Surface and atmospheric radiative processes depend strongly on wavelength. For example, snow albedo shifts from highly reflective at wavelengths <0.7µm (Vis) to highly absorptive at wavelengths >0.7µm (NIR). E3SM and CESM, fully coupled Earth system models, employ the atmospheric radiative transfer model RRTMG-SW, which aggregates solar radiation from 14 spectral bands into two bands (Vis and NIR) for transmission between the atmosphere and surface components. RRTMG-SW splits the flux in the overlap spectral band spanning 0.62–0.77µm evenly, sending half to the Vis band and half to the NIR band, which are then sent to the surface. The hyperspectral radiative transfer model SWNB2 shows that the fractional flux in the overlap band is actually distributed 55%/45% about the 0.7 µm Vis/NIR partition, not 50%/50% as represented in E3SM and CESM. Here we use E3SMv3 to simulate the climate response to the more realistic, asymmetric Vis/NIR partitioning of the overlap band, which shifts over 3.5 W/m² of surface insolation from the NIR to the Vis band.

Century-long fully coupled E3SMv3 simulations show that the greatest statistically significant changes occur over Antarctica during the summer months, where snow cover is constant and insolation is high. Shifting flux from the highly absorptive NIR band to the highly reflective (and transmissive, in the atmosphere) Vis band causes surface and lowest 500 mb tropospheric temperatures to decrease here by up to 1.8 K . Further statistically significant changes include increases in sea-ice area of over 5% in the southern Atlantic and southern Indian Ocean off the Antarctic Coast during local Fall, indicating that the shift in the spectral distribution of solar radiation lowers melt rates and increases sea-ice formation/persistence. Signals in the climatology over the Arctic, where most snow cover is seasonal, are less robust. Reasons why the surface and lower atmospheric temperatures above Greenland remain unchanged, despite similar forcing to Antarctica, will also be discussed. These results highlight the relationship between the spectral distribution of solar radiation and cryospheric surfaces, and the importance of accurately representing the spectral distribution of flux in fully coupled Earth system models.