TO IMPROVE AND DEVELOP SAND AND DUST STORM FORECASTS, PREDICTION AND REANALYSIS DATASETS
In the last years, the development of global and regional Sand and Dust Storms (SDS) forecasts has intensified because of their potential to mitigate impacts upon transportation, energy production, health and agriculture. In addition, operational weather services have started to recognize the importance of representing dust in models to improve weather forecasts. Dust predictions and projections at longer time scales are important for climate prediction and can be of interest for seasonal health early warning systems, and long-term risk assessments for agriculture and other sectors. The development and improvement of short- and medium-range forecasting and long-range prediction capabilities include specific challenges that should be overcome.
This Strategic Goal applies to remain challenges specific to the development and improvement of short- and medium-range forecasting and long-range prediction capabilities based on five focus areas.
Besides the uncertainties in representing dust physical processes, there are enormous challenges in the availability and use of dust observations for data assimilation and evaluation in regional and global dust forecasts. Even these and other limitations, forecasts including data assimilation have shown significant improvements compared to forecasts whose dust initial conditions only depend on model estimates. Another important limitation for the advancement of operational dust forecasts is the lack of standardized evaluation procedures, suitable observations and poorly developed verification systems compared to numerical weather prediction (NWP). Dust and weather forecasts (at both regional and global scales) can be improved by further developing dust data assimilation methods combining new and forthcoming quantitative satellite products available in real-time, and developing procedures and protocols for the evaluation of forecast products.
Taking advantage of the multi-scale capabilities of the new generation of meteorological models, it is feasible over the next decade to provide 10-20 km global dust forecasts with embedded high resolution nests (3-4 km) over the major source regions (Northern Africa, Middle East, West and East Asia, North America, South America and Australia) within a single execution of the model. The 2-way nesting capability of some of these multi-scale models will allow the small-scale dust emission processes at source regions within the high-resolution regional nests to impact the intercontinental dust transport in the outermost global domain improving at the same time global dust forecasts.
The improvement of models will permit a better characterization of the dust cycle and, consequently, more precise forecasts. However, the models will always have some limitations related to the mathematical representation of the simulated processes, the lack of suitable observational data and the imperfect representation of the soil characteristics and state. Ensemble prediction aims to overcome these limitations through the description of the future state from a probabilistic point of view. Multiple simulations are run to account for the uncertainty of the initial state along with any inaccuracy of the model. Multi-model forecasting intends to alleviate the shortcomings of individual models while offering an insight on the uncertainties associated with a single-model forecast.
The ability to forecast dust at long lead times would enable better preparation for the risks presented by SDS. Several agencies produce long-range forecasts for temperature, precipitation and other climate variables for various regions with lead times of several weeks to several seasons. However, none of these agencies produces long-range dust storm predictions. However, this area of opportunity contains major challenges. Given the complicated dependence of dust generation upon slowly-evolving surface conditions (such as vegetation and land use), it remains an open question whether long-range predictions of soil moisture, wind and gustiness over dust source regions can be good enough to produce skilful predictions of dust storm frequency.
One of the challenges in studying dust aerosols and their impact is the paucity of direct in-situ measurements in the areas most affected by SDS. For example, coverage of weather-observing sites in Africa is sub-optimal with only 1/8th of what is considered as minimum coverage, and many historical climatic datasets are still on perishable media. In this respect, model simulations can complement remote sensing and in-situ observations and help address deficiencies in the observing system.