Scale dependence of cirrus heterogeneity effects. Part II: MODIS VNIR and SWIR channels

Abstract. The understanding of the radiative role of clouds is crucial. Ice clouds such as cirrus have, on average, a significant positive radiative effect, while in some conditions it may be negative. However, many uncertainties remain on the role of this type of clouds on Earth's radiative budget and in a changing climate. Global satellite observations are particularly well suited to monitor clouds, retrieve their characteristics and infer their radiative impact. To retrieve ice cloud properties (optical thickness and ice crystal effective size), current operational algorithms assume that each pixel of the observed scene is plane-parallel and homogeneous, and that there is no radiative connection between neighboring pixels. This retrieval representation is far from the reality, where the radiative transfer is 3D, leading to the plane parallel and homogeneous bias (PPHB) and the independent pixel approximation bias (IPAB) impacting both the estimation of top of the atmosphere (TOA) radiation and retrievals. An important factor that constrains the impact of these assumptions is the sensor spatial resolution. High spatial resolution pixels can better represent cloud variability (low PPHB) though the radiative path through the cloud can involve many pixels (high IPAB). In contrast, low spatial resolution pixels poorly represent the cloud variability (high PPHB) but the radiation is better contained within the pixel field of view (low IPAB). In addition, the solar and viewing geometry (as well as cloud optical properties) can modulate the magnitude of the PPHB and IPAB. In this Part II of our study, we have simulated TOA 0.86 μm and 2:13 μm solar reflectances over a cirrus uncinus scene produced by the 3DCLOUD model. Then, 3D radiative transfer simulations are performed by the 3DMCPOL code at spatial resolutions ranging from 50 m to 10 km, for twelve viewing geometries and nine solar geometries. It is found that, for simulated nadir observations taken at resolution higher than 2.5 km, horizontal radiation transport (HRT) dominates biases between 3D and 1D reflectance calculations, but it is mitigated by the side illumination effect for off-zenith solar geometries. At resolutions coarser than 2.5 km, PPHB dominates. For off-nadir observations at resolutions higher than 2.5 km, the dominant effect is that the oblique line of sight passes through many cloud columns, but other 3D effects are also important. Similar to nadir simulations, side illumination effects mitigate the HRT. At resolutions coarser than 2.5 km, the PPHB is again the dominant effect. The magnitude and resolution-dependence of PPHB and IPAB is very different for visible, near-infrared, and short-wave infrared channels compared with the thermal infrared channels discussed in Part I of this study. This strong wavelength dependency of cirrus cloud 3D radiative effects may be a significant issue for retrieval techniques that simultaneously use radiative measurements across a wide range of solar reflectances and infrared wavelengths.