owner-occupied homes (7.4%) than in rental properties (1.7%). On average, 4.93% of homes have solar panels, but this varies markedly from one type of home to another. Detached houses are most likely to have solar panels (in 11.72% of cases) and apartments the least likely (0.65%). See the illustration below for details of the percentage of installed solar panels in each type of home. In some cases, we were not able to obtain details of the type of home involved. It would be interesting to conduct a follow-up study, not only into the type of home involved, but also into the socio-economic status of their residents. Which groups are most likely to live in homes with solar panels?

solar panels per housing type

Regional statistics

CBS believes that, by making use of administrative sources in this way, it will be able to deliver better estimates of the amount of solar power being produced than it can using the current survey-based method. In addition, based on a survey funded by 40 Dutch local authorities and Netbeheer Nederland (the energy network operators’ branch association), CBS recently published details of solar power per local authority. This was the first time it has published data of this kind. In a later phase of this survey, the results of the data analysis concerning VAT refunds will be incorporated, together with data from the Netherlands Enterprise Agency’s (RVO) Energy Investment Allowance Scheme (EIA).

Using big data sources to fine-tune the measurement of solar power

To further fine-tune its statistical analysis of solar power production, CBS wants to use big data sources to pin down the specific locations of solar panels and to determine the amount of solar power being effectively generated, per solar panel. Outline details of the study currently being carried out at CBS are given below. Individual publications will follow when the study has progressed to the point at which it is possible to provide more detailed information.

Determining locations more accurately

CBS is conducting funded research at European level into the use of aerial photography to pinpoint solar panels. This will make it possible to establish the locations of solar panels with greater accuracy. CBS uses aerial photographs because those satellite images that are freely available do not have sufficient resolution to detect domestic solar panel installations. This research combines the use of artificial intelligence techniques and aerial photographs to automate the detection of solar panels. On this basis, we can work towards a harmonised method for accurately determining the number of solar panels in Europe as a whole. This additional resource is being used to validate and supplement our existing registry data.

More accurate estimation of effectively generated solar power

In addition to pinpointing the locations of solar panels more accurately, we are working on two different methods for improving our estimates of the amount of solar power being effectively generated.
The first method uses public data (from and other sources) to make reliable estimates of effective production per unit of installed capacity. This involves taking account of roof characteristics (slope and orientation), the site itself, and the weather conditions.
The second method uses advanced time series models to process data from the high-voltage power grid (Tennet) and solar radiation data from the Royal Netherlands Meteorological Institute (KNMI). Meteorological data on solar radiation is combined with high-frequency data on power consumption measurements for the Dutch high-voltage power grid. The concept underpinning these models is that if the solar panels are generating high yields then less power is supplied to the high-voltage grid and vice versa. In this way, a model can use these sources to indirectly derive a value for solar power. While the initial results give a plausible overall picture, better calibration is still needed. The model was supplied with data for the Netherlands as a whole, as this is freely available. However, this model can also be used at a lower (regional) level. At this level, the correlation between measured power consumption and meteorological data is expected to be even better, leading to more accurate models.

The figure below gives an initial impression of the development of solar power in the Netherlands in the period from 2010 to 2016.

Development of solar power

Your feedback is welcome

If you have suggestions for readily available data sources, please get in touch. We would be delighted to hear from you.