Simulating and forecasting southern Africa's ocean

Development - Phase I

Phase I will focus on implementing, validating and disseminating data from an ocean forecast system. The aim is to provide unassimilated 5-7 day forecasts that are available for a wide variety of users by the end of Phase I. In setting up a regional ocean modelling system for southern Africa we will adopt a downscaling strategy, introducing a series of nested grids downscaling from the global model to a regional high resolution model, which will then provide boundary conditions for a very high resolution limited areas model. However, the present computing hardware capacity in southern Africa is not enough to run a global model operationally. Therefore the boundary conditions for the regional model will obtained by collaborating closely with international institutes that produce global operational products.

In WP-1 the global operational ocean forecast models used by MyOcean and the HYCOM Consortium will be thoroughly evaluated and their viability for use as the lateral boundary conditions of the regional system will be assessed. Using the output from global operational systems to force the boundaries of southern African regional domains will forge and maintain useful international collaborations. The regional and limited area models will be implemented using the Regional Ocean Modelling System (ROMS; Shchepetkin and McWilliams, 2005).

The objective of WP-2 will be to nest regional southern Africa model (SAM), with a horizontal resolution of 1/12 (~9 km) within the global grid (see black line in the Figure below). The model grid will be based on the SAfE configuration (Southern Africa Experiment; Penven et al., 2006), and its geographical boundaries will mimic those of the regional Agulhas model based on the NEMO code (Madec, 2006), developed by the DRAKKAR collaboration (e.g. Biastoch et al., 2008). A long-term hindcast simulation will be run in order to verify that ROMS is capable of reproducing all of the salient oceanographic features surrounding southern Africa. This validated hindcast simulation will provide valuable data for various studies, especially those with the aim of understanding oceanic variability around southern Africa.

In WP-3, a pilot study will introduce an additional nest within the SAM that spans the Benguela upwelling region (see above Figure, dashed white line) and resolves salient features there at a resolution of 1/36 (~3 km). Although the pilot study will focus on the west coast, its implementation will be generic and therefore applicable to any other region of interest. A validation procedure will be designed for hindcast, nowcast as well as forecast products in order to ensure that data of a known quality is supplied to users. In order to develop a robust validation system, it is essential that cooperative activities with the remote sensing and in situ observation pillars of the OceanSAfrica programme, as well as interaction with potential users, commence as soon as possible.

An important aspect of developing operational modelling activities, is the routine and consistent dissemination of the forecast products. The objectives of WP-4 are to develop the reliable dissemination of 5-7 day forecast products (via a website) that have been subject to a systematic and rigorous validation procedure in order to fulfil the vision that the data provided are of known quality. This will ensure visibility of the initiative, and hopefully attract users, developers and students to the project. Good communication with users, and incorporation of their feedback, is an essential component of developing useful ocean monitoring and forecast products. This will be addressed through outreach programmes such as user forums and workshops. Work Packages 5 to 7 are focussed on high resolution forecasts in coastal embayments and nearshore environments.

In Work Package 5 it is intended to implement high resolution forecast models on a coastal embayment scale, based on existing well-validated DELFT3D model set-ups at two pilot sites. The focus in Phase I is on developing appropriate assimilation infrastructure for forcing data streams (largely provided by the existing operation WAVENET and IPOSS systems) and appropriate forecast cycles and dissemination of model forecasts. In Phase II it is anticipated that the requisite open boundary conditions and high resolution atmospheric model results will be available within which to nest these high resolution models. This will remove existing limitations on model accuracy. It is anticipated that progress will be made through both Phase I and II in validating the ROMS systems at these higher resolution coastal embayments scales.

In Work Package 6 the focus is on implementing wave forecast models that will support forecasts in coastal embayment and higher spatial resolutions, i.e. the localised scales important to most users. Presently only certain wave parameters on a relatively coarse grid and for limited parameters are available from the NCEP operational model. The intention is to implement a large-scale regional operational WAVEWATCH model that provides the required wave parameters at appropriate resolutions to support wave forecasting in coastal and nearshore systems (see the WAVENET system).

Implementation of a biogeochemical modelling system at a coastal embayment scale is planned in Work Package 7. Potential implementation sites are St Helena Bay, Saldanha Bay and Algoa Bay. Initial model implementations will be developed during Phase I, however given the reliance of biogeochemical models on accurate hydrodynamic simulations at appropriate scales, it is likely that the dissemination of biogeochemical model will only occur during Phase II.