KUCHING—Disaster management agencies across Sarawak must accelerate the adoption of artificial intelligence and data analytics to create “smarter, faster and more precise” early warning systems, Deputy Premier Datuk Amar Douglas Uggah Embas said on Monday. Addressing senior officials from the Department of Irrigation and Drainage (DID), the Meteorological Department and the state’s Hazardous Materials Unit, Uggah argued that climate-driven extremes are outpacing traditional forecasting models and that only algorithmic, data-driven platforms can close the critical gap between hazard detection and public alerts.
“We have reached an inflection point where historical patterns no longer predict future shocks,” Uggah told the closed-door technical round-table, according to minutes released by the State Disaster Management Committee (SDMC). “AI can correlate river-level sensors, satellite cloud imagery and social-media chatter in real time, compressing what used to be a three-hour modelling exercise into under ten minutes. Those extra minutes save lives.”
Sarawak’s 2.8 million residents are exposed to seasonal floods, landslides and transboundary haze; last November’s monsoon displaced more than 15,000 people and caused RM350 million in agricultural losses. While existing telemetry networks have improved since 2018, Uggah said the state still relies heavily on deterministic models that “assume tomorrow will behave like yesterday.” By contrast, machine-learning systems continuously retrain on fresh data, sharpening probabilistic forecasts and reducing false alarms that erode public trust.
The Deputy Premier’s call to action aligns with a global shift toward AI-augmented emergency management. The United Nations Office for Disaster Risk Reduction estimates that every dollar invested in early warning infrastructure saves up to six dollars in avoided losses. Google’s Flood Forecasting Initiative now covers 100 million people in 33 countries, while the European Centre for Medium-Range Weather Forecasts has deployed graph-neural-network models that cut wind-storm prediction error by 20 per cent.
Locally, Sarawak’s DID already operates 450 water-level sensors and 92 rainfall gauges, but data streams are siloed in separate control rooms. Uggah proposed a centralised “data lake” hosted on the state’s private cloud, with application programming interfaces (APIs) opened to universities and start-ups under a regulated sandbox. “We will not build a monolithic black box,” he stressed. “Instead we want modular micro-services—one for flash-flood routing, another for landslide susceptibility—each continuously benchmarked against global best practice.”
Funding mechanisms are under discussion in the upcoming state budget. Sources close to the Ministry of Finance told Borneo Post that an initial RM40 million has been earmarked for a three-year pilot, contingent on matching grants from federal agencies and development banks. A similar public-private model financed Selangor’s smart-sensor network in 2022, which now delivers SMS alerts in Malay, English and Chinese dialects within 15 minutes of threshold breaches.
Technical partners are already circling. Microsoft Malaysia confirmed it has held exploratory talks on deploying Azure-based geospatial analytics, while a Penang-based Internet of Things (IoT) start-up, Kiotron, has proposed edge-computing nodes that run AI inference on low-power micro-controllers, reducing back-haul bandwidth by 70 per cent. “Sarawak’s topography—mountainous interior, longhouses without fibre—demands edge solutions,” said Kiotron chief executive Ng Wei Liang. “We can solar-power a LoRa gateway that pushes a 64-kilobyte risk vector to the cloud every five minutes for less than USD 200 per site.”
Uggah also wants social-behavioural data woven into risk maps. Researchers at Universiti Malaysia Sarawak have shown that evacuation rates double when alerts reference nearby landmarks—e.g., “the flood will reach Sibu Hospital in 90 minutes”—rather than administrative village codes. Natural-language-generation models can now localise warnings automatically, a technique documented in a 2023 Nature paper on AI for humanitarian response.
Privacy and ethics safeguards remain contentious. The Sarawak chapter of the Malaysian Bar Council warned that aggregating mobile-phone location data for real-time evacuation tracking could violate personal-data laws unless anonymisation protocols are independently audited. Uggah conceded the point, noting that the state attorney-general’s office will draft an “AI governance charter” modelled on the EU’s AI Act, with mandatory impact assessments for any system processing biometric or location data.
Climate scientists say the urgency is justified. Dr. Lai Xiang Yung, a hydrologist at the Malaysian Meteorological Department, projects that extreme daily rainfall could increase 11 per cent by 2035 under intermediate emissions scenarios. “Static intensity–duration–frequency curves updated every decade cannot keep pace,” Lai said. “We need dynamic, AI-updated curves refreshed every storm season.”
Implementation timelines released by SDMC show a phased approach. Phase One (Q3 2026) will integrate existing DID and Met-station feeds into a unified dashboard; Phase Two (Q1 2027) layers in satellite synthetic-aperture-radar data to detect landslide subsidence within 24 hours; Phase Three (Q4 2027) adds predictive citizen-science inputs, allowing longhouse residents to upload geo-tagged photos of rising waters via a lightweight Android app that works offline.
Asked whether such sophistication risks excluding rural communities, Uggah pledged “no village left behind.” He cited the state’s ongoing rural 5G rollout—now covering 62 per cent of designated sites—as the backbone for last-mile alerting. Traditional sirens and mosque loudspeakers will remain, triggered automatically if digital channels fail. “Redundancy is non-negotiable,” Uggah said.
Regional cooperation will also deepen. Sarawak plans to share anonymised datasets with neighbouring Sabah, Brunei and Kalimantan through the Asean Coordinating Centre for Humanitarian Assistance, enabling cross-border AI models that anticipate how upstream rainfall in Indonesia translates into downstream flooding in Sarawak within 12 hours.
Ultimately, Uggah framed the initiative as economic as well as humanitarian. “Investors demand resilient supply chains,” he said. “If we can prove that AI-driven early warnings cut business-interruption losses by 30 per cent, we are not just saving lives—we are underwriting Sarawak’s future growth.”
SDMC will host a two-day “AI for Disaster Resilience” hackathon in Kuching this July, offering RM1 million in seed grants for teams that produce deployable prototypes. The state cabinet is expected to approve the full programme when the legislature reconvenes in August.
0 Comments