Why Not Use Public Resource Data Sets for Solar Resource Assessment?

Vaisala 3TIER Solar Time Series
Naomi Stringfield
Aug 07th 2017
Renewable Energy

You're a developer who is working on a large solar project and you're deciding which solar resource data set to use. You've got a tight deadline and no guarantee that you’ll get a power purchase agreement, which means you don’t want to spend a lot of money. Why not just use some of the free public data that’s available in PVsyst?

Sound familiar? Developers don't have many guidelines for what to use when it comes to resource data except those they impose on themselves and occasional guidance from the financial community. Also there is little to no information available that compares different resource data sets and allows developers to make informed choices. How do you make a good choice?

There are three key questions to ask when choosing your resource data.

  1. Are hourly data available, or only monthly means?
  2. How old is the data set?
  3. What is the uncertainty?

If you can answer these questions about the data you use, then you're already on the path to making a good choice. Based on these answers and your risk appetite, you can make an informed tradeoff between schedule, accuracy, and cost.

Monthly vs. Hourly Data

Let's first tackle the monthly vs hourly debate. PVsyst and some other software programs will create an hourly energy profile for you based on interpolating from annual and monthly average values. At very early stages of project design, this can be a quick way to get indicative numbers. If you have very good monthly data, then this approach may even be ok for fixed PV projects.

The problem with this approach is it can add too much uncertainty, even in the early prospecting stages of project development. This is especially true for tracking PV plants. In one recent study, interpolating the data resulted in a 7% uncertainty, which is too high since it could make or break a project. This is why we recommend using hourly data at all project phases.

Age and Length of the Data Set

The second consideration is how long the data record is and when it was last updated. Any data set being used for long-term project estimates should be at least 10 years in length or longer. This is where some of the public data sets really shine because they have 20 years or more data incorporated. Unfortunately, they are mostly only available as TMY (Typical Meteorological Year) time series so you don’t actually get to use all of the data to evaluate the annual variability in the resource. Some sites vary more or less than the industry standard assumption of 2.5% a year for resource variability. If you rely on a shorter data set, you could be quite surprised when it comes to operating the plant.

When the data set was last updated is critical depending on the overall age of the data set and on your location. For example, in PVsyst the NASA data available globally has a 20+ year record but hasn’t been updated since 2005. Meteonorm offers 30+ years of data, but its records end in 2010. Climate changes at your location in the last decade will not be captured by these data sets. Are you willing to take that chance?

Uncertainty of the Data Set

Finally you want to consider the effect that a resource data set’s uncertainty is going to have on your energy estimates when it comes time to get past your internal review board or to secure external financing. High-quality, private data sets have an annual uncertainty, defined as the standard deviation of bias error, of 5% or less. For example, Vaisala has 5 different models which have a range of GHI uncertainty between 4.4% and 4.8% globally. By contrast, the Meteonorm version 6 information freely available in PVsyst has an uncertainty of 10-12%.

So what kind of impact does that have? We can isolate the resource uncertainty’s impact on energy estimates by keeping all other project components the same and only looking at how the P90 values change. We are concerned about the P90 values because they are sensitive to uncertainty and the banks are sensitive to the P50 to P90 ratio as a measure of project risk.

Taking a hypothetical project in India with a 75 MW DC capacity we get Year 1 P50: 121.6 GWh. When we apply a 5% resource uncertainty we get Year 1 P90 (5% unc): 110.5 GWh and when we apply a 10% uncertainty, keeping all other components equal, we get a Year 1 P90 (10% unc): 105.0 GWh. That is a nearly 5% drop in the P90 value and overall there was an almost 4% increase in the overall energy uncertainty that would be presented to the bank from 7.1% to 10.7%.

Bottom Line

Every developer has a different appetite for risk – and a different budget. By considering carefully what solar resource data set to rely on, you can make a good decision that results in the best balance of up-front investment and bankability.

What if you didn't have to choose between keeping a tight schedule, accuracy of the resource data, and the cost to acquire it? With Vaisala's Solar Time Series Tools you can get access to highly accurate resource data within hours and for as low as $50 a time series for hourly data.

Click for more information and to view online purchase options.

Add new comment