The Earth Clouds Aerosol and Radiation Explorer (EarthCARE) mission is an ESA/JAXA multi-instrument mission to launch in 2023. The mission consists of a cloud-profiling radar (CPR), a cloud/aerosol lidar (ATLID), a cloud/aerosol imager (MSI), and a three-view broadband radiometer (BBR) covering both LW and SW bands. The mission will deliver a unique and powerful suite of cloud, aerosol and radiation products.
The EarthCARE lidar is called ATLID (ATmospheric Lidar) and is a high-spectral resolution lidar (HSRL). Like ALADIN (the HSRL lidar carried aboard Aeolus) ATLID operates at 355nm . However, ATLID is optimized for cloud and aerosol sensing, while ALADIN's main focus is on the retrieval of winds. Accordingly, ATLID will measure with a much higher resolution compared with ALADIN and possesses a depolarization channel (unlike ALADIN).
ATLID uses a Fabry-Perot etalon based design to distinguished the thermally broadened return from atmospheric molecules from the spectraly narrower return from clouds and aerosols. This allows for the determination of both the extinction profile as well as the extinction-to-backscatter ratio (also known as the lidar-ratio or S). This is in contrast to elastic backscatter lidars (e.g. CALIPSO) which must specify the S in order to derive the extinction profile. In principle, rather simple direct methods for retrieving extinction and S profiles using HSRL attenuated backscatters exist, however, they require high signal-to-noise ratios. In general, the low SNR of space-based lidars compared to their terrestrial counterparts, leads to the need for extensive horizontal smoothing windows.
Averaging intervals on the order of 100km are often defensible for aerosol fields, but certainly not for cloud observations. For clouds, intervals on the order of a very few kilometers or even finer are necessary. Therefore, it is necessary to separate the "strong" (e.g. cloud) and "weak"(e.g. aerosol) regions at high resolution before smoothing the attenuated backscatter returns. In the ATLID processing chain, this is largely accomplished by using the output of the Featuremask processor (A-FM) which uses adaptive thresholds and ideas drawn from image-processing to provide a high-resolution target mask.
After the cloud-screened signals are averaged, then direct techniques for retrieving the extinction and lidar-ratio profiles at low horizontal resolution are employed. To retrieve the cloud properties, a second pass employing a forward-modelling optimal estimation approach (using the low-resolution results as priors) is applied at high horizontal resolution. The processor that accomplishes this is called the ATLID profile processor (A-PRO). In addition to the retrieval of the optical properties, targets are also classified (e.g. water, ice, aerosol(type)) using e.g. extinction, temperature, lidar-ratio and linear-depolarization ratio) by the ATLID target classification procedure (A-TC) which is embedded within the A-PRO processor.
A-FM and A-PRO have been extensively tested using detailed simulated L1 data derived from the application of advanced lidar radiative transfer models and instrument simulators to high-resolution atmospheric model data. Adaptations of A-FM and A-PRO, called AEL-FM and AEL-PRO respectively, have also been created and successfully applied to ALADIN observations. In this presentation, A-FM and A-PRO will be briefly introduced, new developments highlighted, and various illustrative examples drawn from the set of simulated tests scenes presented and discussed. Additionally, example AEL-FM and AEL-PRO results will be presented.
The EarthCARE satellite will host a range of instruments, all optimised for improving our understanding of the interactions between clouds, aerosols and radiation. Over the past few years, it has been shown, with increasing confidence, that the cloud radar and lidar observations made by EarthCARE could have a direct impact on numerical weather prediction (NWP) forecasts if they are used in data assimilation to help initialise forecasts. Also, as global atmospheric models begin to resolve convective-scale features, the use of observations related to clouds becomes increasingly relevant. The relationship between EarthCARE and NWP data can be seen as symbiotic; the monitoring of satellite observations by comparing with the expected state of the atmosphere can be an invaluable tool for detecting and diagnosing problems before data products are released to the wider scientific community.
Previously, the studies at the European Centre for Medium-Range Weather Forecasts (ECMWF) investigating the potential for the direct impact of EarthCARE observations on NWP via data assimilation have been limited to radar reflectivity and lidar backscatter. In the first part of this talk, we will extend the capability to assimilate other EarthCARE observables, including Doppler velocity, Rayleigh backscatter and particulate extinction. The potential information content of each observation type will be discussed with reference to the Jacobians of the observation operators, which provide the sensitivity of the simulated observations to changes in the model state. The second half of the talk will evaluate the synergy between the suite of EarthCARE observations and the rest of the observing system at ECMWF by using A-Train observations. The interplay between active and passive observations in the minimization will be investigated, for example by assessing the impact of assimilating radar reflectivity on the analysis departures of AMSU-A microwave radiances. Finally, we will discuss future directions for maximising the synergistic power of EarthCARE in NWP, including the use of two-moment observation operators to reduce uncertainty and biases in the simulated observations.
Expanding the diversity of EarthCARE observations that can be simulated within the ECMWF model allows the potential for a more comprehensive quality monitoring system; the potential for further direct impacts of EarthCARE observations on forecast skill to be achieved if they were to be assimilated; greater opportunities for synergistic approaches to reduce uncertainty in the observations; and facilitates the use of the additional observations in model evaluation.
Aerosols are a key actor in the Earth system, with relevance for climate, air quality, and biogeochemical cycles. Aerosol concentrations are highly variable in space and time. A key variability is in their vertical distribution, because it influences aerosols life-time and, as a result, surface concentrations, and because it has an impact on aerosol-cloud interactions.
The instrument ATLID (ATmospheric LIDar) of the EarthCARE mission shall determine vertical profiles of cloud and aerosol physical parameters (altitude, optical depth, backscatter ratio and depolarisation ratio) in synergy with other instruments. Operating in the UV range at 355 nm, ATLID provides atmospheric echoes with a vertical resolution of about 100 m from ground to an altitude of 40 km.
On the side of model-oriented aerosol research (IPCC), the value of using a lidar aerosol simulator has been demonstrated, to ensure consistent comparisons between the modeled aerosols and the observed aerosols. In the current study, we make use of the lidar aerosol simulator implemented within the COSPv2 satellite lidar simulator (Bonazzola et al., in preparation) to estimate the total attenuated backscattered signal (ATB) and the scattering ratios (SR) that would be observed at 355 nm by the lidar ATLID overflying the atmosphere predicted by the E3SMv1 climate model. This simulator performs the computations at the same vertical resolution as the ATLID lidar, making use of aerosol optics from the E3SMv1 model as inputs, and assuming that aerosols are uniformly distributed horizontally within each model grid-box. It applies a cloud masking and an aerosol detection threshold, to get the ATB and SR profiles that would be observed above clouds by ATLID with its actual aerosol detection capability.
Our comparison shows that the aerosol distribution simulated at a seasonal timescale is generally in good agreement with observations, with however a discrepancy in the Southern Hemisphere, as the observed SR maximum is not reproduced in simulations there. Comparison between cloud-screened and non cloud-screened computed SRs shows little differences, indicating that the cloud screening by potentially incorrect model clouds does not affect the mean aerosol signal averaged over a season. Consequently, the differences between observed and simulated SR values are not due to sampling errors, and allow to point out some weaknesses in the aerosol representation in models. The use of several wavelengths can give further indication on the nature of the aerosols that need to be improved.
Supercooled and mixed phase clouds play important roles in high latitude radiation budgets. But models have difficulty in realistically simulating supercooled water clouds and mixed phase clouds. These biases in model clouds lead to local and regional biases in radiation which can produce errors in both weather forecasts and climate projections. CALIOP algorithms classify the ice/water phase of vertically resolved cloud layers and have been widely used to study the global distribution of supercooled water clouds, but CALIOP does not have a mixed-phase cloud type. Observations of high latitude mixed phase clouds needed to guide model improvements are limited to a handful of field campaigns and a handful of ground-based sites.
The current CALIOP cloud phase algorithm discriminates ice clouds from liquid clouds using 532 nm attenuated backscatter and depolarization signals. Both ice and liquid clouds produce depolarization of the lidar backscatter signal but in liquid clouds this depolarization results from multiple scattering, and from single scattering in ice clouds. One consequence is the correlations between the magnitudes of depolarization and backscatter signals are different for ice and liquid clouds. These signals can be visualized in a 2-D diagram and cloud thermodynamic phase can be classified using 2-dimensional thresholds. While mixed-phase clouds are common, in the current CALIOP algorithm cloud layers are only identified as either ice or liquid. However, the CALIOP phase algorithm is applied to layers detected at both 1/3km (single shots) and 5 km horizontal resolution. Investigation of the phase of 1/3 km layers detected within 5 km layers finds that both ice and water may be detected within a 5 km layer classified as liquid. We find the fraction of ice and liquid within 5km layers varies in a way that makes physical sense, and that 5 km cloud layers of mixed phase appear most frequently near the boundaries of the Ice and Water sectors of the current CALIOP cloud phase diagram.
Further information can be obtained from the CALIPSO IIR. The forward model used in IIR cloud retrievals is constrained by information from co-located CALIOP profiles. This approach is used to retrieve effective diameters of both liquid drops and ice particles, providing additional characterization of supercooled and mixed phase clouds that is complementary to the CALIOP measurements. This presentation will discuss the identification and characterization of mixed phase clouds using these observations from CALIPSO. These techniques can be applied to future ATLID and MSI observations, so these results are highly relevant to the EarthCARE mission.
Ice clouds play a crucial role in Earth’s radiation budget since they are semi-transparent to solar radiation while they are very effective in trapping thermal radiation from escaping into space. Active remote sensing with radar and lidar from space can help to quantify this process by measuring the spatial and vertical distribution of ice clouds on a global scale. Here, the variational algorithm VarCloud has been a workhorse for recent studies of global ice cloud properties to reconcile measurements from the A-Train satellites CloudSat and CALIPSO. Looking ahead, the upcoming ESA/JAXA satellite mission EarthCARE will give the opportunity to apply this variational framework on a single platform. With the launch of EarthCARE already on the horizon, the consolidation of a validation strategy of retrieved ice cloud properties is now of outermost importance.
Within this context, the German research aircraft HALO was equipped with an EarthCARE-like payload consisting of a high spectral resolution lidar (HSRL) system at 532 nm and a high-power cloud radar at 35 GHz. In case studies, coordinated flights were performed along other airborne platforms carrying instruments at different wavelengths or in situ probes. In combination with additional underpasses of the A-Train satellites, we learned numerous lessons when cloud properties are compared at different scales. Besides these case studies, we applied VarCloud to a month of airborne data acquired over the North Atlantic to test if a statistical comparison with ice cloud properties from the ERA5 reanalysis model is possible.
In this presentation, we want to share our insights into the effect of different spatial resolutions and different instrument sensitivities on retrieved ice cloud properties. By comparing retrieval results between platforms, we can differentiate between spatial resolution and instrument sensitivity effects. Besides the validation with scarce in situ data, the comparison with microwave constrained ERA5 reanalysis allows us to identify regions with potential biases. This enables us to anticipate if retrieved cloud properties can be compared on a statistical basis or only along coordinated flight legs and if conversions are necessary. By sharing our knowledge with the wider community, we hope to foster helpful discussions to consolidate an airborne validation strategy for EarthCARE.
The objective of the EarthCARE satellite mission is to help improve numerical predictions of weather, air quality, and climatic change via application of synergistic retrieval algorithms to observational data from its cloud-profiling radar, backscattering lidar, and passive multi-spectral imager (MSI). EarthCARE’s overarching scientific goal is to retrieve cloud and aerosol properties with enough accuracy that when used to initialize atmospheric radiative transfer (RT) models, simulated top-of-atmosphere (TOA) broadband radiative fluxes, for domains covering approximately 100 km2, “typically” agree with their observationally-based counterparts to within W m-2. “Observed” TOA fluxes derive from radiances measured by EarthCARE’s multi-angle broadband radiometer (BBR). As the latter are not used by retrieval algorithms, comparing them to modelled values, obtained by RT models operating on retrieved quantities, affects a “moderately stringent” verification of the retrievals; “moderately stringent” because BBR radiances consist, in part, of photons that also constitute to MSI radiances. This imperfection aside, this radiative closure verification is a well-defined and cost-effective final stage in EarthCARE’s formal “production model”. Its purpose is to both: i) provide, at all stages of the mission, quantitative feedback to algorithm developers regarding performance of their algorithms; and ii) help users focus, for whatever reason, on retrievals ranging from those that “appear” to have performed well to those that “appear” to have encountered difficulties.
This presentation discusses the essential features of EarthCARE’s radiative closure assessment. It begins with a review of the Scene Construction Algorithm (SCA) that applies a radiance-matching technique to MSI data and expands the ~1 km wide cross-section of retrieved profiles into the across-track direction thereby enabling application of 3D RT models that simulate TOA radiances and fluxes. It is worth noting that the EarthCARE mission appears to be the first Earth science project to apply 3D RT models in an operational sense; all others use 1D models. For continuity with previous missions, however, conventional 1D RT models are also applied to each retrieved column with results averaged over the same assessment domains used by the 3D RT models. The assessment domains measure ~5 km across-track by ~21 km along-track, Simulated TOA quantities are then compared with radiances measured directly by the BBR and fluxes inferred from those radiances by angular direction models that were developed specifically for EarthCARE. Results of these comparisons will be shown for a virtual observational system that consists of synthetic observations computed for surface and atmospheric conditions simulated by the Canadian Numerical Weather Prediction (NWP) model. These test scenes correspond to full EarthCARE “frames” that measure approximately 200 km by 6,000 km. There are three in total and they cover a wide range of cloud, aerosol, and surface conditions. Horizontal grid-spacing is 0.25 km, which includes some geophysical variability below the resolution of most of EarthCARE’s observations.