Atmospheric hazards

Robert Landis
World Meteorological Organisation, World Weather Watch Department, 41 Giuseppe-Motta, P.O. Box 2300, 1211 Geneva 2, Switzerland.
Tel.: +41-2-27 30 81 11, Fax: +41-2-27 34 23 26

Because of our reasonable level of understanding of the processes of the atmosphere, warnings of disaster-causing events can often be forecasted on time scales that enable significant reduction of injury, loss of life and damage to property. While the dissemination of atmospheric warnings is largely oriented to sub-national areas (i.e. local populations), the observation, detection and monitoring of the atmosphere must include scales that encompass the globe. Furthermore, observations are carried out both in situ as well as by remote sensing techniques, and can be earth-surface as well as space-based. Because most atmospheric phenomena have time and space scales that are highly related, the ability to prepare warnings and forecasts is largely dependent on three factors: (1) how well we understand the processes of the atmosphere including exchanges at its boundaries (e.g. at the ocean and land surfaces) as well as our ability to numerically represent and accurately reproduce these processes; (2) computational capacity to calculate the necessary time and space parameters (Lagrangian) required for predicting the future state of the atmosphere at resolution sufficient to be of use to the public; and (3) adequate and timely observation of the initial conditions of the atmosphere on a time and space resolution required for warnings. In order to evaluate and collate the large data sets needed for operational preparation of warnings and forecasts, relatively sophisticated techniques of data assimilation are used to represent the initial state of the atmosphere. These techniques can merge both space-based and surface-based information into one useable dataset and at the same time use historical or climatological data to help in data quality management. While the ability to observe, detect and communicate information, and process it into atmospheric predictions, can be automated, the final determination and preparation of warnings normally requires significant human intervention. This is because of the limitations in fully understanding atmospheric processes at high resolution where many atmospheric hazards occur, as well as our ability to assimilate many very high resolution data (e.g. from Radar) into atmospheric prediction models particularly on the meso-scale, often referred to as the ‘people scale’. The ultimate effectiveness of early warnings rest to a large degree on the ability to disseminate them at the local level. This is certainly one of the most common concerns of all warning systems. While technology such as the internet certainly helps on a large scale, it still cannot provide the needed direct dissemination of warnings locally. The best technologies are those that people of all types have close at hand. Normally this is the electronic media (e.g. radio and television), but even these have significant limitations particularly in developing countries. The challenge is to link competent and efficient preparation of warnings and forecasts by an official source to a proper local dissemination mechanism.