EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Flood preparedness decisions and stakeholders' perspectives on flood early warning in Bangladesh

Sazzad Hossain1,3, Hannah Cloke1,2, Andrea Ficchì1, and Elisabeth Stephens1
Sazzad Hossain et al.
  • 1University of Reading, Geography and Environment, United Kingdom of Great Britain and Northern Ireland (
  • 2Department of Earth Sciences, Uppsala University, and Centre of Natural Hazards and Disaster Science, Uppsala, Sweden
  • 3Flood Forecasting and Warning Centre, BWDB, Bangladesh

There is high temporal variability in the occurrence of the monsoon floods in Bangladesh during the South Asian summer monsoon. Detailed flood forecast information about flood timing and duration can play a vital role in flood preparedness decisions. The objective of this study is to understand different stakeholder perceptions about existing forecasting tools and data, and how these can support preparedness and response activities. Forecast users can be divided into three broad categories-national, sub-national and community level. The stakeholders working at national level are involved in policy making while the sub-national level involved in implementation of policies.  In order to identify the appropriate lead-time for better flood preparedness and the challenges in communicating probabilistic forecasts to users, semi-structured interviews with key stakeholders involved in various sectors of flood disaster management at national and sub-national level, community level household surveys, focus group discussions and a national consultation workshop were undertaken during the 2019 monsoon.

It was found all major stakeholders working at national and sub-national levels are aware of the availability of forecasts and receive flood forecasts from the Flood Forecasting and Warning Centre (FFWC). However, about 40% of the respondents at the community do not receive forecast information. Before the flood event, policy level stakeholders need to know the availability of resources and preparedness at the sub-national level for better response activities. On the other hand, sub-national level stakeholders of different government agencies act as a bridge between policy level and the local community. Existing short-range forecasts cannot provide information about the potential flood duration which is essential for resources assessment, mobilization and preparedness activities.

People living in the floodplain are aware about the flood seasons as it is an annual phenomenon. However, they can anticipate floods events only 2 to 3 days beforehand based on the available early warning and their risk knowledge. This short-range forecast can be used for some basic household level response activities such as protecting household equipment or moving their livestock to a safer place. It is essential to know the actual duration and flood extent for their agricultural decisions such as understanding when to transplant young crops into the field. The study found that all stakeholders need forecast information with a lead-time between 15 to 20 days for better flood preparedness decisions. People are likely to have seen deterministic forecasts so far and are not used to probabilistic forecasts with multiple scenarios for a same event. However, national forecast bulletins may include probability of flooding events based on a threshold known as flood danger level. Capacity development of the local community is necessary to improve understanding of the probabilistic forecast and overcome communication challenges.



How to cite: Hossain, S., Cloke, H., Ficchì, A., and Stephens, E.: Flood preparedness decisions and stakeholders' perspectives on flood early warning in Bangladesh, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-1367,, 2019

Display materials

Display file

Comments on the display material

AC: Author Comment | CC: Community Comment | Report abuse

Display material version 1 – uploaded on 01 May 2020
  • CC1: Comment on EGU2020-1367, Md Jamal Uddin Khan, 08 May 2020

    Thanks for such field work in Bangladesh. I have two relavent question regarding the adaptation of probabilistic forecast. From your field experience, do you think the common people will be able to comprehend the probabilistic forecasts, or a deterministic forecast with some error is more suitable for the people to understand? How about decision level?

    Second question is how the understanding of the flood among population differs between GBM and other deltas? (Like Mekong or Mississippi?)

    • AC1: Reply to CC1, Sazzad Hossain, 31 May 2020

      Answer to part 1:

      Thanks for your nice questions.

      First, if you want to increase forecast lead-time, you need to adopt probabilistic forecast. People need longer lead-time forecast for their preparedness activities. The community people are aware of the flood season as it is an annual cycle. However, they do not know the actual timing of floods in a monsoon. Flood may occur in July or August or even both July and August. This flood timing hampers their agriculture activities. From the ongoing weather condition and river water level rising trend, they can understand only few days ahead that floods are going to affect them.

      During our fieldwork, we explored the communication aspect of probabilistic forecasts. It can be presented in the form of spread like maximum, mean and minimum which can be understood by the higher-level audience at the national level. On the other hand, the community people like to have single value instead of multiple values. Our study shows that If forecast is given in the form of probability i.e. 75 % probability of flooding community people likely to use forecast in their decision. People want higher probability for their decision making. For instance, there were no floods in August during the 2019 monsoon though it was a peak monsoon month for floods. The probabilistic forecast at the beginning of August in 2019 showed a very high probability of no floods (Fig. 1). The community people will use this forecast information If this forecast is communicated effectively with them i.e. understanding them forecast uncertainty.  As community people (in the study area) have risk knowledge, the forecast with a certain probability will help the community to make their decision.



      Figure 1: GloFAS extended range forecast (a) 01 August 2019 and (b) 15 August 2019 (Source:

      It is difficult to give a deterministic forecast with the longer lead-time. Both deterministic and probabilistic forecast is associated with some uncertainties and users should be aware of theses. The forecaster responsibility is to communicate these uncertainties. Therefore, “a deterministic forecast with some error is more suitable for the people to understand” may give confusion to users.

      Answer to part 2:

      The second part is very important that basically indicates the risk of knowledge of people. The Floods in the GBM basins are due to monsoon climate, therefore it an annual phenomenon.  Floods are in the other river basins, for instance, Mekong is also due the monsoon climate. People living in the floodplain are familiar with the floods and its impacts. However, understating of floods may differ due to different societal context. We have not studied other river basin than GBM, therefore, it is difficult to make a specific comment on this issue. Based on our study, different hydrometeorological drivers control flood characteristics, and people’s understanding of differs in terms of timing, duration that likely impact their resources.

      • CC2: Reply to AC1, Md Jamal Uddin Khan, 31 May 2020

        Thanks for the detailed and illustrated answer. If I understand correctly from your reply, then probabilstic forecast tries to bridges the gap between uncertainty in the atmospheric forcings and the demand from the people and authorities. As you explained, the probabilistic forecast also demands a basic interpretation skill on the individual level. In other words, to ensure better communication of the information it will be necessary to reduce the probabilistic forecast to a simplified version for general public consumption from the forecaster side.

        Thanks again for experimenting with such approaches which has the potential to highly improve the understanding of long term forecasts, as well as make those forecast actionable from policy/administrative perspective.