- A disaster risk reduction hub for the Hindu Kush Himalayas, was inaugurated last month, to safeguard communities, infrastructure, and ecosystem services from the increasing frequency and intensity of disasters.
- Implementing early warning systems is a strong focus area of the hub this year.
- Experts stressed on the need for a data-driven approach for impact-based forecasting of disasters and highlighted the role artificial intelligence and machine learning can play in developing early warning systems.
“Disasters do not stop at geographical boundaries. What happens upstream, affects downstream and vice versa. That is why we need regional cooperation in the Hindu Kush Himalayan region, which is shared by a group of countries,” said Pema Gyamtsho, Director General of International Centre for Integrated Mountain Development (ICIMOD), an international organisation based in Kathmandu, Nepal. He was speaking at the inaugural session of the launch of Hindu Kush Himalayan Disaster Risk Reduction (DRR) Hub, which aims to accelerate understanding, information-sharing and action to address the increasing number of disasters in the Himalayan mountains which are spread across 3,500 kilometres in Afghanistan, Bangladesh, Bhutan, China, India, Myanmar, Nepal, and Pakistan.
The launch of the DRR Hub on December 9-10, 2024, was held in Kathmandu. ICIMOD’s regional member countries and their governments, academia, practitioners, non-governmental organisations, UN agencies such as United Nations Office for Disaster Risk Reduction (UNDRR) and the United Nations Economic and Social Commission for Asia and the Pacific (UNESCAP), were in attendance.
The Hindu Kush Himalayan DRR Hub, with its secretariat at ICIMOD, aims to protect hill communities, infrastructure and ecosystem services from the rising numbers and intensity of cloudbursts and floods, landslides, glacial lake outburst floods (GLOFs), avalanches and droughts. There are shared risks and vulnerabilities across borders, which are exacerbated by climate change and hence, the need for a collaborative approach.
It is estimated that over 241 million people inhabit the Hindu Kush Himalayan region, the source of 10 major river systems, including the Indus, Ganges and Brahmaputra. But the region also faces high levels of poverty — 31% of the people live below the official poverty line of their respective countries and face food insecurity. Climate-induced disasters are further marginalising these vulnerable mountain communities.
“Hindu Kush Himalayas, also known as the Third Pole, provide ecosystem services to a quarter of the global population. But it is a global hotspot for disasters,” reiterated Gyamtsho. The region’s steep topography, fragile ecosystems, coupled with rapid socio-economic changes, growing population density and a changing climate, has increased its vulnerability to disasters. Scientists have already declared the Hindu Kush Himalaya, one of the most biodiverse regions on Earth, a ‘biosphere on the brink’.
Diana Patricia Mosquera Calle, Deputy Chief, Regional Office for Asia Pacific, UNDRR, said, “The United Nations’ Early Warnings for All initiative aims for universal coverage, but to meet the target of this ambitious and urgent global drive we need countries to come together. That’s why we’re very supportive of this initiative.” Launched in 2022, the Early Warnings for All initiative, aims to ensure that everyone on Earth is protected from hazardous weather, water, or climate events through life-saving early warning systems by the end of 2027. This is also a strong focus area of the DRR Hub this year.
Implementing early warning systems
The Hub plans to ensure that everyone is protected from hazardous weather, water, or climate events through life-saving early warning systems by the end of 2027 through disaster risk knowledge and management; detection, observation, monitoring, analysis, and forecasting; warning dissemination and communication; preparedness and response capabilities.
Arun Bhakta Shrestha, a Senior Climate Change Specialist with ICIMOD, pointed out how timely information sharing can prevent casualties during a disaster. “During the Yigong LDOF [landslide dam outburst flood] in 2000 in Tibet Autonomous Region, China, a lack of information sharing led to casualties. But during the Pareechu LDOF in 2004, information was shared and there was no loss of life, though there were damages.” Yigong is a tributary of Brahmaputra and Pareechu river is the upper reaches of Sutlej River in Tibet.
During the Yigong LDOF, due to a sudden increase in temperature, a huge amount of snow and ice melted, which led to a massive, complex landslide on April 9, 2000, in the upper part of the Zhamulongba watershed on the Yigong river. Within eight minutes, about 300 million cubic metres of debris, soil, and ice was dumped across the river bed, forming a landslide dam, which was 100 m high, 1.5 km wide (along the river), and 2.6 km long (across the river). The blocked river formed a lake behind the dam. An attempt was made to dig a large trench to release the water but this failed.
On June 10, 2000, the dam broke and led to a huge flash flood downstream, killing 30 people in Arunachal Pradesh, India, with more than 100 missing, as noted in an ICIMOD publication, Resource Manual on Flask Flood Risk Management.
Community-based flood early warning systems
Participants at the launch also shared their projects and experiences of working on early warning systems, which can be replicated in other countries.
Anju Jha, president of Mandwi, a non-profit that works on developing early warning systems with the local communities in the Terai region of Nepal which often faces flash floods. Mandwi has developed a community-based flood early warning system (CBFEWS) in the Lal Bakaiya River Basin in the Rautahat district of Nepal.
“There is a three hour lead time for flood water to flow from upstream parts of the basin to the downstream. Through a CBFEWS, we pre-warn the communities and save lives and reduce damages,” Jha told Mongabay India.
As part of the early warning system in the Lal Bakaiya River Basin, a water monitoring system at a highway bridge in Nijgadh and a rain gauge at Singaula, Nijgadh have been installed. Data from both systems help provide flood early warnings to communities living downstream, who have been trained through a series of capacity building workshops.
“The success of this initiative can be gauged from the fact that 13 flood-affected municipalities along the Lal Bakaiya River have recently signed an agreement to establish a basket fund for the sustainability of this early warning system. Each municipality will contribute to the basket fund, ensuring the continuous operation and maintenance of the early warning system,” said Jha.
Similar projects on CBFEWS have been implemented in India’s two most flood-prone states, Bihar (Ratu river) and Assam (Jiadhal and Ranganadi rivers). “Grassroots organisations and non-governmental organisations can only show models of early warning systems that work on the ground. The government needs to come forward and scale up such projects,” Partha Jyoti Das, head of the Water, Climate & Hazard Division of Aaranyak, a Guwahati-based non-profit, told Mongabay India.
Artificial Intelligence for early warning
In China, artificial intelligence and machine learning (AI/ML) are being used to forecast geo-hazards (landslides), said Xuanmei Fan, Director of the State Key Laboratory of Geohazard Prevention and Geoenvironment Protection (SKLGP), Chengdu University of Technology, China.
“There are nearly 300,000 geo-hazards in China, most of which occur in the southwest part of the country. These geohazards cause 700 to 1,000 fatalities every year,” said Fan. “We have mapped these geo-hazards and use low-cost sensors to collect real time data. The sensors send us three billion data sets daily, which is humanly not possible to analyse and we use artificial intelligence to do so. This way we have been able to forecast 400 landslides and save many lives,” she explained. These sensors are part of the LiDAR (Light Detection and Ranging) technology which China uses for identifying locations where landslides are likely to occur. LiDAR detects objects by emitting rapid laser pulses (like radar which uses radio waves) and using sensors to measure the time it takes for those pulses to bounce back after hitting surfaces.
This process is repeated a million times and it ends up producing a complex 3D map of the surveyed area, which shows in case there are any changes in the mountain landscape and its slopes. Since there are billions of data sets, AI plays a crucial role in enhancing cloud computing by automating the tasks of data analysis and improving overall efficiency.
Fan explained how if there is some movement on a hill slope, it can be detected through the 3D maps that are generated using LiDAR and AI enhanced cloud computing. According to her, the best model of an early warning system has to be based on AI, field monitoring, and remote sensing. “Quality of data is crucial for a good AI-based warning system. And data is still a challenge,” she acknowledged.
Sanjay Srivastava, Chief of Disaster Risk Reduction, UNESCAP, stressed on the need for a data driven approach for impact- based forecasting and highlighted the role AI/ML can play in developing early warning systems in the Hindu Kush Himalayan region. “Countries are already digitising early warning and risk communication. Cloud computing is a blessing for AI and AI/ML has a big role to play in early detection of an impending disaster,” he said.
Transboundary data sharing
Jamyang Zangpo, Senior Meteorologist/ Hydrologist Officer with the National Centre for Hydrology and Meteorology, shared concerns of the Himalayan kingdom. “Transboundary data sharing is crucial and Bhutan as an upper riparian country shares and transmits the river water levels and water information to the designated stations in India, on sub-daily and sometimes on an hourly basis. But, Bhutan has no data sharing arrangement with China in the north,” Zangpo highlighted.
Zangpo also revealed that Bhutan has a total of 567 glacial lakes of which 17 are potentially dangerous. The Bhutan government is working on GLOF early warning systems in areas where potentially dangerous glacial lakes are situated. For instance, PunatsangChhu basin has 11 of the 17 such glacial lakes. The early warning system there includes 10 automatic water level stations, 18 siren towers and a basin control room in Wandue.
Bangladesh, which is a lower riparian country, also highlighted the importance of transboundary data sharing. “93 per cent of the Ganga, Brahmaputra and Meghna basins lie outside of Bangladesh, hence transboundary data sharing is crucial for us” Jiban Kumar Sarker, Additional Chief Engineer, Bangladesh Water Development Board, stated at the launch
“During the monsoon, 20-25 percent of Bangladesh faces riverine floods. The northern and eastern zones of the country are prone to flash floods. There are already communication channels for in-situ data sharing with India and China. With Nepal it is going to start,” he said.
Read more: [Commentary] Up close and personal with the fragility of the Himalayas
Banner image: Raphstreng Tsho glacial lake. The Bhutan government is working on GLOF early warning systems in areas where potentially dangerous glacial lakes are situated. Image by ICIMOD via Flickr (CC BY-NC 2.0).
Credits
