Impact of reflected shortwave anisotropy on satellite radiometer measurements of the Earth's energy imbalance

<p>The Earth's energy imbalance is the difference between incoming solar radiation and outgoing reflected and emitted radiation from the Earth, and quantifies the current ongoing accumulation of energy in the Earth's climate system. There are indications that the imbalance is growing, and it is important to be able to measure and monitor this quantity to better constrain future changes. The reflected shortwave component of the outgoing radiation depends on surface and atmospheric properties, which leads to strong directional variations associated with the angular geometry relative to the incoming sunlight and the observer. The effect of this anisotropy on global average measurements from wide-field-of-view radiometers has been the topic of some investigation in the past, and results from an earlier study suggest that this effect could potentially lead to substantial systematic biases in the context of the global mean reflected shortwave radiation. Here we simulate wide-field-of-view instruments on satellites in polar, sun-synchronous and precessing orbits, as well as constellations of these types of satellite orbits, with both Lambertian and anisotropic shortwave reflection. Reference top-of-atmosphere radiative fluxes are taken from the Clouds and the Earth's Radiant Energy System synoptic data product and combined with angular dependence models that were developed for the Earth Radiation Budget Experiment. We find that the polar regions appear brighter and the midlatitudes appear dimmer with anisotropic reflection, but both the estimated global annual mean and the estimated interannual trend only exhibit limited sensitivity to whether Lambertian or anisotropic reflection is assumed. With anisotropic reflection, the estimated global annual mean root-mean-square sampling error is at most 0.11 <span class="inline-formula">W m⁻²</span> provided that at least two complementary satellites are used, compared with at most 0.09 <span class="inline-formula">W m⁻²</span> in the case of Lambertian reflection. The magnitude of the difference in the estimated interannual trend is at most 0.07 <span class="inline-formula">W m⁻²</span> per decade, and typically only <span class="inline-formula">∼</span> 0.01 <span class="inline-formula">W m⁻²</span> per decade. Analysis of the angular sampling of these satellites reveals that the anisotropic reflection requires sufficient sampling of viewing zenith angle and relative azimuth angle, in addition to the solar zenith angle. However, we conclude that it is possible to choose satellite orbits so that the sampling error is not substantially affected by reflected shortwave anisotropy.</p>