Figure 2 - uploaded by Xuemei Chen
Content may be subject to copyright.
(a, b) Distributions of simulated TOA reflectance spectra in overcast conditions ρ ovc for the different viewing geometries in the look-up table and for a solar zenith angle of 30 • , with a thick liquid cloud (COT = 150). (a) CTH = 15 km; cloud base height (CBH) = 2 km. (b) CTH = 0.5 km; CBH = 0.2 km. (c) Error on ρ ovc caused by a misattribution of cloud height to the "low thick cloud" category. Green, red, and blue arrows indicate spectral regions with main absorption features from O 3 , O 2 , and H 2 O, respectively.
Source publication
We develop a new way of retrieving the cloud index from a large variety of satellite instruments sensitive to reflected solar radiation, embedded on geostationary and non-geostationary platforms. The cloud index is a widely used proxy for the effective cloud transmissivity, also called the “clear-sky index”. This study is in the framework of the de...
Contexts in source publication
Context 1
... spectral radiative transfer simulations of ρ ovc show that there is also a significant dependency between the TOA cloudy reflectances and other variables. In Fig. 2, we represent two-dimensional histograms of TOA reflectances calculated from such simulations, with a solar zenith angle set to 30 • as an example. For most wavelengths, a significant spread of the distribution is observed (Fig. 2a and b), corresponding only to different viewing geometries defined by a linear mesh grid in the cosine of ...
Context 2
... of ρ ovc show that there is also a significant dependency between the TOA cloudy reflectances and other variables. In Fig. 2, we represent two-dimensional histograms of TOA reflectances calculated from such simulations, with a solar zenith angle set to 30 • as an example. For most wavelengths, a significant spread of the distribution is observed (Fig. 2a and b), corresponding only to different viewing geometries defined by a linear mesh grid in the cosine of the viewing zenith angle (θ v ) and difference φ of the solar and viewing azimuth angles (Table ...
Context 3
... this paper, we assume a cloud optical thickness (COT) of 150 to define optically thick clouds and overcast conditions. This assumption relies on COT statistics from Tr- ishchenko et al. (2001). The simulations for a low thick cloud (cloud top height (CTH) at 500 m) and a high thick cloud (CTH at 15 km) show in general good agreement (Fig. 2c), except in absorbing bands of O 2 (mainly at 690 nm, O 2 -B band, and 762 nm, O 2 -A band) and H 2 O (mainly at 725, 820, and 950 nm) and for short wavelengths where scattering becomes increasingly significant (e.g., Jin et al., 2011). For these wavelengths, the TOA reflectances with low clouds can be much lower than for high clouds ...
Context 4
... is worth noting that we use MCD43C1v6 BRDF data regardless of their quality flags. We observe though that keeping only the highest-quality data improves significantly statistics (Fig. B2). Also, the choice of a spectral linear interpolation between MODIS channels to simulate surface reflectances in SEVIRI channels is supposed to contribute significantly to biases observed in ρ clear simulations, in particular for the 0.8 µm channel with vegetated surfaces due to the red edge spectral pattern (low reflectivity below ...
Context 5
... spectral radiative transfer simulations of ρ ovc show that there is also a significant dependency between the TOA cloudy reflectances and other variables. In Fig. 2, we represent two-dimensional histograms of TOA reflectances calculated from such simulations, with a solar zenith angle set to 30 • as an example. For most wavelengths, a significant spread of the distribution is observed (Fig. 2a and b), corresponding only to different viewing geometries defined by a linear mesh grid in the cosine of ...
Context 6
... of ρ ovc show that there is also a significant dependency between the TOA cloudy reflectances and other variables. In Fig. 2, we represent two-dimensional histograms of TOA reflectances calculated from such simulations, with a solar zenith angle set to 30 • as an example. For most wavelengths, a significant spread of the distribution is observed (Fig. 2a and b), corresponding only to different viewing geometries defined by a linear mesh grid in the cosine of the viewing zenith angle (θ v ) and difference φ of the solar and viewing azimuth angles (Table ...
Context 7
... this paper, we assume a cloud optical thickness (COT) of 150 to define optically thick clouds and overcast conditions. This assumption relies on COT statistics from Tr- ishchenko et al. (2001). The simulations for a low thick cloud (cloud top height (CTH) at 500 m) and a high thick cloud (CTH at 15 km) show in general good agreement (Fig. 2c), except in absorbing bands of O 2 (mainly at 690 nm, O 2 -B band, and 762 nm, O 2 -A band) and H 2 O (mainly at 725, 820, and 950 nm) and for short wavelengths where scattering becomes increasingly significant (e.g., Jin et al., 2011). For these wavelengths, the TOA reflectances with low clouds can be much lower than for high clouds ...
Context 8
... is worth noting that we use MCD43C1v6 BRDF data regardless of their quality flags. We observe though that keeping only the highest-quality data improves significantly statistics (Fig. B2). Also, the choice of a spectral linear interpolation between MODIS channels to simulate surface reflectances in SEVIRI channels is supposed to contribute significantly to biases observed in ρ clear simulations, in particular for the 0.8 µm channel with vegetated surfaces due to the red edge spectral pattern (low reflectivity below ...
Similar publications
A majority of snow radiative transfer models (RTMs) treat snow as a collection of idealized grains rather than an organized ice–air matrix. Here we present a generalized multi-layer photon-tracking RTM that simulates light reflectance and transmittance of snow based on X-ray microtomography images, treating snow as a coherent 3D structure rather th...
Citations
... Because of the computational cost of running 43 such models, they also have relatively coarse spatial and temporal scales (Verbois et al., 2020). Satellite-44 derived solar irradiance estimations rely on both observations -the satellite channels -and models to 45 estimate the SSI from these channels (Rigollier et al., 2002;Tournadre et al., 2021). They can be seen as 46 a compromise between ground-based sensors and numerical weather models: like the latter, they cover 47 a large area and period, but their temporal and spatial resolution is usually finer (Ineichen, 2014). ...
Estimations of solar surface irradiance (SSI) derived from meteorological satellites are widely used by various actors in the solar industry. However, even state-of-the-art empirical and physical SSI retrieval models exhibit significant errors; the estimations provided by these models are thus traditionally corrected using ground-based measurements of SSI as references. The literature is rich with such correction methods, often called adaptation techniques. Most of the proposed models, however, are local or site-specific, i.e., they do not extrapolate the correction in space and are only applicable to the location of the ground-based measurements.
In this work, we propose a novel global adaptation technique, that can extrapolate the correction in both space and time. To that end, we leverage (1) a dense network of measurement stations across France, (2) a relatively large number of predictors, and (3) a non-linear, sophisticated regression algorithm, the Extreme Gradient Boosting. The model is applied to the HelioClim3 database; its performance is benchmarked against raw HelioClim3 estimations, and alternative, simpler adaptation techniques.
Our analysis shows that this global model significantly improves satellite-derived SSI estimations from the HelioClim3 database, even when the evaluation is carried out on measurement stations that were not part of the training set of the algorithm. Our proposed model also outperforms all tested alternative global adaptation techniques.
These results suggest that global adaptation techniques leveraging advanced machine learning and high dimensionality have the potential to significantly improve satellite-derived SSI estimations, notably more than traditional adaptation approaches. There is certainly room for improvement, but the development of such techniques is a promising research topic.
