Consensus Seasonal
Forecast: Experience from SARCOF
Several international initiatives are in progress to develop models with stronger orientation to users. There are areas in the world, mainly the tropics, for which forecasting is more reliable and consequently seasonal forecasts are already considered to be relatively reliable for practical use. In Brazil, seasonal forecasts are also sufficiently reliable to enable farmers to plan for the anticipated drought. Lessons can be learnt for South Asia by examining forecast strategies applied in above mentioned areas. Globally, the research on seasonal products is increasing while the research related to seasonal forecasting in Bangladesh is just beginning. Some of the recent findings show a great potential for seasonal flood forecast in Bangladesh.

The success of seasonal forecasts requires a closer cooperation between the forecast providers and the forecast users. One example of a project designed to bring together forecasters and users is entitled SARCOF (the Southern Africa Regional Climate Outlook Forum), a pilot scheme in which many of the issues of forecast creation and provisions for estimation of benefits are being examined. The SARCOF process first operated prior to the 1997/98 wet seasons in southern Africa. The process consists of an annual program of three meetings, with the first meeting held in September prior to the onset of the wet season. The principal participants to the Forum are delegations from the National Meteorological and Hydrological Services (NMHS) of the Southern Africa Development Community (SADC) region, which have the ultimate responsibility to disseminate the forecast product to their user communities. Within a workshop format, technical and discussion sessions are held on global and regional climate dynamics, forecast methodologies and seasonal forecast presentations.

Based on the predictions provided by each of the forecast groups for different homogeneous zones of the region, a consensus forecast is achieved through discussions among the participating climate community. For each zone rainfall forecasts for the coming season are expressed as probabilities of occurrence of rainfall in three tercile classes, average, below average, and above average rainfall. This probabilistic product is then distributed to users. In December, a mid-season correction is held. Again, through a process of consensus an assessment of the early season forecast is made. This assists in the update of Forum forecasts for the remaining main wet season period. Finally a post-season meeting is held in April or May in which extensive validation of the Forum product is conducted and the process is debated within the context of user feedback. This allows the aims and methods of the process to be related directly to the requirements of a range of users.

An important component of the process is the capacity building exercises in which training on climate science and seasonal forecasting methodologies is provided by international and national experts through technical sessions at the Forum meetings. Although initially experimental, the SARCOF process is continuing each year and has been adopted worldwide as a model for seasonal forecasting. Regional Climate Outlook Fora are currently operating for other regions including south and Central America, other parts of Africa and South-east Asia (Harrison, 2000). Products from a number of RCOFs are available on-line (http://www.cpc.ncep.noaa.gov/products/african_desk/rain_guidance/).

It is envisioned that the model structure of SARCOF is an applicable tool to bring the forecasters and users together in Bangladesh. Moreover, SARCOF model handles on climate information (wet/dry), flood is essentially a component of total climate system -- here in Bangladesh the focus is on the combined effect of rainfall and run-off. Additionally, the socio-economic setting in Bangladesh, which is presumably agrarian, makes Bangladesh a potential candidate country that can benefit from the experience of SARCOF model.

Consensus Seasonal
Forecast: An alternative in Bangladesh
Decision-makers worldwide face a difficult challenge in developing an effective response to the threat of water-induced disasters. In Bangladesh, this task is more complex, not only because of the multitude of such disasters, but also because of the country's vulnerable geography. Major parts of Bangladesh occupy the deltaic plain of the Ganges-Brahmaputra-Meghna river system. Bangladesh is criss-crossed by a labyrinth of rivers. As the country is located at the lowest reach of the fluvial system, and in the area affected by coastal surges, flood disasters of different types are frequent, including river, urban and tidal or storm surge. It has been observed that the technical components of the FFWRS are relatively better developed in Bangladesh, but the non-technical (soft) components need more attention that can only be achieved by clear understandings and cooperation between various agencies. An operational framework for an ideally suitable FFWRS for Bangladesh may be synthesized in the following 5-stages:

Stage 1: Forecast Preparation -- On the basis of previous and current meteorological, oceanic, and hydrological conditions, this stage predicts the occurrence and magnitude of a flood in advance.

Stage 2: Interpretation and Stakeholders -- This stage identifies the probable impacts of flood upon vulnerable communities, to warn people about the onset of flood and its probable impacts.

Stage 3: Process of Dissemination -- This stage communicates and distributes the warning messages to disaster management agencies and vulnerable communities

Stage 4: Response -- This stage generates actions by concerned agencies and threatened communities for protection against flood hazards in response to warning.

Stage 5: Review and analysis -- This stage continuously monitors the performance of various components of the FFWRS for possible improvement.

(Note: A detailed description of these stages are available in Environmental Management 35:61, pp 716-725)

Concluding remarks
In order to enhance the existing FFWRS, Bangladesh currently faces two types of problems: i) seasonal flood forecasts -- the research of which has just begun, and ii) warning dissemination and response -- the research of which has yet to start. For the forecast part, Bangladesh has to rely on the latest ideas and technology from developed countries. Regular exchange of data and a close co-operation between the local institutions [e.g., Space Research and Remote Sensing Organization (SPARRSO), Bangladesh Meteorological Department (BMD), Flood Forecasting and Warning Center (FFWC) of the Bangladesh Water Development Board (BWDB)] and those of existing regional and international forecast centers would be very essential. For the dissemination part, the model structure of SARCOF as described and discussed in this paper in the context of Bangladesh could be an applicable tool that can bring the forecasters and users together. The advantage of this model is that it increases the skill of forecast for a better comprehension of seasonal flooding variability and change through providing significant opportunities to synthesize the regional climate dynamics ahead of time by virtue of meetings among the local, regional, and international institutions.

It is very important that there is a feedback between forecasters and users so that forecasters focus on providing what users need. Forecasters must explain what might be possible, and users must take on the responsibility of providing definitions of adverse weather, and appropriate values for costs of treatments and losses. The process of feedback can be strengthened by the participation of the flood response group (FRG) -- that can be formed by organizing Government agencies and NGO's working at the field -- who will periodically contact the local authorities with information about the error of forecasts. This information will then be transmitted to the 'Outlook Fora' for corrections. It is in this sense that this is the best way to ensure people participation in the total decision-making system.

In conclusion, it is not only prudent but also essential that the social aspects of FFWRS be thoroughly assessed. In some cases, it is necessary to re-conceptualize FFWRS, with the involvement of effective multi-disciplinary teams. There is no doubt that information and communication technology has opened up new avenues for accessible and rapid flood warning dissemination in a manner that can be designed to address the needs of the at-risk community more than ever before. The government should, therefore, strive to enhance the FFWRS for flood forecasting in Bangladesh.

Dr. Rashed Chowdhury is the Research Scientist (Faculty) of the Pacific ENSO Applications Center (PEAC), University of Hawaii, USA; Email: rashed@hawaii.edu