Bangladesh is in essence an agrarian country in a deltaic environment that relies on floods and rich soils to produce its annual grain requirement. However, early monsoon floods, late onset of floods, and other climate variations significantly impact food production and quality of life.
Recognition of the danger of unexpected flooding led to the development of short-range flood forecasts (24 to 72 hours in advance) more than a decade ago. While the role of short-range forecasts is important, they are insufficient to help vulnerable communities prepare, particularly in terms of dislocation of cropping practices.
What is needed is better comprehension of seasonal climate variability and change, the consequences of this for the key features of seasonal flood hydrography, predictability of these features, and improved translation of this information into production and coordination on the ground level.
What are seasonal forecasts?
By definition, seasonal forecasts are predictions of average seasonal conditions over a region that are made many months in advance based on slowly changing parts of the climate system. Analyzing ocean temperatures provides some ability to forecast average conditions months in advance.
If it is possible to create successful seasonal forecasts, then generation of a probabilistic outlook of flooding well ahead of time (three to six months before the occurrence of flooding) is also possible.
In the case of short-range deterministic flood forecasts, we can generate information three to seven days in advance, and probabilistic seasonal forecasts can produce information months in advance with reasonable accuracy. However, while the deterministic forecasts are mostly accurate because of their shorter time scales (hours to days), the probabilistic forecasts have some uncertainties because of their longer time scales.
What is a probabilistic forecast?
When reading a probabilistic flood forecast in three-tercile format, for example, 40:40:20 means the possibility (or probability) of higher-than-normal flooding is 40-percentage (upper tercile), normal flooding is 40-percentage (middle tercile), and lower-than-normal flooding is 20-percentage (lower tercile). Therefore, even if we forecast a 40% possibility for a higher-than-normal flooding, there is still a 20% possibility for a lower-than-normal flooding.
Compared with deterministic forecasts, this is the kind of limitation probabilistic forecasts have. However, while the deterministic method can produce forecasts only during the occurrence of an event (for example, the Flood Forecasting and Warning Center of the Bangladesh Water Development Board produces information on daily rise/fall in river water-level in a flooding season), the probabilistic method can produce forecasts in a hot/dry spring on the possibility of a rainy summer and monsoon flooding.
The El Niño-Southern Oscillation (ENSO) climate cycle, which has two phases – El Niño and La Niña – has been widely used to develop probabilistic seasonal forecasts. In Bangladesh, El Niño is associated with drought and La Niña with flooding.
What is La Niña?
La Niña refers to the appearance of colder-than-average sea surface temperatures (SSTs) in the central or eastern equatorial Pacific region (see the illustration below), the opposite to conditions during El Niño. It is a cold event where the SSTs become anomalously colder compared with the long-term average for the central and eastern equatorial Pacific.
La Niña episodes also feature large-scale changes in the atmospheric winds across the tropical Pacific, including increased easterly (east-to-west) winds across the eastern Pacific in the lower atmosphere, and increased westerly (west-to-east) winds over the eastern tropical Pacific in the upper atmosphere.
These conditions reflect an enhanced strength of the equatorial Walker circulation. When the Walker circulation is strong, the upper tropospheric winds in the Australasian region are easterly and, consequently, tropical disturbances are transported westward into the Bay of Bengal. Therefore, rainfall becomes very active in the region of western Pacific, including Bangladesh.
There is evidence of teleconnections between La Niña strength and seasonal climate anomalies (rainfall, flooding and cyclones) in Bangladesh, which normally faces a surplus of rainfall during La Niña years. For example, all previous La Niña years (1964, 1973, 1988 and 1998) recorded excessive basin-wide rainfall.
During any La Niña year, the trade wind strengthens and as a result rainfall increases significantly along the greater Ganges-Brahmaputra-Meghna (GBM) basins, causing flooding along the whole catchments. This, in turn, severely floods Bangladesh, as it is the lowest riparian country in these basins. The stream-flow of the major rivers in Bangladesh is the result of monsoon rainfall upstream in India.
How helpful are seasonal forecasts?
Although La Niña-based seasonal products are used widely and successfully for flood hazard management in one-quarter of the globe, the scientific research in Bangladesh related to seasonal products is just beginning.
The government and water experts in Bangladesh will decide how effectively they can use the products of seasonal forecasts. For example, the onset of this year’s La Niña was visible in March and April. At that time, the seasonal SST forecast showed a cooling tendency (an indication of borderline La Niña) for seasons July-August-September and August-September-October.
Currently, as the season advances, a trend of further cooling has been projected and a weak-to-moderate La Niña is likely to continue in season October-November-December. This information, which was available by April, can be utilized now with reasonable accuracy for probabilistic flood forecasting for Bangladesh.
To improve the forecast, the observed climate data for Bangladesh and state of the science global datasets for other climate features could also be used. This information could help develop a strategy to address stakeholders’ needs through the flood response group. Therefore, in addition to short-term deterministic forecasts, medium-to-long-term seasonal forecasts are essential in developing a real-time response plan for hazards management.
This would significantly enhance the agricultural support system in Bangladesh.