Cloud Droplet Number Concentration – satellite retrievals Advanced by Atmospheric models for Assessing Aerosol-Cloud Interactions – CDNC4aci

CDNC4aci will improve our current understand of the global distribution of the cloud droplet number concentration by developing and exploiting the results of a new method allowing for precise and dedicated retrievals of this properties from satellite observations

The development of this new method will be done in synergy between satellite observations and simulations of cloud fields from high-resolution models (spatial resolution of about 150-m).

Analytic (“fake”) satellite observations will be obtained from these simulations of cloud structures and corresponding estimates of the cloud droplet number concentration will be obtained. These estimates will then be computed to the model “truth” and biases can be estimated and reduced. This project aims to use this method in order to tend to cloud droplet number concentration retrievals that are as precise as possible for a large sort of cloud fields and atmospheric configurations.

Once the retrieval method is considered optimal, it will be applied to a large amount of satellite observations (over a long period of time) in order to create a new database of the cloud droplet number concentration.

From this database, combined with information on aerosol concentrations, we will propose an updated estimate of the radiative forcing associated with aerosol-cloud interactions. This estimation will be more precise than the precious attempts but also more consistent with similar estimations obtained from modelling only.

The first year of CDNC4aci project, in agreement with initial estimations, focused on two main aspects: i) the characterisation of an algorithm allowing to directly estimate the cloud droplet number concentration and ii) the simulation of realistic satellite observations obtained from high-resolution modelling. An important collaborative framework was done to achieve this objective. We have successfully formalised this algorithm and also adequately plugged the retrieval algorithm to cloud field simulations from the ICON model. In particular, a specific cloud scene from ICON simulations representative of both stratiform and cumuliform cloud types has been selected and focused on during this part of the developments.

A major result from this work was then the creation of a retrieval algorithm capable of simulated, based on high-resolution model outputs, observations from a large number of satellites. An important validation work was done to check these analytic observations again actual satellite measurements (this work was presented during the IRS symposium). This algorithm, named S3COM (Satellite Simulator and Sandbox for Cloud Observation and Modelling) has been relieved under a free license on the GitHub platform (in agreement with the data management plan) and is thus fully accessible for the scientific community.

Future work planned for the next trimester will focus on improving S3COM and in particular its optical representation of liquid clouds. Indeed, S3COM uses the radiative transfer code RTTOV (developed by EUMETSAT) that only treats in an approximate way the optical properties of liquid clouds. Based on a new collaboration with the Centre d’Études en Météorologie Satellitaire (CNRM) we will propose a significantly improved representation of liquid clouds in RTTOV. Then, the sensitivity of clouds to atmospheric parameters will be computed; leading to preliminary retrievals of the cloud droplet number concentration from 2023.

The S3COM algorithm was released on GitHub : github.com/odrans/S3COM. For the moment in v0.9-beta, but the version 1.0 is expected in October 2022. A R package allowing the community to conveniently run and use the outputs of S3COM is also published : https:// github.com/odrans/Rs3com . All the technical developments of CDNC4aci are here published under a free BSD-3 license in accordance with an open-science policy, and will allow for a close, transparent and facilitated collaboration with the international community.

An improved representation of liquid clouds in RTTOV is planned. These results could be implemented in the operational version of the code.

S3COM is being used in the context of the European project FORCeS (https://forces-project.eu/) to compute the sensitivity of satellite observations to cloud fields, in the view of developing artificial intelligence applications.

It is planned that the results of CDNC4aci will benefit the work of a Master’s thesis and a future PhD expected to start in October 2022. The S3COM algorithm will then be used to quantify the radiative impact of aerosol above clouds (not studied in the context of this project).

Two publications referring to the CDNC4aci funding are being revised:
- S. Dipu, M. Schwarz, A. Ekman, E. Gryspeerdt, T. Goren, O. Sourdeval, J. Mülmenstädt, J. Quaas: Exploring satellite- derived relationships between cloud droplet number concentration and liquid water path using a large-domain large- eddy simulation, TellusB, in revision.
Goren, T., G. Feingold, E. Gryspeerdt, J. Kazil, H. Jia, J. Kretzschmar, J. Quaas, Aerosol Tune the Stratus-to-Cumulus Transition, GRL, revised

Aerosol-cloud interactions imply an effective radiative forcing that is a key uncertainty when understanding and interpreting observed climate change. Global data are needed to better quantify the relevant processes, but a key quantity - the cloud droplet number concentration (CDNC, Nd) - is not available from operational products. Building on preliminary work, CDNC4aci will work towards reliable retrievals of Nd from satellites in close observations - model interaction: newly-available cloud-resolving simulations will inform the retrieval development and refinement, and the data, in turn, will be used to improve understanding and quantification of aerosol-cloud interactions in the model and from statistical analysis. Specifically, the project will include multi-angle and polarimetric observations for better Nd data, it will revise retrieval approaches using model-informed cloud vertical stratification conditioned on cloud regime and thoroughly quantify and correct retrieval errors and biases and assess aerosol-cloud interaction processes from data. The project will assess the cloud-process information in the retrieved data using model sensitivity analyses, it will make model and data comparable by forward-simulating measured polarized radiances and retrieval products, and assess aerosol-cloud interactions in a global model evaluated using the data and the process understanding in model-data assessment. The final goal is a consistent quantification of the aerosol-cloud forcing between model and data analysis.