Impact of Wildfires on Solar Projects

The link between wildfires and solar generation has been highlighted in recent years. In 2023 specifically, several articles were published, such as the article in June in Wired Magazine, “Wildfire smoke slashed solar power generation” and in more academic studies, such as NREL and the University at Albany’s report in July “Impact of Wildfires on Solar?Generation, Reserves, and?Energy Prices.”

As the occurrence of wildfires increases due to climate change and anthropogenic causes, we felt it was important to consider the potential impact of wildfires, and specifically wildfire smoke, on solar energy production. Natural Power’s team investigated the energy impacts of wildfires and the effect on assumptions as it relates to individual projects. As the satellite datasets have started to include high impact wildfire years (such as 2020) in typical year calculations, we wanted to evaluate wildfires as a risk to our solar assessments.??

For this study, Natural Power reviewed solar resource datasets and specifically Global Horizontal Irradiance (GHI) for an area in the central valley of California. The location was chosen for its proximity to operational PV projects.? Figure 1 shows the daily averages of GHI at this site for 2020, and the smoothed curve is Clear Sky GHI, which ideally represents maximum GHI without any cloud coverage?

Knowing that August through October were high impact wildfire months for this location, the team evaluated the Particulate Matter 10, that is Particulate Matter of 10 micrometers or less in diameter (PM10), as CAISO suggested, and saw that the peaks in PM10 corresponds with the drops in the green, confirming that the presence of particulate matter can result in lower GHI, as shown in Figure 2.?

As comparing smoke losses to cloudless conditions may be an unrealistically high bar, we added in the Typical GHI Year (TGY) dataset for the high impact months. This allows us to approximate the losses in GHI from the smoke which is represented in red in Figure 3.?

It is important to note that for solar production modeling, the ratio of Typical Meteorological Year (TMY) to Typical GHI Year(TMY/TGY), also known as themedian approach is what is generally used to model a P50 energy year. For this period highlighted in red, GHI is about 20% lower than TGY. 2020 was known as a record setting year for low GHI, so Natural Power wanted to assess every GHI year as provided by the time series data sets.?

Natural Power assessed the combined GHI for August through September for all the years for which time series are available, that is 1998 through 2023. Figure 4 shows the GHI plotted as a histogram with monthly sums binned. It is important to note that three of the last four years (2020 through 2023) are in the low-side tail. The year 2022 was a normal one, and in the middle of the pack.

Next, we plotted PM2.5,? that is Particulate Matter of 2.5 micrometers or less in diameter, monthly sums as an indication for fire events, shown in Figure 5. Using the 90th percentile as the rule for defining a wildfire event, we have highlighted those events in red. Based on this visual rendition, there have been more wildfire events in recent years and in fact half the forest fires as defined by the 90th percentile happened after 2017.??

Although the trend for the frequency of wildfires is changing and is not well understood, evidence suggests that wildfires are increasing and their impacts tend to be site specific. Three of the last four years have experienced wildfire events that have resulted in outliers in GHI, and for the California site Natural Power studied, half of the fires for the period evaluated, have happened after 2017?

As previously mentioned, the ratio TMY/TGY is used as the basis of P50 energy models. Wildfire impacts may not yet be captured in TMY or other median-based approaches, as they are outlier events which by definition are hard to capture in a Typical Year approach. Although we are using the best information from the past as the basis of our estimates for the future, there will always be uncertainty in future projections. Full long-term time-series modeling may start capturing these wildfire effects.??

On the basis of our study, we recommend using sensitivity and downside analyses which will help in the evaluation of potential losses and the negative impact that wildfires may have on the performance of specific solar PV installations in areas which are particularly prone to the effects of wildfires.?