Large-scale floods, which result in loss of life and the displacement of thousands of residents, serve as a stark reminder of the unpredictable and destructive power of water. This vulnerability prompted researchers from the Indian Institute of Technology (IIT) Bombay to create a complex dual-operation artificial intelligence (AI) system. This system is designed not only to identify where the next major flood might occur but also to accurately predict the water level rise.
The team combined satellite radar data with advanced machine learning to create a highly detailed map that identifies flood-prone areas with an accuracy exceeding 93%. This system covers an area of 55,000 square kilometers, stretching from Tadri in Uttara Kannada district of Karnataka to Kanyakumari along the Western Ghats coast in South India. The new tool is intended to be a powerful means of protecting millions of people living in India's most vulnerable coastal regions.
Traditionally, flood forecasting relied on extensive historical precipitation data and physical sensors. However, the IIT Bombay team, including researcher Dr. Kashish Sadhwani and Professor T.I. Eldho, focused on pattern recognition by analyzing a number of key factors. Notably, the study showed that surface runoff was a more important predictor than the volume of precipitation itself.
As explained by Dr. Sadhwani, although precipitation is the primary trigger for floods, it does not always directly lead to inundation in a specific location. Surface runoff reflects the complex hydrological response of the landscape, taking into account the combined influence of rainfall intensity, soil moisture, land use, infiltration capacity, and drainage characteristics.
A two-stage procedure was used to process this data. First, a classification model determines whether an area is at risk. Then, a regression model estimates the potential depth of accumulated water, creating a continuous map of potential flooding. Sentinel-1 satellite imagery from the European Space Agency (ESA) in Synthetic Aperture Radar (SAR) mode was used to train this model, as these images can penetrate monsoon clouds and capture observations. The model was trained to recognize dark shades in the images indicating standing water by comparing pictures taken before and during past floods.
Ultimately, the model provides a highly detailed map with a resolution of up to a 30-meter grid, but currently has an error (RMSE) of about 0.99 meters. Although a deviation of approximately one meter may be significant for urban planning, Dr. Sadhwani notes that the current value of the system lies in its scope and speed. Dr. Sadhwani emphasizes that the model is designed for rapid regional flood analysis, ensuring high computational efficiency and the ability to quickly generate information on the area and depth of inundation over large territories. This is particularly useful for early-stage planning, prioritizing vulnerable areas, and supporting emergency response.
The system currently focuses on terrain with a slope of less than 7%. This was a deliberate methodological choice because the study indicated that flood depth calculations were limited to areas with slopes less than 7% to ensure accurate mapping of inundation using SAR images, considering possible water movement during the survey. Furthermore, on steeper sections, radar signals are distorted by geometric deformations such as shadow and layover, and water movement during data collection can lead to inaccuracies in determining the boundaries and depth of inundation. Thus, setting a slope threshold helps guarantee the reliability and physical consistency of the resulting inundation depth data.
For Kerala and Karnataka, this tool could radically change the situation. 30-meter resolution maps in areas where clay soils retain water and low-lying coastal plains prevail will allow local authorities to precisely determine which hospital, school, or road is likely to be flooded. Dr. Sadhwani stresses that this structure is capable of 'identifying flood-prone areas, guiding urban planning and land use management' and plays a 'critical role in disaster preparedness and response, allowing authorities to effectively allocate resources and prioritize vulnerable regions for evacuation and aid.'
Although the current study focused on the southern part of India's West Coast, researchers believe that this foundation is ready to be scaled up to complex urban centers like Mumbai or the East Coast. However, transitioning to these areas will require accounting for new variables and possibly recalibrating and retraining the model. Dr. Sadhwani explains that 'coastal conditions introduce additional complexities, such as tidal fluctuations, storm surges, sea-level changes, and backflow effects. The methodology can be effectively adapted by incorporating these coast-specific parameters into the existing structure.'
As climate change increases the frequency of extreme weather events, this AI system offers more than just a forecast; it helps disaster management groups prepare and plan appropriate measures to reduce flood losses and increase regional resilience.
Modern technologies, while simplifying life, are simultaneously creating a number of problems. Actorra and her family, including Archana Puran Singh, shared an example of growing difficulties related to artificial intelligence (AI).
On social media, Archana Puran Singh, along with her sons Aryaman and Ayushman, stated that the planet is experiencing a shortage of water due to AI. According to their assertion, the more questions asked to AI, the more water is consumed.
This issue was raised against the backdrop of the water problem in Mumbai. When the family was in the Malad area, they encountered a severe water shortage. The water supply problem in various parts of Mumbai has worsened due to the lack of rain.
Younger son Ayushman Sethi noted that even with technologies like AI, it is impossible to satisfy the most basic needs. His brother Aryaman added that AI consumes excessive amounts of water. Archana supported this view, stating that AI tools use a lot of water, which exacerbates the shortage. She emphasized that every time a user asks a question in a chat, AI spends a significant amount of water generating the response.
Ayushman recounted that once, while using his AI avatar to create a video, he was told that it resulted in a huge amount of wasted water. After this, he decided to stop using AI because he felt he could not do such things. Aryaman concluded that AI is mainly used for pointless activities, such as chatting for entertainment.
It is worth noting that Archana and her family are actively active on YouTube, releasing daily family vlogs that are popular with many viewers.
Due to accelerated climate change and unpredictable weather conditions, Bangalore faces serious flash floods caused by extreme rainfall. Researchers from the CSIR Fourth Paradigm Institute and AcSIR created a high-precision hydrological simulation using the United States Environmental Protection Agency's (SWMM) stormwater management model. This model is built on a complex 'digital twin' of the Koramangala-Challaghatta catchment area. The team modeled historical flood events to analyze how stormwater overwhelms the city's concrete infrastructure. Now, the administration can use this model as a decision support system to forecast inundation zones and optimize drainage capacity. It was noted that Bangalore's transformation into a technology hub has led to environmental problems, and the lack of natural landscape contributes to the rapid overflow of old drainage systems.
