HOW DOES VIRTUAL IRRADIANCE WORK?

Virtual Irradiance provides a satellite-image-based estimation of the sunlight shining down on the surface of Earth. It works on the principle that the amount of sunlight hitting the surface is determined by two questions:

  1. How much sunlight is hitting the top of the atmosphere at that moment? (i.e., Where is the sun in the sky?)
  2. How is the atmosphere affecting the amount of sunlight that gets through?

The question of where the sun is located is a long solved astronomical question, so we can calculate the available sunlight at the top of the atmosphere in a straightforward manner.

The question of how the atmosphere is affecting the light getting through can be answered by assessing factors like cloud cover, atmospheric turbidity, etc. Virtual irradiance uses satellite images and other related data to make this assessment and predict the amount of sunlight hitting any location in the Continental US, Hawaii, and Asia.

WHY IS IRRADIANCE DATA IMPORTANT FOR UNDERSTANDING AND OPTIMIZING APV ASSET?

Quality irradiance data is critical to understanding a solar PV site’s performance. Unless you know how much sunlight energy has gone into the system, you cannot tell if the system is operating as expected.

On-site irradiance sensors can be expensive and challenging to maintain because of their sensitivity to soiling, snow cover, inter-row shading, positioning error, and calibration errors. Virtual Irradiance avoids these challenges by using a methodology based on satellite images and consistent calibration.

WHAT INPUT DATA SOURCES ARE USED BY VIRTUAL IRRADIANCE?

Virtual Irradiance uses a variety of input data sources, including:
• Satellite images from the geostationary satellites
• Elevation data
• Snow data
• Other atmospheric data

HOW MANY SITES CAN VIRTUAL IRRADIANCE PROCESS?

We don’t have a limit on the number of locations for which customers can purchase data. The Virtual Irradiance system has roughly a 1 km2 resolution across the Continental US and Hawaii, and 5 km2 resolution in Asia. 

WHAT IS THE ACCURACY OF VIRTUAL IRRADIANCE IN THE US?

We are currently on our sixth generation irradiance model, and we have made significant improvements to the model over time.

The Virtual Irradiance model has been optimized and validated against high quality irradiance stations in the US Surfrad and ISIS weather station networks, using a leave-one-out cross-validation methodology to provide a reliable estimate of accuracy across a wide variety of climates and weather conditions.

We use a % mean absolute insolation error metric to quantify the accuracy of Virtual Irradiance, because this metric most closely matches the type of errors in total energy difference that are generally of interest to our customers. The % mean absolute error metric provides the typical error a user should expect when comparing VI data to a perfectly accurate measurement of sunlight energy, for a giving time scale.

Where:
• i is the interval for the calculation (e.g., hourly, daily, monthly, yearly intervals)
• f i is the VI prediction, and
• yi is the value from the weather station

Virtual Irradiance error in the US is ~2.6% for monthly data, and ~1.8% for yearly data.

The errors above have been quantified across all the different climates and weather conditions in the Surfrad and ISIS weather station locations. Due to the nature of how it works, Virtual Irradiance will have slightly higher error in locations with long-term snow cover or dense long-term cloud cover, and slightly lower error in locations that tend to be sunnier and have less prolonged periods of snow cover. This means that for many solar PV applications, which tend to be installed in sunny locations, the Virtual Irradiance system can have even better accuracy than quoted above.

Note: A comprehensive study of the error rate in Asia has not been completed. Because the resolution is lower in Asia,error rates may be higher, but the overall quality and methodology of analysis remains the same.

HAS VIRTUAL IRRADIANCE BEEN INDEPENDENTLY VALIDATED?

Virtual Irradiance has been independently validated by a number of our clients using their own proprietary irradiance data.If you would like to validate VI with a data set of your own, please discuss this with your sales representative. We do not currently have a publicly available independent report validating the accuracy of Virtual Irradiance.

HOW CAN I GET ACCESS TO VIRTUAL IRRADIANCE DATA?

Virtual Irradiance is available through the LocusNOC web application’s charting and reporting tools, through our Datalink Excel plugin tool, and directly from our Locus API. Using these tools, you can directly visualize the data, or download it for use in other tools.

Virtual Irradiance data can be provided for any location in the Continental US, Hawaii, and Asia, even if you are not using Locus monitoring for those locations.

HOW DOES VIRTUAL IRRADIANCE COMPARE TO ON-SITE IRRADIANCE SENSORS?

Virtual Irradiance can be used be used to complement or replace on-site irradiance sensors. On-site irradiance sensors can be expensive and challenging to maintain because of their sensitivity to soiling, snow cover, inter-row shading,positioning error, and calibration errors. Virtual Irradiance avoids these challenges by using a methodology based on satellite images and consistent calibration.

For small and mid-scale sites, Virtual Irradiance can be a very cost effective way to acquire irradiance data without the need for on-site irradiance sensors.

For large-scale sites, Virtual Irradiance can be a great complement to on-site sensors. Virtual Irradiance can be used as an independent check on the data coming from on-site sensors, enabling identification of sensor shading, soiling, calibration issues, or installation errors. Virtual Irradiance can also be used to backfill gaps in the data acquired from on-site sensors.

HOW DO ON-SITE IRRADIANCE SENSORS AND VIRTUAL IRRADIANCE COMPARE TO RELATIVE PERFORMANCE ALGORITHMS FOR ASSESSING SITE PERFORMANCE?

Locus switched from relative performance analyses to the use of Virtual Irradiance for assessing site performance, despite investing significantly in relative performance algorithms (we hold 3 patents in this area), because relative performance approaches have some inherent weaknesses that cause them to be very inaccurate compared to Virtual Irradiance or on-site irradiance sensors.

Relative performance algorithms generally work on the principle that you can determine the weather impact on a solar PV system’s output by looking at neighboring sites’ performance. The idea is that if you use a set of comparison sites for a benchmark, the collective set of comparison sites will represent weather effects and idiosyncratic performance issues washout. The algorithm’s basic steps are:

  1. For a given location, find all solar PV systems within a given distance
  2. Normalize the performance of the nearby systems by their expected output level
  3. Filter the sites to remove outliers or otherwise non-comparable systems (e.g., sites with clear production problems,data that looks incorrect, different orientations, different technologies, etc.)
  4. Compare the normalized benchmark of selected sites

The reasons this type of algorithm has a much higher error than Virtual Irradiance or on-site sensors are:

  1. Limited coverage areas: The algorithm is limited to areas with a significant number of solar PV systems that pass all the filters.
  2. Nearby sites do not always experience similar weather: In some locations nearby sites experience very different weather due to microclimates (e.g., around San Francisco, or other coastal versus nearby inland site situations).
  3. Many sources of error affect all sites in an area together: E.g., snow or soiling often affect broad areas, and slow accumulation of soiling is particularly problematic. This causes relative performance to underestimate the actual sunlight received
  4. Different sites can behave differently under the same general weather conditions: For example, different orientations behave differently due to time-of-day weather effects, different technologies (crystalline vs thin film) behave differently under diffuse light or different temperatures, etc.
  5. High sensitivity to bad data: Because relative performance compares peer-sets, any sites with bad data that slip through the quality filters will corrupt results for all nearby sites, which makes the analysis very sensitive to data quality problems

WHAT IS THE COVERAGE AREA FOR VIRTUAL IRRADIANCE?

Virtual Irradiance currently provides coverage for the Continental US, Hawaii, and Asia. The Asia release covers data across India, Pakistan, Iran, UAE, Malaysia, Singapore, Sri Lanka, Oman, Azerbaijan, Uzbekistan, Turkmenistan, Kyrgyzstan, Tajikistan, Afghanistan, Nepal, Bhutan, Bangladesh, Myanmar, Thailand, as well as parts of Saudi Arabia, Iraq, Yemen,Eastern China, Kazakhstan, Mongolia, Laos, Cambodia, Indonesia.

DO YOU PROVIDE A FORECASTING PRODUCT?

We currently provide Virtual Irradiance on a historical basis only, with data available from March 2012 to present time for the Continental US and Hawaii, and from October 2015 to present for Asia. We do not currently offer a Virtual Irradiance for irradiance forecasts.

Did this answer your question?