KUALA LUMPUR: Inaccurate flood forecasting can erode public trust when individuals perceive warnings as unreliable, potentially decreasing the likelihood of people heeding accurate warnings in the future, says an expert.
The concerns follow the release of the Auditor General's report on Tuesday, revealing the National Flood Forecast and Warning Programme (PRAB) has fallen short of expectations, particularly in issuing flood warnings at least two days in advance and improving warning accuracy to within 0.5 metres of actual flood levels.
Universiti Sultan Zainal Abidin, East Coast Environmental Research Institute (ESERI) fellow researcher, Ir Dr Azimah Ismail emphasised the severe consequences of inaccurate flood forecasting from telemetric hydrology.
She highlighted that communities may become sceptical after constantly receiving inaccurate warnings, hindering the preparation for flood measures and thus impeding effective disaster recovery.
"People may not evacuate when necessary because of inaccurate forecasts; this then leads to displacement and evacuation issues. This, in turn, has potential consequences such as confusion, community disruption and added strain on emergency response resources," said Azimah.
Hence, she called for a long-term approach, urging meteorological and emergency management agencies to continuously enhance forecasting models and invest in technology and infrastructure for accurate and timely information on the importance of preparedness.
Azimah suggested that the government first invest in technology for advanced monitoring systems.
"The government should deploy monitoring technologies such as satellite imagery, remote sensing, and ground-based sensors to gather real-time data on weather conditions, river levels, sea level, soil moisture and other relevant parameters.
"Also invest in advanced data analytics and modelling tools to process the vast amount of data collected. Machine learning algorithms can help improve the accuracy of flood predictions by identifying patterns and trends.
"There should also be an infrastructure development of early warning systems by developing and maintaining a robust system that can rapidly disseminate alerts to vulnerable communities. This includes investing in communication infrastructure, such as sirens, text messages, and mobile applications, to reach a wide audience," she added.
Therefore, she said alert or warning systems must provide immediate notification to flood-prone areas, enabling evacuation to be carried out promptly, preferably within 24 hours.
Regarding the Auditor General's suggestions to the Drainage and Irrigation Department (DID), Azimah underscored the importance of stable data transmission from telemetric hydrological stations.
To address this, she emphasised the need for developing ICT infrastructure from 3G to 5G to enable real-time flood accuracy for the public.
"It is crucial for DID to develop a hydrological data tracking system, establish a database system, and upgrade ICT infrastructure from 3G to 5G.
"Developing a flood prediction and modelling system based on recorded rainfall data, studying precipitation trends, and in advance, creating a warning and dissemination system for the public in affected stations," she added.
Meanwhile, the Association of Water and Energy Research Malaysia (AWER) president, S. Piarapakaran, said inaccurate forecasting would also impact relief efforts.
"For example, if the National Disaster Management Agency (Nadma) receives a lower risk forecast, they would deploy gears that suits lower risk first
"However, when the actual risk is higher, these gears may not be suitable and would require realignment in rescue operations, wasting critical time for rescue," he said.
Piarapakaran echoed concerns about the impact of inaccurate forecasting on relief efforts and emphasised the need for addressing root causes such as outdated modelling data.
He said flood forecasting accuracy and an early warning time of more than two days are major problems that need immediate attention.
He also urged the government to form an expert team to review the modelling input data.
"When impermeable and permeable surfaces change due to land-use change, a flood forecasting model that was built using older data may fall short of the actual situation.
"Forecasting models are used to resemble the actual condition of a flood occurrence. If the gap gets wider or does not meet the expected results, we need to refine the methodology.
"On technology issues, contractors or project proponents should be aware of the ground zero operating conditions and changes taking place in technological advancements," he added.
Previously, the Auditor General's report highlighted forecasts for seven monsoon flood predictions were not issued a week in advance, and 51 warnings were given less than two days before the expected flood date.
Additionally, based on the Northeast Monsoon Report, 28 alerts showed variances exceeding 0.5 metres between predictions and actual occurrences.
Predictions, warnings, and notifications issued were also inaccurate, with 306 areas failing to align with forecasts and 183 areas encountering floods without corresponding flood warning bulletins.