Researches on flood monitoring systems worldwide

There are many systems used in practice by disaster management agencies for monitoring flood levels across the world. However, most of these systems are not only very costly and sophisticated to be used and maintained, but also lack of realtime monitoring of water levels which might lead to overflow in the canals and flash floods. With the aim of designing a flood warning system that is cost-effective, efficient, and user-friendly, a research team comprising Yasir Hashim, Abdul Hafiz Bin Mohd Idzha, and Waheb A. Jabbar from the Faculty of Engineering Technology at the Universiti Malaysia Pahang, successfully designed and implemented a Wireless Flood Monitoring System. This system utilizes ultrasonic sensors combined with a Arduino microcontroller to measure the water level and identify the situation whether it is “safe”, “cautious” or “dangerous” based on the predefined levels. The implemented prototype alerts individuals and/or authorities separately via SMS using GSM module or Bluetooth. In addition, the monitored water level data can be visually displayed using a LED.

Flood Monitoring system block diagram was conducted by Yasir Hashim, Abdul Hafiz Bin Mohd Idzha, and Waheb A. Jabbar

The application of new technologies in flood monitoring and dissemination has also garnered significant interest and research attention from scientists outside the hydro-meteorological field worldwide. For example, a group of IT engineers at the Jalgaon College of Engineering, Thailand, has applied IOT platform to develop a real-time monitoring system for water level, flow rate, and rainfall to monitor floods in Nakhon Si Thammarat province, located in southern Thailand. Designed to deliver real-time alerts to both authorities and residents, the developed system is composed of three major components: (1) sensor network, (2) processing/transmission unit, and (3) database/ application server.

These real-time water condition data can be remotely monitored through a wireless sensor network that utilizes mobile General Packet Radio Service (GPRS) communication to transmit measured data to the application server equipped with VirtualCOM software, enabling the server to communicate with remote sensors. The system, implemented with the latest Internet of Things (IOT) technology, delivers alert information through a smartphone application, capable of seamlessly sending any warning information to residents via wireless networks. Additionally, the device displays alerts using LED lights and activates sirens to alert both authorities and residents in the vicinity of the installation area.

Incorporating radio frequency in the development of an early flood warning system has been undertaken by a research team composed of A. Finawan, R. Tahir, E. Eliyani, A. Fauziah, and A. Jannifar, affiliated with the Departments of Electrical Engineering and Department of Mechanical Engineering at Banda Aceh Institute of Technology, Indonesia. The flood Early Warning System comprises two interconnected parts: (1) the sender, responsible for transmitting flood status using a flood detection sensor and a flood data sender, and (2) the receiver, including a receiving radio, siren, sound module, and speaker. This system makes it possible for a disaster information center or disaster management center installed in a local to send disaster information simultaneously and immediately to residents to protect public safety when a disaster occurs.

The flood Early Warning System was developed by A. Finawan, R. Tahir, E. Eliyani, A. Fauziah, and A. Jannifar

The provided information indicates that flood monitoring and warning system research activities are a prominent focus for numerous researchers and developers in various countries within the region. These systems share certain inherent features, notably: the provision of real-time alerts directly to residents through different alarm levels, facilitated by water level sensors connected wirelessly to processing units, data storage, and the dissemination of warning information to residents via mobile applications, speakers, and signal lights.

Researches on flood warning systems in Vietnam

Floods are the most catastrophic natural phenomena in Vietnam, typically occurring after heavy downpours, tropical depressions, or the landfall of storms. These disasters expose affected regions to the compounding effects of multiple natural hazards, resulting in profound consequences, including significant impacts on human lives, property, material assets, and infrastructure. In response to the perilous, unpredictable, and challenging-to-forecast nature of flood-related disasters, numerous projects and research initiatives aimed at developing flood monitoring and warning systems have been implemented in recent years. 

The hydrological station 308, a part of the Nam La - Nam Pan flood and landslide warning system

Notable among these projects is the initiative by the Vietnam Meteorological and Hydrological Administration (VNMHA) in 1998 to construct a flood warning and forecasting system for the Nam La and Nam Pan river basins, catering to the forecasting and warning needs of flash floods in Son La city and Mai Son district (Son La province). The deployment of this system took place between 1999 and 2000, with official activation in 2001 under the supervision of the National Centre for Hydro-Meteorological Forecasting, in collaboration with the Northwest Regional Meteorological and Hydrological Station. This initiative marked a significant milestone as the first system within the industry to employ automated measurement and transmission technology. However, challenges and limitations were encountered in equipment selection and station construction.

Nevertheless, since its inception, the system has demonstrated timely and accurate forecasting of rainfall and floods in the Nam La and Nam Pan river basins, thereby minimizing human and property losses during flood events in the basin. Particularly noteworthy was the flood event on September 26, 2008, where flash floods occurred simultaneously in numerous areas within Son La city and Mai Son district. The floodwater level in Son La city reached alarm level III, while at the Hat Lot Bridge in Mai Son district, it matched the peak level of the flood on July 27, 1991. The timely and accurate bulletins issued by the Station on rainfall and flood conditions resulted in minimal casualties, property damage, and livestock losses. From August 28th to 30th, 2018, an exceptionally large flood event occurred in the Nam Pan River basin, registering a magnitude of 3.57 meters. The peak floodwater level, reaching 517.28 meters at 4:40 PM on August 30th, surpassed the historical flood level of September 2008 (516.77 meters) by 0.51 meters. The monitoring stations within the Nam Pan River basin promptly provided data, aiding in effective forecasting, warning, and disaster management efforts in Mai Son district. Despite the occurrence of a historically significant flood event, minimal damages to both human life and property were reported.

In 2009, phase 1 of the project "Investigation, Survey, Zoning, and Warning of Flash Flood Potential in the Mountainous Regions of Vietnam" was implemented by the Vietnam Institute of Meteorology, Hydrology and Environment (now known as the Viet Nam Institute of Meteorology, Hydrology and Climate Change). As part of the project, flash flood risk zones were mapped at a scale of 1:100,000 for 14 northern mountainous provinces, and flood warning maps for 37 river basins and streams within these provinces were created at a scale of 1:5,000.

To bolster flash flood alert systems for smaller river basins, the project implemented the Threshold Line method (CL method) in accordance with the guidelines provided by the Japanese ministry of Land, Infrastructure and Transport. This method aimed to establish flash flood thresholds by examining the relationship between cumulative rainfall and maximum rainfall intensity during precipitation events. Consequently, 37 CL threshold diagrams were devised for 37 river basins. However, due to sparse and brief rainfall data recorded by the rain gauge stations, these diagrams did not accurately represent the actual flash flood thresholds for each location.

Although the project received commendation upon conclusion, several deficiencies persisted. These encompassed the limited effectiveness in supporting flood prevention efforts, the small scale of the maps, which inadequately addressed the majority of locations at the commune or village level. Furthermore, flood risk zone maps predominantly relied on static data, lacking integration with Geographic Information Systems (GIS) for dynamic flash flood warning purposes.

From 2012 to 2017, Phase 2 of the project "Investigation, survey, and mapping of flood-prone areas in the Central Highlands, Central regions of Vietnam, and the establishment of a pilot system to provide warnings for high-risk areas prone to floods, serving disaster preparedness and climate change adaptation planning" was carried out by the the Viet Nam Institute of Meteorology, Hydrology and Climate Change. The project has been implemented with the aim of: Developing current status and flood risk mapping at a scale of 1:50,000 for 23 major river basins covering 19 provinces and cities in the Central Highlands and Central regions; Establishing rainfall threshold maps capable of causing floods in the Central Highlands and Central regions to support flood warning efforts; Establishing the Vietnam Flood Forecasting and Warning System (VNFFG), which includes two sub-systems: the VNFFG flood warning operational system, operating by utilizing real-time online rainfall estimates, and the VNFFG analysis system, updating system parameters and analyzing warning results as well as intermediate information. The Vietnam Flood Forecasting and Warning System is based on the Flood Forecasting Guidance (FFG) approach tailored specifically to Vietnam. The system is partially open and subdivides the watershed system into more detailed sub-basins (ranging from 10-30 km²) to proactively issue warnings and align with the characteristics of flood generation.

In the domain of scientific research, application, and system design, the emergence of flood warning systems in Vietnam has been observed since as early as 1995. Over this period, a number of notable research endeavors have been undertaken. For instance, Artificial Neural Networks (ANN) were applied by Nguyen Huu Khai (1995) to simulate and predict floods in several small river basins, such as the Dinh River in Binh Thuan (with a basin area of 474km²), the Nam La River in Son La (92.4km²), and the Ve River in Quang Ngai (1,260km²)... The outcome resulted in the development of a set of forecasting and warning tools for several small river basins. However, utilizing these tools necessitated sufficient input data such as rainfall, discharge, or water level, and often required significant expertise and time to run due to the complexity involved. In 1998, a study was conducted by Đỗ Xuân Sâm and colleagues to construct a series of maps (at a scale of 1:250,000) for flood warning in the Central South region of Vietnam. The research employed methods of map analysis, database establishment, Geographic Information System (GIS) tools, and factor overlay analysis. The primary factors contributing to flood occurrence were identified as: rainfall, surface conditions (including topography, hydrogeology, and land cover), and human activities within the watershed. It was emphasized that each factor typically possesses a flood-inducing threshold, albeit with considerable variability, owing to the multifaceted nature of flood generation. Flood events often manifest in regions characterized by pronounced fragmentation, steep terrain, limited obstruction, and vulnerable soil substrates prone to erosion and instability-induced impediments to flow, resulting in dam formation and subsequent breaches, particularly prevalent in areas with high river and stream density, short flow paths, and rapid water concentration. Despite not being practically implemented, the study laid the groundwork for future endeavors aimed at developing flood warning maps.

A WebGIS-based integrated flood and drought monitoring and early warning system in real-time was developed by Dr. Tran Anh Phuong and colleagues

In terms of real-time monitoring and alerting, a WebGIS-based integrated flood and drought monitoring and early warning system in real-time was conducted by Dr. Tran Anh Phuong and colleagues. This system is the culmination of a collaborative effort between traditional hydrological and hydraulic models, synergistically combined with machine learning technology, developed by the research team to provide forecasts and deliver emergency response support. The system is characterized by high automation and requires minimal human intervention. Real-time updates on rainfall, water levels, and flood warning maps are provided, alongside the visualization of flood shelter locations and evacuation routes to facilitate safe navigation. Moreover, the system incorporates functionalities for issuing SMS and email flood alerts when water levels surpass predefined alarm thresholds, starting from level 2 onwards. The study also underscores the pivotal role of accurate quantitative rainfall forecasting in anticipating flood events, a challenging task not confined to Vietnam but resonating globally. Consequently, the forecasting system is deemed most reliable within a forecast horizon of 12 to 24 hours.

The Real-time Monitoring and Alert System currently undergoing testing by the Water Resources Institute (WRI) is a result of the research conducted by the team of Nguyen Anh Duc, Duong Hong Son, and their colleagues. This system, designed for flood monitoring and warning, consists of three main components:

(1) The observation and monitoring system: This component encompasses real-time measurement instruments for precipitation, water levels, and live observation cameras. Observation data is transmitted to the server in real-time via internet, 3G/4G networks. Additionally, the system is capable of receiving community-provided information, such as field photos, to augment rescue and relief efforts.

(2) The information storage, orocessing, and forecasting system: This system is tasked with gathering data from the monitoring system and feedback from the public. It leverages forecasted rainfall data from the Global Forecast System (GFS), amalgamated with measured data and various hydrological and hydraulic models, as well as artificial intelligence algorithms, to generate forecasting scenarios.

(3) The information dissemination and emergency response support system: It serves to disseminate warning information via the WebGIS platform. Upon flood levels exceeding alarm thresholds 2 and 3, the system autonomously dispatches warning messages to designated phone numbers responsible for flood and storm prevention. Additionally, the system facilitates emergency response by showcasing rescue and relief support information, such as safe refuge locations and routes to safety.

From 2008 to 2010, a state-funded project titled "Research and development of automatic hydro-meteorological measurement devices," with project code KC.03.08/06-10, was undertaken by Le Duc Hong and colleagues at the Institute of Automation and Military Engineering. During this period, successful development of various automatic measurement devices for hydro-meteorological parameters was achieved. These devices included temperature/salinity sensors, wave gauges, sea level gauges, and an automated center for data measurement, collection, management, and storage. Additionally, wireless data transmission equipment utilizing UHF/VHF/HF frequencies, as well as devices for connecting measurement stations with communication vehicles and other media outlets, were also developed. 

However, due to technological constraints a decade ago, the products still exhibit several limitations. The primary components of the products, designed for measuring salinity, temperature, and water level, rely on data logging and memory partitioning methods employed by Programmable Logic Controllers (PLCs). Consequently, they overlook the measurement of the most critical parameter in hydrology: the flow rate of the current. Furthermore, the conductivity and temperature measurement devices, crucial for inferring salinity, are exclusively sourced from HACH (inclusive of both the sensor and SC100 transmitter). While employing industrial computers for data storage and communication may seem viable, this approach is only feasible in indoor settings with adequate power supply. Transitioning to microprocessor-based devices for outdoor applications poses myriad technical challenges, including power supply management, lightning protection, environmental resilience, energy efficiency optimization, data compression for transmission, on-site storage capacity, and system-wide reliability enhancement protocols. Although domestically-manufactured UHF and VHF communication modules are utilized, the software infrastructure still relies on PLCs.

In Vietnam, early warning systems for large flood events on small rivers and streams have not fully met the increasingly high demands of society. The lead time for flood warnings can only be implemented 3-6 hours in advance for small river basins, and detailed forecasts regarding the timing, location, intensity, and extent of new occurrences are limited to qualitative rather than quantitative risk alerts. These challenges and limitations primarily stem from several factors such as complex terrain. The mountainous regions of Vietnam feature rapidly fluctuating water levels in small and medium-sized rivers and streams. The flow rates are high, and flash floods occur within a short period, often lasting only 1-3 days. Floodwaters rise within a few hours to 10-15 meters, with a descent period ranging from one to several days. The amplitude of floods typically ranges from 5-15 meters, with an increase of 1-3 meters per hour. Extraordinarily large flood events may witness amplitudes of 15-20 meters.  With its sudden and rapid occurrences, the demand for an advanced, modern, and robust monitoring, observation, and warning system is imperative. However, the monitoring network supporting forecasting remains insufficient, lacking radar rainfall data and high-resolution radar systems for mountainous areas and small watersheds where the risk of flash floods is high. Nationwide, there is still no comprehensive network of rainfall monitoring stations in mountainous regions and headwater areas, where floods on small rivers are forming, and where automated data collection and transmission systems are needed to accurately reflect spatial and temporal rainfall patterns. Limitations in scientific and technical capabilities mean that current forecasting models do not allow for detailed, specific, and highly accurate predictions of quantitative rainfall, particularly in small areas, especially in mountainous regions affected by terrain. New flood zoning maps only delineate areas at risk, unable to forecast or warn when floods will occur.

In recent years, there have been significant advancements in the capabilities and technologies for flood forecasting in Vietnam. However, the warning level still remains at the stage of predicting the risk of occurrence within 3-6 hours beforehand, with qualitative warnings outweighing quantitative ones. Detailed predictions regarding the timing, location, intensity, and extent of flood events are still elusive. The national hydro-meteorological monitoring network is not yet sufficiently dense, failing to cover all drainage basins, which hampers the provision of information and data for flood forecasting and warning on small rivers and streams. The complexity of the terrain, unpredictable changes in rainfall patterns due to climate change, and alterations in vegetation cover resulting from deforestation and the release of water from hydroelectric reservoirs further compound the challenges. As a result, the dynamics of flooding are no longer governed by predictable seasonal patterns.

The limitations highlighted above in flood forecasting activities have implications for the management and support of flood prevention efforts in mountainous regions, posing numerous difficulties and challenges. To proactively address the sudden, rapid, unpredictable, and difficult-to-forecast nature of flooding events, it is essential to explore and establish a system for directly disseminating flood warning information from real-time monitoring and observation systems to communities residing in small river basins and streams./.

Tran Quang Ngoc
Center for Hydro–Meteorological Observation
(Published in the Special Issue of Science & Technology, issue number 2, 2023, of The Natural Resources and Environment Magazine)