Met-8 Natural Colour RGB

Impact of California 2020 fires on atmospheric composition

15 August-22 September 2020

Met-8 Natural Colour RGB
Met-8 Natural Colour RGB

Data from multiple satellites and instruments can be used to investigate the impact the extensive California fires that ravaged the US state in 2020 had on atmospheric composition.

Last Updated

04 May 2023

Published on

02 December 2020

By Federico Fierli (EUMETSAT) and Ivan Smiljanic (CGI)

Each year we observe extensive and prolonged wildfires in many regions — the US state of California is known to be one of the most regularly affected. However, 2020 appears to be exceptional with preliminary data and estimates showing record-breaking extent and fire emissions. The first week in September 2020 was characterised by an even higher intensity of such phenomenon, as documented by satellite observations.

This case uses a series of different measurement techniques, to highlight some key elements of the wildfires during summer 2020 in California. The season was particularly intense in terms of burnt areas and intensity of fires, with intense trans-continental transport of smoke and pollutants. In addition, NASA reported that from 6 September there was an unusual pyrocumulonibus (Piro Cb) in southern California. Such clouds, that develop relatively often during intense wildfires at high latitudes, are much less frequent in temperate regions. They are comparable to convective towers and are generated by fire at high temperatures. A key feature of these clouds is to project at higher altitudes the fire-emitted materials, such as gas and particles, thanks to the intense uplift amplifying the impact of the fire on the composition of the atmosphere.


This same cloud is shown in the Metop-C infrared image (Figure 1) as a white corona-like structure (the high cloud) with a plume extending out of it (the particle outflow). Vertical uplift has a role in spreading fire emission products much further distances. Horizontal transport is more efficient at higher altitudes in the atmosphere due to stronger winds and less efficient removal. The cloud is also observed as a composite 'RGB', especially for Dust RGB contrast. It exhibits unusual green and yellow shades — possibly linked to stronger vertical ash ejections of ash and SO2 gas. (Figure 2).

Metop-C infrared
Figure 1: Metop-C infrared (10.9 um) 7 September 7th, 01:12 UTC. The image is centered over California.
Figure 2: GOES-17 ABI Dust RGB capturing Piro Cb plumes on 6 September 2020, hourly loop from 00:30 to 23:30 UTC.

Active fires

Widespread fires can be easily detected with the Fire Temperature RGB (Figure 3). This RGB shows information on the intensity of the fires, utilising respective temperature sensitivity of the following channels: IR3.9, NIR2.2 and NIR1.6. However, this RGB product is 'blind' to thinner clouds that might mask the fires, i.e. dampening the perceived intensity of the fire. Also, during the night it only shows the fires, with no situational awareness about the clouds and other features in the atmosphere or on the ground.

In attempt to tackle these two pitfalls, similar options for the 'Fire Temperature' RGB were explored (Note: names are only working titles for these RGBs).

GOES-17 Fire Temperature RGB image comparison

GOES-17 ABI Fire Temperature RGB during the night compare1

Figure 3: GOES-17 ABI Fire Temperature RGB during day and night fire scene on 9 September 2020, at 09:00 and 21:00 UTC.

The 'Fire RGB' (Figure 4) - with the following assignment for R-G-B combination: IR3.9 - NIR2.2 - NIR1.3. Here the NIR1.3 serves the role of thin (high) cloud detection, and, to a degree, cloud height assignment. Information on fire temperature is kept even with no NIR1.6 contribution (though missing out the information on the temperature of the hottest fire). Situational awareness is still limited to daytime applications.

GOES-17 ABI 'Fire RGB' image comparison

GOES-17 ABI 'Fire RGB' during the night compare1

Figure 4: GOES-17 ABI 'Fire RGB' (IR3.9, NIR2.2, NIR1.3) during day and night fire scene on 9 September 2020, at 09:00 and 21:00 UTC.

The '24h Fire RGB' (Figure 5 and Figure 6) - with the following assignment for R-G-B combination: IR3.9 - IR10.4 - BTD (IR12.3-IR10.4). By utilising infra-red channels this RGB can provide meaningful information both day and night, not only about the fire intensity (to a degree), but also on cloud and ground properties (limited information). Also, despite operating in the infrared range, smoke can be detected if thick and high enough (limited information), especially in the animated imagery.

GOES-17 ABI '24h Fire RGB' image comparison

GOES-17 ABI '24h Fire RGB' during the night compare1

Figure 5: GOES-17 ABI '24h Fire RGB' (IR3.9, IR10.4, BTD12.3-10.4) during day and night fire scene on 9 September 2020, at 09:00 and 21:00 UTC.

Figure 6: GOES-17 ABI '24h Fire RGB' (IR3.9, IR10.4, BTD12.3-10.4) full day loop, 30 min time step, on 19 August 2020.

Long-range transport

The second aspect is the effect of long-range transport of fire particle and gas emissions happening in the specific time-window of the event. Long-range transport is a known feature and can be observed satellites on several species.

Firstly we look at carbon monoxide (CO) that is produced by combustion and stays in the atmosphere for weeks. Figure 7 shows the accumulated concentration of CO from TROPOMI on board Sentinel 5P, from 9-11 September, with an illustrative view from Google Earth in Figure 8. The red area west of California had CO tropospheric concentrations four times higher than those recorded in Amazonia and Central Africa.


Figure 7: Carbon Monoxide tropospheric columns TROPOMI onboard Sentinel 5P, accumulations from 9 to 11 September 2020 



California 2020 fires on atmospheric composition (Google Earth)
Figure 8: View of the CO tropospheric concentrations using Google Earth


The intensity of transport could also be seen in the atmospheric particle products, such as the Aerosol Index product from the Metop-B and Metop-C GOME-2 instruments (Figure 9).

Metop-B GOME-2 AAI
Figure 9: Absorbing Aerosol Index from GOME-2 on board the Metop-B and Metop-C constellation on 15 August. Elaboration from AC-SAF.

Smoke transport to Europe is also detected, directly, or even indirectly, on SEVIRI imagery (more specifically the Natural Colour RGB). Normally, smoke is seen as homogeneous cyan-shaded veil in this product. However, Figure 10 shows very thin smoke that is only indirectly detected by looking at the lower-level water cloud field. In the areas with thin smoke aloft, water clouds have pink shades, instead of white. When shining on these clouds, through a thin smoke layer, Sun radiation is more absorbed in visible than near-IR region, hence the more red component of this RGB (NIR1.6) is reflected back to the satellite.


Met-8 Natural Colour RGB
Figure 10: Meteosat-8 Natural Colour RGB detecting smoke over Europe on 22 September 2020, 07:00 UTC.

More than a month after the fires abated the burn scars were still clearly visible in satellite imagery. On the GOES-17 Natural Color RGB from 29 October  (Figure 11), they can be seen as the quite large dark brown regions surrounded by green vegetation.

GOEs-17 Natural Color RGB
Figure 11: GOES-17 Natural Color RGB, 29 October 21:00 UTC


Additional content

California's Creek Fire Creates Its Own Pyrocumulonimbus Cloud (NASA)