Lowering uncertainty is fundamental to managing solar investment risk

Humans dislike uncertainty

Humans aren't big fans of uncertainty. We go to great lengths to protect ourselves from the risks and consequences of life's surprises, but the nature of the future is that it is uncertain. In life, we can't change this fact, but we can minimise the consequences of various outcomes.

Most often, we turn to financial mechanisms to limit the outcomes of life's uncertainty. Income insurance is a great example of managing the risks in the uncertainty of future employment. We take out income insurance, not to change the uncertainty itself, but to limit the consequences of an outcome where you find yourself unemployed or unable to work.

In the financial world, uncertainty greatly influences the risk management practices of investors. The future nature of financial markets are rife with uncertainty, reacting daily to unexpected global events. As a result, investors have devised ways of minimising the impact of the many possible negative outcomes, using derivatives or hedge-funds to provide protections. But still the outcomes remain uncertain, it is only the consequence of the risk that could be managed. Does the same framework apply to investing in solar development projects? Let's explore...

Uncertainty in historical solar data

An essential element of the solar development timeline is the process of due diligence done during the financing and/or purchasing stage for a solar farm.

As with all investments, there are a number of risks involved with bringing a solar generation asset into your portfolio. Setting the quality of operation and management of the asset aside (out of scope for us to consider here), chief among the risks is the asset failing to deliver on the rate of return promised by the developer.

The principal input to the rate of return calculation is the expected revenue generation from the solar farm (followed then by operational expenditure). And this is where we arrive once again at the importance of high quality solar irradiance data: revenue generation is primarily influenced by the availability of solar irradiance at that location over time.

Solar irradiance data for decision making

Beyond ~7 days ahead (e.g. the data produced by our rapid update solar forecasting system), there is a very quick drop off in the accuracy of 'forecasts' of solar irradiance. It is well beyond the capability of our numerical weather models to make accurate predictions of cloud cover beyond a few days ahead. This means that the best proxy for what solar radiation is likely to be available over the next several years or next few decades, is looking at the past availability of solar irradiance at that location.

This is precisely why Solcast offers three historical solar data products in support of the investment and due diligence process, these are:

1. Historical Time Series data (timestamped data built through direct monitoring of the local cloud cover conditions at those specific times)

2. Typical Meteorological Year "TMY" data (constructed from 15+ years of historical data to build representative scenarios of expected values of hourly solar irradiance and weather variables)

3. Monthly Averages (averages of the solar irradiance available at a location for a given month, using our historical database of 15+ years of data)

Uncertainty in solar data services

Each of the above data products are relevant to different stages of the solar farm development and financing process, and are designed specifically to help represent, accurately and confidently, the expected solar irradiance available over time at a given location. These data therefore form the basis of the expected revenue generation for the solar farm asset.

Let us now return to the question I prompted earlier - what role does uncertainty play in the risk management of solar farm investment?

The uncertainty in the historical solar data products used to evaluate investment risk inherently influence the risk profile itself. This is to say that in using solar data to assess risk, part of the risk profile is that the solar data itself has its own level of uncertainty.

Are we going in circles here? Let me approach this a different way.

Not all solar data services are created equally. Many, like Solcast, use satellite imagery, but may not use the latest in state of the art solar radiation modelling. Others create synthetic data which are influenced by coarse weather models or weather observations that are 10s to 100s of kilometers away.

As a consequence of the widely varying approaches, as each solar dataset comes with its own unique level of uncertainty in that data. Meaning, uncertainty in the ability of those data to accurately model the solar irradiance arriving at the Earth's surface. This uncertainty is estimated by comparing the dataset estimates of solar irradiance to measurements from high quality measurement stations.

Solcast - dedicated to lowering uncertainty

Let's look at some numbers to further explore this concept. Uncertainty in historical solar data is best understood as a measure of confidence in the representativeness of our the dataset as a whole to actual measured conditions. For this purpose, Solcast has developed a validation tool, where you can explore similar statistics for your local climate region. This validation is based off of the actual measurements of solar irradiance from high quality weather stations that I just mentioned.

One of the key differentiators of the three historical data services on offer by Solcast compared to other solutions on the market, is that our data offer the lowest uncertainty available. What this means, is a direct translation to lower risk for investors or financiers making investment decisions on solar farm assets.

Uncertainty comparison across the market

In the below table we summarise the results of our global validation for global horizontal irradiance (the most widely used and relevant value of solar irradiance). Uncertainty in this context is represented not by the overall mean bias, but by the standard deviation of the bias error. The higher this standard deviation climbs, the wider the distribution of possible outcomes based on the quality of the solar irradiance estimates becomes (therefore increasing uncertainty).

To help you dive deeper, we've published comparison pages that contrast the quality of our historical data services to those of several other options on the market. As you review these, be sure to pay attention to the uncertainty of each dataset, which can be best estimated by the "Bias Deviation (GHI)" value in each table.

• SolarGIS
• Vaisala 3Tier
• Meteonorm

Solcast plans to continue to add to this list of comparisons over the next few months.

Work with low uncertainty data - in minutes!

Through the Solcast API Toolkit, we have made it possible for those evaluating solar investment risk, as well as for the solar farm developers creating those opportunities, to access our historical data products in just a few minutes. Registration is fast and free, comes with free historical data requests for your personal evaluation and does not require a credit card or subscription. You can register for the API Toolkit here, or, if you'd like, explore detailed information on our Historical and TMY data services.