Task 3.7

OSE/OSSE (Observing System Experiment/Observing System Simulation Experiment) technology

Lead: CMCC (Tomislava Vukicevic)

Email address: Please login or register to view contact information.

The objective of this task is to improve technology for assessment of observing strategies with respect to spatio-temporal analysis of coupled biochemical and physical state in coastal regions. As a first major step toward the full-coupled state assessments, already existing technology for performing OSEs and OSSEs with high-resolution data assimilation (DA) systems in different coastal regions will be improved with respect to the biochemical transport analysis. This activity will draw upon, but will significantly extend, accomplishments of the pilot OSSE studies in the WP9 of JERICO(FP7)), which have focused on observation impacts on physical state at regional scales and have used sub-optimal OSSE configurations. In this task an advanced OSE/OSSE infrastructure will be developed, focusing on the high-resolution biochemical transport in coastal areas and using well-established standards for optimal configuration of OSSEs. This will ensure harmonization of the methodology and compatibility of assessments across diverse regions and using different DA systems. The technology developed in this task will be used in the JRAP4.6.

SubTask: 3.7.1 Biochemical transport in high-resolution DA systems (led by CMCC)
3D biochemical transport modules will be added to the high-resolution DA systems. The modules will be developed in AIFS-DA (Adriatic Ionian Forecast and DA System based on Ensemble Kalman Filter (EnKF) technique; CMCC) and German GALATON (Geesthacht Assimilation System; HZG) systems. A version of 3D transport model from the Mediterranean Sea Biochemistry Analysis and Forecast system will be imbedded into the AIFS-DA EnKF cycling on a grid of 2 km horizontal resolution and 121 vertical levels. The transport will be evaluated in the Adriatic and NW Med coastal areas using in-situ observations and the analysis of surface currents from HF radar observations in front of Gargano and in Gulf of Manfredonia. For the German Bight, a 3D transport module will be added by means of diagnostic module to the forecast model in DA at 1 km resolution and 21 sigma layers. Verification will be performed for typical wind conditions as well as neap and spring tide phases. This assessment will also include substances with specific buoyancy properties like, e.g., micro plastics. Transports associated with the wind and the tides will be analysed separately.

SubTask: 3.7.2 OSE/OSSE infrastructure (led by CMCC)
The new infrastructure for configuring optimal OSSEs will be developed using well-established standards (Halliwell et al., 2014) that require simulation and validation of a full-blown nature run and performing coupled OSE/OSSE for the existing observation network in each domain where OSSEs will be performed. The nature runs for producing simulated observations will be based on model simulations using a different forecast model or at least a significantly different version of the model than the one used in the DA system, and will be verified using observations, The error statistics for each OSSE will be calibrated based on the equivalent OSE, with both experiments being performed for the existing observation network. In this way, the assessments of value-added (i.e., the reduction of error) by potential new observations would be fully objective (the error would be measured relative to the quality of analysis with already available observations). This would ensure that the results produced in different regions and using different DA systems will be compatible. For the OSSEs in NW Mediterranean and Adriatic coastal regions (AIFS-DA) the nature run will be produced using very high-resolution unstructured-grid forecast model SANIF (South Adriatic North Ionian Forecast System, based on SHYFEM) that was especially designed for resolving the coastal ocean processes. HZG will carry out fraternal-twin experiments in the German Bight where a reference model run will be perturbed by applying modifications to the wind forcing, the bathymetry, the bottom roughness, and the boundary forcing. Nature runs and calibration of the error statistics in the South-eastern part of the Bay of Biscay (Ifremer/CNRS) will be done by ensemble simulations based on coastal model configurations (sigma vertical layers, horizontal resolution around 1-2Km). Perturbations on the reference run will affect model parameters (e.g. turbulent-closure and light extinction coefficients), wind forcings, and initial conditions.

SubTask: 3.7.3 Optimization of HF-radar DA for the tracer transport (led by CMCC)
The sampling characteristics of HF radar systems (high spatial and temporal resolution over relatively large areas) is quite different from other observations that are traditionally used in DA (e.g., tide gauges or altimeter) and thus requires dedicated assimilation methods. In this task the HF radar assimilations systems run by the different partners will be optimised with respect to the tracer transport analysis. HZG will work on the optimisation of 3D transport in the German Bight using additional information from tide gauges and ADCP data, salinity and temperature profiles from the FINO-1 and FINO-3 stations. An optimized model error covariance matrix will be estimated for the 3D current field and evaluated for different stratification conditions. Free model runs and analysis will be compared in terms of transports and langrangean trajectories. A radar observation model will be implemented in AIFS-DA EnKF in collaboration with CNR-ISMAR. This will be first evaluated in a quick OSSE assuming radar locations as for the actual radar observations in Gargano and in Gulf of Manfredonia (CNR-ISMAR) and then in OSEs for the same coastal region in WP4 CS6. The RM Spectrum method (Le Hénaff et al., 2009) which has been developed to optimise multi-parameter observing systems and so far has been applied only to analysis of temperature and salinity will be extended to estimation of the surface currents with HF radar observations to radar location in the Bay of Biscay (Ifremer/CNRS). The integration of SOCIB HF radar observations in data assimilation system will be evaluated in the Ibiza Channel (key section in the Balearic Sea) using the Western Mediterranean OPerational system, (WMOP, SOCIB, resolution of 2km). 

Deliverables:

Deliverable 3.11: Optimal OSE/OSSE infrastructure including biochemical transport (M12) The document will describe nature run simulations and their verification, and calibration of error for both equipment types. It will also report implementation and verification of the biochemical transport modules in DA systems, results of the transport analyses and their verification will be presented the targeted coastal areas and statistics for OSSEs via comparison between OSSE and OSE results for the existing observation networks.

Deliverable 3.12: Improved Radar DA technology for biochemical transport analysis (M18).The document will present results from demonstration and sensitivity experiments (either OSE or quick-OSSE type) on HF-radar assimilation and impact on the transport analysis for the same regions mentioned in deliverable 3.7.2