What we did
An earlier result in this research area left one question open: at the Railroad Valley reference site (RVUS), an offset of about five percent was not explained by the sensor, the viewing geometry or the atmosphere. This entry asks when the difference appeared and what can and cannot yet be said about its cause.
We compared Landsat 8 and Landsat 9 top-of-atmosphere readings directly with RadCalNet's modelled top-of-atmosphere values for RVUS, over the same area around the site, with no atmospheric processing in between. Landsat 8 covers 2013 to 2025, with more than 120 dates; Landsat 9 covers 2021 to 2025, with about fifty. For each year we measured how the ratio between satellite and reference changes with the height of the sun.
As a second check, we reread RefCal's Sentinel-2 surface results for 2024 and 2025 at six RadCalNet sites in the same way. Sentinel-2 uses an independent satellite calibration chain. This comparison uses RefCal surface results, not a Sentinel-2 top-of-atmosphere comparison.
What we saw
From 2013 to 2021 the ratio did not depend on sun height: summer and winter dates agreed. Between December 2022 and June 2023 that changed, in both Landsat satellites at once. Since 2023, at low sun, Landsat top-of-atmosphere values are about five to eight percent higher than RadCalNet's modelled values at RVUS; at high sun, in summer, they are close.
This is the largest change in the thirteen-year record and the only one that has lasted. A smaller one-year change of the same kind appeared in 2017, after a wet spring, and reversed by 2018.
The Sentinel-2 surface comparison shows a similar low-sun pattern at RVUS in 2024 and 2025, consistent across the bands compared. No other RadCalNet site shows it. La Crau shows a seasonal effect of the opposite sign, strongest in the visible bands, which points to a different, still unresolved cause there.
What it means
The comparison at RVUS now changes with the seasons, and it began doing so between late 2022 and mid 2023. These data do not establish whether the cause is the playa surface, the reference measurements, how they are carried to the top of the atmosphere, or a satellite-side effect. The timing matches the wet 2022 to 2023 season on the playa.
That narrows the open question from this research area to a dated change with a known shape. The next step is to extend the Sentinel-2 comparison back to 2022 and 2023, so that an independent satellite can be tested across the period in which the change appeared, before any claim about which side changed.