The initial design for CMIP6 has recently been published [Meehl et al., 2014]. It includes a set of ‘DECK + CMIP6 Historical’ experiments to be run by modelling groups whenever they develop a new model version:
1% yr-1 CO2 increase up to 4xCO2
CMIP6 Historical run
These experiments are called the CMIP Diagnostic, Evaluation and Characterization of Klima (DECK) plus CMIP6 Historical experiments. In this document, we present the proposal of the list of COSP diagnostics to be requested to for the DECK + CMIP6 Historical experiments and additionally the CMIP6 CFMIP experiments. This proposal is the outcome of discussions by the COSP Project Management Committee (PMC) and the CFMIP Committee. The COSP diagnostic request for CMIP5/CFMIP2 is summarised and motivated in the CFMIP-2 proposal document [Bony et al., 2009], and documented in detail in the CMIP5 Standard Output documentation at http://cmip-pcmdi.llnl.gov/cmip5/output_req.html in excel spreadsheet format (Worksheet ‘CFMIP output’ indicates which tables appear in which experiments and for which periods, which other worksheets such as cfMon, cfDay etc indicate the variables in each table). Our view is that the CFMIP-2 diagnostics set fundamentally sound and forms a suitable basis for the COSP request for the DECK, CMIP6 Historical and CMIP6 CFMIP experiments, subject to some modifications. Thus, we present this proposal as changes with respect to the CMIP5/CFMIP-2 protocol in the accompanying spreadsheet. The request scoping sheet also shows which outputs are requested in which experiments. We have tried to address the concerns raised in the CMIP5 survey by simplifying the technical difficulty of the requests (sometimes at the expense of extra data) and basing the requests upon a frozen well-tested and already-released version of COSP (v1.4). In the sections below we present and motivate the specific requested changes.
2Description of proposed changes
2.1Change #1: Replacement of curtain data by full 3D fields, and deletion of cfOff table (proposed by Alejandro Bodas-Salcedo)
In CFMIP-2, the production of data along the A-train track (“curtain” data, table cf3hr offline) involved a substantial amount of post-processing. A second post-processing step required the gridding and time-averaging of these data to produce the monthly means requested in the cfOff table. This proved quite difficult for many modelling centres. Although not from the ESG archive, this type of data has been used in several model evaluation papers [Bony et al., 2009; Bodas-Salcedo et al., 2008; Field et al., 2011; Williams et al., 2013] involving case-study comparison of models with along-track observations from CloudSat and Calipso. We believe that by simplifying the request, the modelling centres will find easier to contribute these data. Hence, we propose to drop the orbital sampling, i.e. to request globally-complete fields on a standard lat/lon grid. Given this change, the calculation of monthly-averages from gridded 3-hourly data is trivial, and therefore we propose to delete the cfOff table.
2.2Change #2: New table cfMonExtra. Add CloudSat and CALIPSO CFADs to cfMonExtra (proposed by Alejandro Bodas-Salcedo and Mark Webb)
Optimisations to the code in COSP v1.4 mean that it is now practical to run the CloudSat simulator inline in models and so for longer periods. We propose the introduction of a new table cfMonExtra for the inclusion of monthly mean COSP diagnostics used for model evaluation in the AMIP DECK experiment, but which we don’t consider appropriate for coupled or climate change experiments. In this new table we include Cloud Frequency/Altitude Diagram (CFAD) diagnostics for CloudSat and CALIPSO for the entire AMIP integration. CFADs for CloudSat and CALIPSO have appeared in a number of published studies [e.g. Nam et al., 2014; Franklin et al., 2013; Bodas-Salcedo et al., 2011; Bodas-Salcedo et al., 2012; Nam and Quaas, 2012; Nam and Quaas, 2013; Kay et al., 2012; Kodama et al., 2012; Marchand et al., 2009; Abel and Boutle, 2012] and their inclusion as monthly means in the AMIP DECK experiment will make them available for analysts in a more convenient form than the higher frequency outputs currently requested in CMIP5.
2.3Change #3: Standard monthly COSP and daily COSP 2D outputs in all of the DECK, CMIP6 Historical and CMIP6 CFMIP Experiments (proposed by Mark Webb and Steve Klein)
Many of the standard monthly COSP and daily COSP 2D have been shown to be valuable in the CMIP5 experiments, not only for cloud evaluation [e.g. Franklin et al., 2013; Bodas-Salcedo et al., 2012; Nam and Quaas, 2013; Lacagnina and Selten, 2014; Bodas-Salcedo et al., 2014; Klein et al., 2013; Cesana and Chepfer, 2012; Tsushima et al., 2013] but also in quantifying the contributions of different cloud types to cloud feedbacks and forcing adjustments in climate change experiments [e.g. Tsushima et al., 2013; Zelinka et al., 2012a; Zelinka et al., 2012a; Zelinka et al., 2013; Zelinka et al., 2014]. We propose to include these in all of the DECK, CMIP6 Historical and CMIP6 CFMIP experiments as standard for the entire length of the runs, to support evaluation of cloud, cloud feedbacks and cloud adjustments and to investigate trends in the observational record.
2.4Change #4: Move PARASOL reflectance to cfMonExtra (proposed by Robert Pincus)
Top-of-atmosphere reflectance measurements from PARASOL were part of the standard request for CMIP5. They have been used in some applications [e.g. Nam et al. 2012] but have not been widely exploited. The proposal is to move them from the cfMon to cfMonExtra tables to reduce the number of integrations for which they are requested and to focus on model evaluation applications.
2.5Change #5: Add MISR CTH-COD to cfMonExtra. Add MISR CTH-COD and ISCCP CTP-OD histograms to cf3hr (proposed by Roger Marchand)
Histograms of cloud-top-height (or cloud-top-pressure) and optical-depth produced by ISCCP have been widely used in the evaluation of climate models, often in combination with the ISCCP-simulator now part of COSP. Because top-of-atmosphere outgoing longwave fluxes are related to cloud-top-height and outgoing shortwave fluxes are related to cloud-optical-depth this framework provides a way to evaluate the distribution of model clouds in a way that is closely related to their radiative impact. Similar histograms of cloud-top-height and optical-depth are being produced from observations by the Multiangle Imaging Spectro-Radiometer (MISR). While similar, the cloud-top-height in the MISR dataset is obtained using a stereo-imaging technique that his purely geometric and insensitivity to the calibration of the MISR cameras. This technique provides more accurate retrievals of cloud-top-height for low-level and mid-level clouds, and more reliable discrimination of mid-level clouds from other clouds, while ISCCP provides greater sensitivity to optically-thin high-level clouds. In addition, ISCCP and MISR histograms can be combined to separate optically-thin high-level clouds into multi-layer and single-layer categories [Marchand et al. 2010]. We therefore recommend using both ISCCP and MISR observations and instrument-simulators in the evaluation of climate model, and such an analysis is underway using a few CFMIP5 models that have run the MISR simulator [Hillman et al. 2014]. While monthly data are useful for the broad evaluation of models on monthly or longer time scales, the acquisition of high frequency (Three hourly) data will enable analysis of events that are not well resolved with monthly data, including the diurnal cycle, the Madden-Julian Oscillation (MJO) and various synoptic states or weather patterns, such as frontal passages. We recognize that this represents a large increase in data-volume compared with monthly averages and propose collection of this three hourly data only for a period of about 1 year.