Structure of the EarthCARE project

The project is divided to the following work packages (WP):

Developments for assimilating cloud observations from radar and lidar (WP-1000)

WP-1100 Observation operator and observation processing for cloud radar

The work package focused on a number of areas: improvements to the radar forward operator, estimation of observation errors, developing of a quality control strategy, and setup of a bias correction scheme for CloudSat reflectivities. In each of them, the following results have been achieved:

  • The reflectivity forward operator ZmVar has been further developed improving the representation of scattering from frozen particles and implementing a model for multiple scattering.
  • Tangent-linear and adjoint versions of the updated reflectivity forward operator, which are required for data assimilation when using a variational technique, have been developed and tested.
  • Observation errors for CloudSat have been estimated considering separately the contribution from the instrument, the forward modelling and the spatial representativeness.
  • A strategy for quality control of CloudSat observations has been developed that rejects those data that could degrade the performance of the assimilation system.
  • Finally, a bias correction scheme has been put in place that is able to minimize the systematic differences between simulated and observed CloudSat reflectivity.

Report:

Di Michele, S., E. Martins, and M. Janisková, 2014: Observation operator and observation processing for cloud radar, WP-1100 report for the project Support-to-Science-Element STSE Study - EarthCARE Assimilation, AO/1-6441/10/NL/CB, ECMWF, 59 pp.

WP-1200 Observation operator and observation processing for lidar

In order to prepare for the monitoring/assimilation of EarthCARE lidar observations using CALIPSO data, the work package focused on development of a lidar forward operator, estimation of observation errors, development of a quality control strategy and a bias correction scheme for the CALIPSO lidar backscatter. The following results have been achieved:

  • The reflectivity forward operator ZmVar has been extended to simulate the lidar signal in clouds. The new lidar operator also includes a model to take into account multiple scattering.
  • Tangent linear and adjoint versions of the lidar backscatter operator have been developed and tested.
  • Observation errors for the CALIPSO lidar have been estimated considering separately the contribution from the instrument, the forward modelling and the spatial representativeness.
  • A strategy for quality control of CALIPSO observations has been developed in order to reject those data that cannot be used in the assimilation system.
  • In order to reduce the systematic differences between simulated and observed CALIPSO backscatter, a bias correction scheme has been prepared and tested.

Report:

Di Michele, S., E. Martins, and M. Janisková, 2014: Observation operator and observation processing for cloud lidar, WP-1200 report for the project Support-to-Science-Element STSE Study - EarthCARE Assimilation, AO/1-6441/10/NL/CB, ECMWF, 40 pp.

Data monitoring (WP-2000)

WP-2100 Monitoring of radar data

A basic framework for monitoring time series of radar observations has been established. Observations from CloudSat have been compared to the corresponding reflectivity first guess departures simulated from the ECMWF model. The study has shown:

  • After applying a quality control screening, there is a reasonable degree of consistency between simulated reflectivity and observations.
  • The extra information brought by the forecast model can be exploited to reduce the size of the minimum anomaly that could be detected through the continuous monitoring of the temporal evolution of statistical parameters derived by the observations.
  • Similar conclusions can be drawn when using reflectivity-derived cloud top height (CTH) as monitoring variable since the study has shown that there is a remarkable agreement between observed and simulated CTH.
  • Given the good agreement between the model first guess (FG) and observations, a warning system can be put in place where a set of thresholds can be defined for the monitoring quantities that are small enough to be used as checking limits in a warning system.

Report:

Di Michele, S., E. Martins, and M. Janisková, 2014: Monitoring of radar data, WP-2100 report for the project Support-to-Science-Element STSE Study - EarthCARE Assimilation, AO/1-6441/10/NL/CB, ECMWF, 22 pp.

WP-2200 Monitoring of lidar data

Similar as for radar observations, a basic monitoring system for lidar observations has been put in place. Time series of cloud observations from the CALIPSO lidar and corresponding first-guess departures simulated from the ECMWF model have been generated. The performed studies have shown:

  • When a quality control screening was applied, reasonable consistency between simulated and observed backscatter in clouds has been observed.
  • Through the continuous monitoring of the temporal evolution of statistical parameters for the differences between observations and their model equivalents, the extra information obtained from the forecast model can contribute to a reduction of the size of the detectable minimum anomaly.
  • A better agreement between observations and simulations has been shown when considering backscatter derived CTH as monitoring variable. Monitoring of the CTH departures has revealed very high correlation between CALIOP-derived and ZmVar-derived CTH which can be related to the ability of the model to reproduce the top of cloud.
  • Based on the continuous monitoring, a set of thresholds can be defined as checking limits to be used in a warning system.

Report:

Di Michele, S., E. Martins, and M. Janisková, 2014: Monitoring of lidar data, WP-2200 report for the project Support-to-Science-Element STSE Study - EarthCARE Assimilation, AO/1-6441/10/NL/CB, ECMWF, 21 pp.

Data assimilation and monitoring experiments (WP-3000)

WP-3100 Data monitoring experiments for radar and lidar

The feasibility of detecting possible problems with CloudSat and CALIPSO data has been investigated by monitoring statistics over longer period during which radar and lidar instrument anomalies have been simulated in the measurements. A basic alert system based on threshold levels has been designed to test if these anomalies could be identified thorough monitoring. Time series of CloudSat reflectivity and CALIPSO cloud backscatter together with their corresponding first-guess departures have been evaluated to investigate if the problems would be identified earlier in the FG departure time series.

In the radar case, the study has shown that:

  • Statistics of CloudSat observations are stable so that their time monitoring can provide indications of possible drifts.
  • The extra information brought by the forecast model leads to an earlier identification of a simulated drift in CloudSat observations.
  • The monitoring of the reflectivity-derived CTH is less efficient than reflectivity in identifying trends despite very high correlation between CloudSat-derived and ZmVar-derived CTH.

In the lidar case, monitoring experiments have indicated that:

  • A systematic change in the lidar calibration coefficient can be detected through analysis of the temporal evolution of lidar cloud backscatter.
  • Monitoring of the differences between observations and model equivalents does not improve the detection skill. Planned improvements in the description of clouds in the model and refinements to the lidar forward operator should reduce the discrepancies between the simulated and observed cloud backscatter, thus to lead to increased value of the monitoring of FG departures for data quality in the future.

Report:

Di Michele, S. and M. Janisková, 2014: Data monitoring experiments for radar and lidar, WP-3100 report for the project Support-to-Science-Element STSE Study - EarthCARE Assimilation, AO/1-6441/10/NL/CB, ECMWF, 27 pp.

WP-3200 Data assimilation experiments for radar and lidar

1D-Var assimilation experiments have been performed using observations of cloud radar reflectivity and lidar backscatter, either separately or in combination. Information on temperature and specific humidity retrieved from 1D-Var was used as pseudo-observations in the 4D-Var system to study the impact of these new observations not only on analyses, but also on the subsequent forecasts. The study has shown:

  • 1D-Var analyses get closer not only to assimilated, but also to independent observations. The impact of lidar backscatter from clouds is smaller than that of cloud radar reflectivity.
  • Analysis increments of both temperature and specific humidity are modified by the assimilation of cloud radar reflectivity and/or lidar backscatter. Comparing to radar, the lidar increments occur at higher altitude than the radar ones and are therefore complimentary.
  • 1D+4D-Var reduces analysis departures for specific humidity pseudo-observations and provides analysis departures closer to temperature pseudo-observations than would be obtained if temperature pseudo-observations were not used.
  • Impact of the new observations on the first-guess and analysis departure statistics when verified against other observation types assimilated in 4D-Var is rather small. There are also small, but systematic improvements coming from the lidar observations when combined with the radar.
  • Generally, a positive impact of the new observations on the subsequent forecast has been observed. Even though this positive impact decreases in time, is is still noticeable up to 48-hour forecasts.

Report:

Janisková, M., 2014: Assimilation experiments for radar and lidar, WP-3200 report for the project Support-to-Science-Element STSE Study - EarthCARE Assimilation, AO/1-6441/10/NL/CB, ECMWF, 56 pp.

Conclusions and recommendations

WP-4000

This work package provides:

  • conclusions from the work carried out during the project with an aim to assess the suitability of using spaceborne radar and lidar data for monitoring and assimilation in operational NWP models;
  • summary of work required to complete the work for possible operational assimilation and/or monitoring of EarthCARE radar and lidar cloud observations;
  • some recommendations for future technical and scientific developments required for lidar aerosol assimilation and data quality monitoring;
  • a brief analysis of the required work and possible benefits of assimilating the observations of EarthCARE Multi-Spectral Imager (MSI).

Report:

Janisková, M., S. Di Michele, A. Benedetti and S. English, 2014: Conclusions and recommendations, WP-4000 report for the project Support-to Science-Element STSE Study - EarthCARE Assimilation, AO/1-6441/10/NL/CB, ECMWF, 21 pp.

 

See also

Work breakdown