NA5: Clouds and aerosol quality-controlled observations
WP Leader: Anthony Illingworth

Objectives

This activity will focus on extending the scope of the FP5 Cloudnet project to include new stations so that eight sites will have the capability to make continuous quality controlled observations of clouds as before and to include aerosols, all observations to have quantified errors and to be available in near real time.

  • To extend the existing cloud observations network developed under FP5 Cloudnet project by incorporating new sites in Northern Europe and the Mediterranean region.
  • To extend the scope of the existing cloud observing network to include aerosols. 
  • To further develop and combine existing variational techniques for deriving quality controlled retrieved profiles of cloud and aerosol parameters with quantified errors in near real time (NRT). This is in response to requirements for activities such as the FP7 MACC project for operational model validation and, ultimately data assimilation and to evaluate CRMs within the FP7 EUCLIPSE project.
  • To implement new metrics for evaluating the performance of models in representing clouds and aerosols based on the newly developed equitable and unbiased skill scores.

The enhancement of the existing near-real-time data base and its access will be dealt with in WP6 and WP18.

 

 Task 5.1  Extension of the existing cloud observing network (TUD + UREAD, CNR, CNR, UHEL, NUIG, DWD).
The existing Cloudnet stations at Chilbolton (UREAD), Cabauw (TUD), Sirta (CNRS) and Lindenberg (DWD) will be extended to include four new stations at Mace Head (NUIG), Potenza (CNR), Sodankyla (UHEL), Hyytiala (UHEL).
The current stations make high resolution observations of vertical profiles with cloud radar, ceilometers, and radiometers. The first stage is target classification in terms of hydrometeors (liquid, ice or mixed phase). The second stage is to retrieve the cloud variables such as cloud fraction, cloud liquid water and ice water content held in operational models. The third stage is to map these retrieved parameters on to the grid of the six operational NWP models (Met Office (UK), MeteoFrance (F), SMHI(S), KNMI (NL), DWD (D) and ECMWF); they can then be compared with the profiles held in the model over the station. forecast mode and evaluate their performance.
The four new stations are Mace Head (IRE), Potenza (I), Södankylä (FL) and Hyytiälä (FL). The first two have recently been equipped with cloud radars and the final two are in the process of acquiring them. The purpose of this task is to install the systems operating at the existing cloud profiling stations at the new stations. Specifically this task will arrange workshops and visits to stations to:
i) Workshop to establish operational procedures, maintenance and calibration methods for the new radars, ceilometers and radiometers.
ii) Implement appropriate target classification algorithms for the specific instruments being used.
iii) Modify the retrieval algorithms for the specific instrument fit at that station.
   
 Task 5.2  Additional aerosol observations (IFT + CNR, CNRS, TUD, DWD)
In this task the existing Cloudnet FP5 infrastructure will be extended to include new and planned prognostic variables being used to represent both clouds and aerosols in the next generation of NWP, climate and CRM models. For example, in the FP7 MACC project, ECMWF will have prognostic variables for various types and sizes of aerosol. Recent technological advances mean that unmanned continuously operating lidars have been installed at the eight stations in the network. In this task, techniques for quality control, data processing and retrieval of aerosols will be developed and tested. In addition, existing EARLINET routines for deriving aerosol profiles, which are being further developed in NA2 and JRA1 to provide profiles of the microphysical properties
of aerosols will be adapted to run continuously in a Cloudnet environment.
Specifically this task will:
i) Hold a workshop to define a common set of retrieval algorithms to derive aerosol properties from lidars and define a common NetCDF format for all the data.
ii) Develop modified software for deriving the new variables
 
 Task 5.3  Development of new techniques for quantified errors of derived parameters (UREAD + all)
 Quantified errors are essential if the geophysical parameters derived from profiling observations are to be used for climatological studies, process studies, model evaluation, or, ultimately, for data assimilation. This task will develop further the variational technique clouds which uses the known errors of the ancillary instruments and the observed radar and lidar backscatter profiles to provide derived profiles of ice water content, ice particles size, etc, accompanied by quantified errors. Error calculation techniques have also been developed under EARLINET for the lidar retrievals of aerosols. In this task the techniques for aerosols, and ice clouds will be combined and extended to include liquid water clouds (in collaboration with WP22) and mixed-phase clouds. Specifically, this
task will:
i) Establish a common set of definitions for the instrument errors for the eight stations.
ii) Define requirements and protocols in a joint workshop with operational modellers and those in the FP7 EUCLIPSE project who wish to use Cloudnet and Earlinet type data to evaluate their ESM climate models, including SCM (single column mode) versions run in forecast mode.
iii) Establish a common set of variational algorithms for deriving geophysical variables for aerosols (e.g. extinction profiles) and for cloud profiles (e.g. cloud fraction, ice and liquid water content, ice particle size and concentration) and their error covariances.
And in collaboration with WP6 and WP18:
iv) Set up and test the infrastructure to store this data in a secure data base with long-term stewardship so that it can be easily accessed and analysed.
v) Implement a near-real-time access so that model and observations can flow to the data base, and the various parameters and their errors can be derived and stored.
 
 Task 5.4  Implementation of new metrics for evaluating model performance (UREAD + all)
 Evaluating model performance is not a simple matter. Simple parameters such as mean values of cloud fraction, liquid water content, and their pdfs and cloud top and bottom are very useful to give a general indication of how well the model is performing. However, many cloud variables have non–Gaussian distributions, so, instead of the commonly used ‘equitable threat score’, w0e will use the recently proposed (Hogan et al, 2009) ‘symmetric extreme dependency score’, which has the advantage that it is equitable, difficult to hedge, and independent of the frequency of occurrence of the quantity being verified. In this task the structure to implement such new metrics will be set up, so that the data obtained in the first two tasks can be rapidly analysed and the
performance of any new model parameterisation for clouds and aerosols can be evaluated. In this way we will realize the goal of providing rapid feedback within one or two months on any changes in model performance when changes are made in the model representation of clouds and aerosols. Currently, evaluation often involves targeted field projects which take several years to plan, execute and analyse which deliver their verdict on model performance after several years delay, by which time the model has probably evolved through many subsequent versions.
The work will involve:
i) A joint workshops with operational and CRM/LES modellers to define a comparison strategy for the observations and the model representation;
ii) Testing, development and implementation of the new metrics.
and, in collaboration with WP6 and WP18,
iii) The implementation and testing of the new metrics within the data-base so that monthly skill scores can be calculated and accessed by users.

Engagement with users:
Our principal users will be those running operational weather forecast models most of which are now incorporating aerosols as prognostic variables; they require the observations to evaluate their forecast models, and subsequently improve parameterisation schemes, and ultimately for data assimilation into the models. In addition those running CRM/LES (Cloud Resolving Models/Large Eddy Simulation Models) wish to use large continuous data sets which will be available from the system we are setting up. Accordingly, we have planned two workshops to which these users will be invited:
1) A workshop with operational weather forecasters and CRM/LES and aerosol modellers to define the variables they wish us to retrieve (see Task 5.3, iii, and deliverable 5.3.2)
2) A workshop with operational weather forecasters and CRM/LES/aerosol modellers to define the metrics needed to evaluate model performance when comparing with observations (see Task 5.4 i and deliverable 5.4.1.

 
Select the search type
 
  • Site
  • Web
Search
     Login