19 March 2024
12 March 2024
About
The Ice Cloud Imager (ICI) is a passive conically scanning radiometer set to be deployed on the upcoming EPS-SG satellite. ICI will operate at millimetre and sub-millimetre wavelengths, featuring 13 channels ranging from 183 GHz to 664 GHz. The primary mission objective is to quantify cloud ice, aiding climate monitoring and contributing to the parameterization of ice clouds in climate and weather models.
At the EUMETSAT Central Facilities, the main retrieval quantities to be produced include the ice water path (IWP), mean ice particle size by mass, and mean mass height. Previously, a retrieval algorithm based on Bayesian Monte Carlo integration (BMCI) was developed, presenting the inversion as a description of the posterior probability distribution. A cloud radiation retrieval database is a crucial part of this algorithm.
Objectives
The objective of the study was to create a comprehensive cloud radiation database for integration into the retrieval algorithm for ICI. The database comprises pairs of atmospheric and surface states alongside corresponding simulated ICI observations. The underlying physics and variability of the database should statistically represent reality and the database should demonstrate sufficient accuracy within the retrieval algorithm. Key features of the database generation method include considerations of instrument characteristics, various particle habits, and polarisation, among others.
Overview
A preliminary database was developed before the commencement of the study to assess the retrieval algorithm. The initial stage of this study involved a thorough evaluation of this preliminary database to identify any limitations or gaps. It was concluded that the development of a completely new database was essential for the study to address these shortcomings. Several requirements were placed on the new database, including accounting for polarisation, three-dimensional variability within cloud structures, and enhancements to the models for snow and sea ice surface emissivity.
Cases in the database are associated with atmospheric scenes. In order to represent the three-dimensional variability of cloud structure, three-dimensional atmospheric scenes were generated. Two-dimensional stretches of radar reflectivity from CloudSat were used as the primary source of information on the spatial structure of clouds, which were then combined with MODIS multispectral data and ERA5 atmospheric/surface data to effectively enhance the dataset and produce a three-dimensional scene. The scenes used in the study extended 2000 km in the along-track direction and 50 km in the across-track dimension.
The radiative transfer calculations within the simulations were performed using the Atmospheric Radiative Transfer Simulator (ARTS). Inputs to the simulation include sensor characteristic data, state-of-the-art atmospheric absorption models and surface emissivity models extended for snow and sea-ice surfaces. Various particle models, incorporating both habit and particle size distribution, are selected to ensure a statistically represented portrayal of reality. The incorporation of ozone interference within the new database marks a significant improvement, addressing a previous oversight in the preliminary database.
The introduction of particle polarisation into the database is another completely new feature. A randomised scaling factor was introduced to approximate the azimuthal orientation of particles. This enabled simulations to be conducted for both vertically and horizontally polarised ICI channels, thereby improving single polarised channels as well. Additionally, it facilitated the inclusion of polarisation resulting from surface effects within the simulations.
The data within the database include all-sky and clear-sky brightness temperatures for each ICI channel, along with antenna-smoothed cloud ice products. The simulations and antenna smoothing are consistent with data remapped onto the field of view of ICI-01V.
For validation purposes, ice water path (IWP) present in the database was statistically compared to DARDAR. Since actual Ice Cloud Imager (ICI) observational data were not available, observations from ISMAR/MARSS and GMI were simulated and then compared to existing observations. Inversions of real GMI observations were conducted as an additional test of the database generation techniques, demonstrating realistic spatial distributions consistent with DARDAR. Finally, the retrieval performance of the new database was evaluated. The performance was found to be highest in high-latitudes. However, a decrease in performance was observed with decreasing surface temperature or cloud ice mass height, which was expected due to the presence of cloud ice mass at low altitudes in high latitudes.
The final cloud radiation retrieval database comprised almost 10 million cases and covered approximately 50,000 atmospheric scenes. To the best of our knowledge, the database consists of the most realistic ICI simulations ever performed. The study was conducted by researchers from Chalmers University of Technology, and the Swedish Meteorological and Hydrological Institute (SMHI), both located in Sweden.
