Mediterranean Forecasting System Toward Environmental Predictions

 

The MFSTEP project is aimed at the consolidation and further expansion of a Mediterranean marine forecasting system based upon the elements implemented during the previous EU funded MFSPP project.

In the framework of MFSTEP, GOS coordinates SST, and ocean colour data acquisition and processing in WP3 (Task 3200) and is responsible for the analysis of assimilation results and satellite ocean colour data within WP12. GOS contribution to MFSTEP consists in the:

 

Improvement of the cloud detection and of the operational algorithms (subtask 3210)

Development of daily SST products for data assimilation (subtask 3220)

Production of daily fields on the model grid before, during and after the TOP. (subtask 3230)

Re-analysis of past SST data from 1999 up to the end of MFSTEP (subtask 3230)

Analysis of the statistical relationships between physical and biogeochemical state variables (subtask 12610)

 

MFSTEP SST data
MFSTEP ocean colour data
MFSTEP scientific results

 

Task 3200 : Sea surface temperature

The NRT system developed for MFSPP will be upgraded as follows:

Subtask 3210: Improvement of the cloud detection and of the operational algorithms

The cloud detection methods will be improved in order to recover SST pixels that are now flagged as clouds. This part is essential for regional and coastal areas applications as most of the cloud detection algorithm have problems to discriminate coastal fronts and clouds. In addition, within this task, particular attention will be paid on the review and test of the operational algorithms applied for SST computation in order to improve the SST estimates. At present the best algorithm for AVHRR SST estimate is the delayed time Pathfinder algorithm (e.g. Kearns et al. 2000), which computes SST from monthly coefficients that are calibrated against in situ data within a temporal window of 5 months. A comparison of MFSPP operational SST fields with Pathfinder data showed a bias error with a clear seasonal behaviour (MFSPP special issue). As Pathfinder coefficients cannot obviously be available in NRT, tests will be made on MFSPP TOP period data to check the possibility of using previous year coefficients to remove part of this error. Available match-up database will be used to have a quality control of the data products and to test and validate the satellite SST data products for AVHRR sensors. If tests yield positive results, the new scheme will be adopted to reprocess MFSPP and MFSTEP data (subtask 3230).

 

Subtask 3220 : Development of daily SST products for data assimilation

The processing of AVHRR developed during MFSPP will be improved to allow a higher frequency assimilation of the sea surface temperature data in the MFSTEP model. Daily maps of SST (computed from night passages) instead of weekly means as chosen during MFSPP will be produced. Two types of products will thus be defined and developed: SST maps at the model resolution (1/16°x1/16°) with data voids in the areas covered by clouds by (to be used for the assimilation in WP8), and daily interpolated SST maps. The maps will be produced using a space/time objective analysis scheme that takes into account the presence of continents and islands and the SST spatial and temporal decorrelation scales typical of the Mediterranean basin.

 

Subtask 3230 : Production of daily fields on the model grid before, during and after the TOP. Reanalysis of past SST data from 1999 up to the end of MFSTEP.

Once the new processing chain will be developed, all the daily SST maps with data voids will be provided once a week (on Wednesday, the start day of forecast, see WP8), while interpolated maps will be produced in delayed time. In addition, during the TOP period (6 months), full resolution SST maps will be distributed.

The task will also include a reanalysis of MFSPP SST data to obtain a homogenous time series of SST data from 1999 to the end of MFSTEP. The data will be reprocessed with the Pathfinder algorithm, that for the period 1999-2001 will use the calibrated coefficients. Reanalysed data will be produced on the 1/8° model grid.

 

 

Task 12600 Estimation of covariance between physical and biogeochemical data.

Subtask 12610 Analysis of the statistical relationships between physical and biogeochemical state variables

The main objective is the analysis of model simulated fields at the light of available satellite ocean colour derived products: chlorophyll-a; normalised upwelling living radiance at 443, 490, 555 nm; diffuse attenuation coefficient at 490 nm (K-490). Ecosystem model derived quantities (e.g. mixed layer depth, vertical distribution of chlorophyll, etc) and ocean colour data will be used to study the statistical relation between space estimate of the surface chlorophyll (Optically Weighted Pigment concentration, OPW) and the vertical profile of chlorophyll. Moreover, satellite chlorophyll data will be analysed with in situ chlorophyll profiles available from project and other data sources. The final goal of this study is to optimise the assimilation methods and data product to be assimilated. Moreover the knowledge of surface signal and 3-d structure of the pigment distribution will certainly contribute to the improvement of satellite primary production estimates. Map of upwelling leaving radiance for the TOP phase of the project will be produced for the comparison with in situ data.