Robust cloud products are critical for DSCOVR to make a significant contribution to climate studies. Building on our team’s track-record in cloud detection, cloud property retrieval, oxygen band exploitation, and DSCOVR-related studies, we propose to develop a suite of algorithms for generating the operational EPIC cloud mask, cloud height and cloud optical thickness products. Multichannel observations will be used for cloud masking; the cloud height will be developed with information from the oxygen A- and B- band pairs (780 nm vs. 779.5 nm and 680 nm vs. 687.75 nm); for the cloud optical thickness retrieval, we propose an approach that combines the EPIC 680 nm observations and numerical weather model outputs. Preliminary results from radiative transfer modeling and from proxy data applications show that the proposed algorithms are viable. Product validation will be conducted by comparing EPIC observations/retrievals with counterparts from coexisting Low Earth Orbit (LEO) and Geosynchronous Earth Orbit (GEO) satellites. The proposed work will include a rigorous uncertainty analysis based on theoretical and computational radiative transfer modeling that complements standard validation activities with physics-based diagnostics. We also plan to evaluate and improve the calibration of the EPIC O2 A- and B-band absorption channels through tracking the instrument performance over known targets, such as cloud free ocean and ice sheet surfaces. The deliverables for the proposed work include an Algorithm Theoretical Basis Document (ATBD) for peer-review, products generated with the proposed algorithms and supporting research articles. The data products, which will be archived at the Atmospheric Science Data Center (ASDC) at the NASA Langley Research Center, will provide essential inputs needed for the community to apply EPIC observations to climate research and to better interpret NISTAR observations. The proposed work directly responds to the solicitation to “develop and implement the necessary algorithms and processes to enable various data products from EPIC sunrise to sunset observations once on orbit”, and also for improving “the calibration of EPIC based on in-flight data”.

DSCOVR_EPIC_L2_CLOUD_03 is the Deep Space Climate Observatory (DSCOVR) Earth Polychromatic Imaging Camera (EPIC) Level 2 Cloud version 03 data product. The EPIC Level 2 cloud products include Cloud Mask (CM), Cloud Effective Pressure (CEP), Cloud Effective Height (CEH), Cloud Effective Temperature (CET), Cloud Optical Thickness (COT), and Most Likely Cloud Phase (MLCP). All the products are provided at the EPIC original temporal and special resolutions. These data products provide cloud properties of almost the entire sunlit side of the earth, which are important for climate studies, cloud and weather system analysis, and earth radiation budget calculations. Data collection for this product is ongoing. Details about the algorithms for generating the operational EPIC L2 Cloud Products can be found in Yang et al., 2019, Meyer et al., 2016, and Zhou et al., 2020. A brief description is provided below: (1) The EPIC CM is based on the threshold method; surface is classified into three categories: land, deep water and snow/ice; CM with confidence level is determined independently for each surface type. (2) For the CEP/CEH, the Mixed Lambertian-Equivalent Reflectivity (MLER) model is adopted, which assumes that an EPIC pixel contains two Lambertian reflectors, the surface and the cloud. This assumption simplifies the radiative transfer equation and cloud pressure can be retrieved using the oxygen A- and B-band pairs. Since the MLER model does not take into account the effect of photon penetration into clouds, the retrieved cloud pressure is an effective pressure. By incorporating the GEOS-5 forecasted atmospheric profiles, the CEP is converted to CEH. (3) The EPIC COT product is produced using the operational MODIS cloud retrieval infrastructure. Since EPIC does not have particle size sensitive channels, a single channel retrieval algorithm was developed assuming fixed values for cloud effective radius (CER). In addition, cloud phase determination capability for EPIC is limited; hence the EPIC COT product provides two retrievals for each cloudy pixel, one assuming liquid phase and the other ice phase. A likely cloud phase is also provided based on the CEH.

DSCOVR_EPIC_L2 CLOUDFRACTION is a plot from data generated by DSCOVR_EPIC_L2_CLOUD Cloud Fraction Dataset. DSCOVR_EPIC_L2_CLOUD_03 is the Deep Space Climate Observatory (DSCOVR) Earth Polychromatic Imaging Camera (EPIC) Level 2 Cloud version 03 data product. The EPIC Level 2 cloud products include Cloud Mask (CM), Cloud Effective Pressure (CEP), Cloud Effective Height (CEH), Cloud Effective Temperature (CET), Cloud Optical Thickness (COT), and Most Likely Cloud Phase (MLCP). All the products are provided at the EPIC original temporal and special resolutions. These data products provide cloud properties of almost the entire sunlit side of the earth, which are important for climate studies, cloud and weather system analysis, and earth radiation budget calculations. Data collection for this product is ongoing. Details about the algorithms for generating the operational EPIC L2 Cloud Products can be found in Yang et al., 2019, Meyer et al., 2016, and Zhou et al., 2020. A brief description is provided below: (1) The EPIC CM is based on the threshold method; surface is classified into three categories: land, deep water and snow/ice; CM with confidence level is determined independently for each surface type. (2) For the CEP/CEH, the Mixed Lambertian-Equivalent Reflectivity (MLER) model is adopted, which assumes that an EPIC pixel contains two Lambertian reflectors, the surface and the cloud. This assumption simplifies the radiative transfer equation and cloud pressure can be retrieved using the oxygen A- and B-band pairs. Since the MLER model does not take into account the effect of photon penetration into clouds, the retrieved cloud pressure is an effective pressure. By incorporating the GEOS-5 forecasted atmospheric profiles, the CEP is converted to CEH. (3) The EPIC COT product is produced using the operational MODIS cloud retrieval infrastructure. Since EPIC does not have particle size sensitive channels, a single channel retrieval algorithm was developed assuming fixed values for cloud effective radius (CER). In addition, cloud phase determination capability for EPIC is limited; hence the EPIC COT product provides two retrievals for each cloudy pixel, one assuming liquid phase and the other ice phase. A likely cloud phase is also provided based on the CEH.

DSCOVR_EPIC_L2_CLOUDHEIGHT_01 visualizes the Deep Space Climate Observatory (DSCOVR) Earth Polychromatic Imaging Camera (EPIC) Level 2 Cloud version 03 data product. The image shows Cloud Effective Height (CEH) derived using Oxygen A and B-band pairs from the DSCOVR_EPIC_L2_CLOUD_03 product. The data is shown on an orthographic projection of the Earth, and a color map is used to indicate the altitude of clouds. CEP is derived using the Mixed Lambertian-Equivalent Reflectivity (MLER) model, which assumes an EPIC pixel contains two Lambertian reflectors, the surface and the cloud. This assumption simplifies the radiative transfer equation, and cloud pressure can be retrieved using the oxygen A- and B-band pairs. Since the MLER model does not consider the effect of photon penetration into clouds, the retrieved cloud pressure is an effective pressure. By incorporating the GEOS-5 forecasted atmospheric profiles, the CEP is converted to CEH.

DSCOVR_EPIC_L2_CLOUDHEIGHT_01 visualizes the Deep Space Climate Observatory (DSCOVR) Earth Polychromatic Imaging Camera (EPIC) Level 2 Cloud version 03 data product. The image shows Cloud Effective Height (CEH) derived using Oxygen A and B-band pairs from the DSCOVR_EPIC_L2_CLOUD_03 product. The data is shown on an orthographic projection of the Earth, and a color map is used to indicate the altitude of clouds. CEP is derived using the Mixed Lambertian-Equivalent Reflectivity (MLER) model, which assumes an EPIC pixel contains two Lambertian reflectors, the surface and the cloud. This assumption simplifies the radiative transfer equation, and cloud pressure can be retrieved using the oxygen A- and B-band pairs. Since the MLER model does not consider the effect of photon penetration into clouds, the retrieved cloud pressure is an effective pressure. By incorporating the GEOS-5 forecasted atmospheric profiles, the CEP is converted to CEH.

DSCOVR_EPIC_L2_AER_03 is the Deep Space Climate Observatory (DSCOVR) Enhanced Polychromatic Imaging Camera (EPIC) Level 2 UV Aerosol Version 3 data product. Observations for this data product are at 340 and 388 nm and are used to derive near UV aerosol properties. The EPIC aerosol retrieval algorithm (EPICAERUV) uses a set of aerosol models to account for the presence of carbonaceous aerosols from biomass burning and wild fires (BIO), desert dust (DST), and sulfate-based (SLF) aerosols. These aerosol models are identical to those assumed in the OMI algorithm (Torres et al., 2007; Jethva and Torres, 2011). Aerosol data products generated by the EPICAERUV algorithm are aerosol extinction optical depth (AOD) and single scattering albedo (SSA) at 340, 388 and 500 nm for clear sky conditions. AOD of absorbing aerosols above clouds is also reported (Jethva et al., 2018). In addition, the UV Aerosol Index (UVAI) is calculated from 340 and 388 nm radiances for all sky conditions. AOD is a dimensionless measure of the extinction of light y aerosols due to the combined effect of scattering and absorption. SSA represents the fraction of extinction solely due to aerosol scattering effects. The AI is simply a residual parameter that quantifies the difference in spectral dependence between measured and calculated near UV radiances assuming a purely molecular atmosphere. Because most of the observed positive residuals are associated with the presence of absorbing aerosols, this parameter is commonly known as the UV Absorbing Aerosol Index. EPIC-derived aerosol parameters are reported at a 10 km (nadir) resolution.

DSCOVR_EPIC_L2_AER_03 is the Deep Space Climate Observatory (DSCOVR) Enhanced Polychromatic Imaging Camera (EPIC) Level 2 UV Aerosol Version 3 data product. Observations for this data product are at 340 and 388 nm and are used to derive near UV aerosol properties. The EPIC aerosol retrieval algorithm (EPICAERUV) uses a set of aerosol models to account for the presence of carbonaceous aerosols from biomass burning and wild fires (BIO), desert dust (DST), and sulfate-based (SLF) aerosols. These aerosol models are identical to those assumed in the OMI algorithm (Torres et al., 2007; Jethva and Torres, 2011). Aerosol data products generated by the EPICAERUV algorithm are aerosol extinction optical depth (AOD) and single scattering albedo (SSA) at 340, 388 and 500 nm for clear sky conditions. AOD of absorbing aerosols above clouds is also reported (Jethva et al., 2018). In addition, the UV Aerosol Index (UVAI) is calculated from 340 and 388 nm radiances for all sky conditions. AOD is a dimensionless measure of the extinction of light y aerosols due to the combined effect of scattering and absorption. SSA represents the fraction of extinction solely due to aerosol scattering effects. The AI is simply a residual parameter that quantifies the difference in spectral dependence between measured and calculated near UV radiances assuming a purely molecular atmosphere. Because most of the observed positive residuals are associated with the presence of absorbing aerosols, this parameter is commonly known as the UV Absorbing Aerosol Index. EPIC-derived aerosol parameters are reported at a 10 km (nadir) resolution.

DSCOVR_EPIC_L2_GLINT_01 is Version 1 of the Deep Space Climate Observatory (DSCOVR) Earth Polychromatic Imaging Camera (EPIC) Level 2 glint data product. This product indicates the presence of glint caused by the single scattering specular reflection of sunlight either from horizontally oriented ice crystals floating in clouds, or from smooth, highly reflective water surfaces. Such glints can prevent accurate retrievals of atmospheric and surface properties using existing algorithms but can also be used to learn more about the glint-causing objects. The glint detection algorithm relies on the fact that EPIC takes images at different wavelengths at slightly different times. For example, red images are taken about 4 minutes after blue images. During these few minutes, the Earth’s rotation changes the orientation of the scene by one degree, which can affect whether EPIC observations at a specific wavelength will capture or miss the narrowly focused specular reflection from ice clouds or smooth water surfaces. As a result, sharp brightness differences between EPIC images taken a few minutes apart can identify glint signals. The glint product includes three parameters for each pixel in the part of EPIC images where the alignment of solar and viewing directions is suitable for sun glint observations: (1) The surface type flag shows whether the area of a pixel is covered mainly by water, desert, or non-desert land; (2) The glint angle—the angle between the actual EPIC view direction and the direction of looking straight into the specular reflection from a perfectly horizontal surface—tells how favorable the EPIC view direction is for glint detection and can help in estimating the distribution of ice crystal orientation; (3) The glint mask indicates whether or not glint has been detected.