ETCMIP

EarthCARE-Tropical Cyclone

Model Intercomparison Project

The EarthCARE–Tropical Cyclone Model Intercomparison Project (ETCMIP) aims to improve tropical cyclone (TC) prediction by using new satellite observations to better evaluate cloud microphysics and vertical motion in numerical models. By comparing EarthCARE observations with simulated signals in global storm-resolving models, this project will investigate how models represent the structures of dynamics and cloud microphysics in TCs. The resulting observational benchmark will help guide for improving cloud microphysical parameterizations and ultimately enhance TC forecasts.

Welcome to ETCMIP

If you would like to participate in ETCMIP, or if you wish to receive related information, contact the coordinators:

Masaki Satoh (University of Tokyo, satoh@aori.u-tokyo.ac.jp)

Jin-De Huang (University of Tokyo, jdhuang@aori.u-tokyo.ac.jp)

Robin Hogan (ECMWF, robin.hogan@ecmwf.int)

Experimental protocol

The tentative experiment protocol is organized into three stages:

Stage 1: Humberto case experiment

All participating models will simulate Hurricane Humberto as the first common benchmark case.
Sensitivity experiments will test different cloud microphysical parameters, especially hydrometeor fall speeds.
Model outputs will be evaluated against EarthCARE observations using radar reflectivity, Doppler velocity, vertical motion, and cloud particle distributions.

Stage 2: Multi-case hindcast experiments

The protocol will be extended to multiple TC cases observed by EarthCARE.
Cases will include TCs with different intensities and sizes to understand model sensitivity to TC characteristics and environmental conditions.

Stage 3: Seasonal free-run experiment

Models will conduct free-running simulations for a Northern Hemisphere TC season.
These simulations will be used to evaluate TC occurrence, track, intensity distribution, and cloud microphysical characteristics under less constrained conditions.
This stage connects process-level evaluation with seasonal-scale TC prediction.

Initial Data

The initial condition strategy for this project is currently under preparation. Ideally, participating models will be initialized from high-resolution operational analysis datasets, such as ECMWF operational analyses, to provide a realistic representation of TC structure and intensity at the initial time. ERA5 reanalysis may also be used as an alternative option because of its broad accessibility, although its lower resolution may limit the representation of intense TC inner-core structures. If a participating model has its own analysis system or is normally initialized from another analysis dataset, that dataset may also be used, provided that the initial condition source is clearly documented.

For the first benchmark experiment, all participating models are expected to simulate Hurricane Humberto. We are testing the sensitivity of this case to different initialization dates (left figure). Details of recommended initialization times and available initial datasets will be provided once the preparation is complete.

Output Data

The output data protocol will initially follow the structure of existing EarthCARE model intercomparison experiments. Because high-resolution TC simulations are computationally expensive, the priority is to provide outputs for the selected EarthCARE overpass periods and the most relevant TC case studies.

Mandatory 3D Variables

Temperature
Pressure
Horizontal wind components
Vertical velocity
Water vapor mixing ratio
Hydrometeor mixing ratios
Hydrometeor number concentrations for double-moment microphysics schemes

Mandatory 2D Variables

Sea-level pressure
10-m horizontal winds
Surface temperature
Precipitation
Outgoing longwave radiation
Outgoing shortwave radiation
Clear-sky radiation fluxes
Incoming solar radiation at the top of the atmosphere
Liquid water path
Ice water path
Precipitable water

Optional Variables

Cloud fraction
Precipitation fraction
Convective mass flux
Convective precipitation, if available
Large-scale precipitation, if available
Surface elevation or topography
Land-use data
Soil moisture

Data submission

JAXA will provide a data server for archiving the model output. Due to security requirements, data cannot be uploaded directly to the JAXA server. Instead, an intermediate server at The University of Tokyo is available for data transfer.

Please contact Takuji Kubota (JAXA; kubota.takuji@jaxa.jp), Masaki Satoh (The University of Tokyo), and Jin-De Huang (The University of Tokyo) for instructions on data submission.

Page updated

Google Sites