Google-UCAR Africa Meningitis Belt Initiative
at NCSU

In an effort to aid the prototype Earth-gauging system UCAR Google initiative, the NCSU Climate Modeling Laboratory is performing dynamical downscaling of operational forecasts for the West Africa region. For this, the Weather Research and Forecast Model (WRF) Version 3 is currently being tested for optimization purposes and to diagnose added value that dynamical downscaling may provide to operational forecasts. In addition, we are providing real-time forecasts of relevant variables for West Africa and Ghana.

Preliminary Results

Dynamical Downscaling Advantage

Using the WRF climate mode at scales finer than the available coarse-scale operational models (1 degree), we recognize that running the model at 30km, for example, allows for the forecast of specific districts or other geopolitical divisions within a country. Figure 1 atests to the advantages of using finer scale forecasting. The arrows point to a district in northern Ghana and plots of relative humidity as generated in a WRF simulation at 30km resolution (left) and NCEP/NCAR reanalysis (NNRP1) at 2.5deg resolution (right). Some features worth noting include the generalization in NNRP1 with gridpoints roughly equaling a third of the country of Ghana. A similar image from the WRF simulation has much higher resolution with several gridpoints within the district Also, notice how the drier, cooler colors cover northern Ghana in the WRF simulation. Because this particular type of meningitis has been found to be so climate sensitive to changes in humidity regime, such changes in relative humidity can offer valuable information for stakeholders in the region in terms of supplies and transportation for inoculation purposes.

Figure 2 highlights the advantage of using the limited area Regional Model when compared against observed meterological conditions, large scale reanalysis and meningitis attack rate. We see that WRF (red) accurately reproduces the anticorrelation between onset of observed high humidity (blue) and the abrupt decline in the transmission of the disease (grey-black) over Kano, Nigeria. GCM-based data (green) overestimates humidity.
-----------
Attack Rate: No. cases / 100,000, per week
WHO Multi-disease Surveillance Centre, Ouagadougou, Burkina Faso



Figure 1. Model resolution comparison


Figure 2. Meningitis Attack Rates & Relative Humidity at Kano, Nigeria


Ensemble Forecasting

Efforts currently underway at UCAR will seek to use an ensemble of large scale models to provide operational forecasts for the Africa initiative. The NCSU Climate Modeling Laboratory will also explore the use of ensemble forecasting by downscaling each of the models in operational mode using WRF. Additionally, we will explore the relative skill and economic value of the ensemble prediction
forecast using the recently published EROC method. Currently underway is a WRF climate mode ensemble using Final Analyses (FNL) initial conditions and varying physics in the model to simulate multiple model ensemble. Real-time, initial-conditions varying forecasts are also provided in a bi-weekly timescale during the meningitis season. Click on the icon on right to access the forecasts.

Related Research

As part of the work at NCSU, seasonal and intraseasonal varibility of the boreal spring over West Africa is being studied. We employ several methods, including parcel trajectory analysis, data assimilation, Principal Component Analysis, among others.
Early results show the importance of different source regions for the airmass throughout the spring season. Source regions vary in the horizontal as well as vertical scales.






-----


Conference papers and presentations




Media




Share |

References and Links