A new study Global prediction of extreme floods in ungauged watersheds published in a recent issue of scientific journal Nature says AI and machine learning technologies have significantly improved global flood forecasting, particularly in regions where flood-related data is scare.
According to the study the use of AI-based technologies in riverine flood forecasting can be a significant change, extending the reliability of global warnings from zero to five days on average. This advancement can help save lives by providing real-time river forecasts up to seven days in advance, especially in regions of Africa and Asia.
The study led by University of California assistant professor (land, air, and water resources) Grey Nearing, however, said, “Further work is needed to expand coverage to more locations and for other types of flooding, underscoring the importance of continued collaboration between tech companies, academic institutions, and governments.”
The study revealed that:
$50bn are the annual economic damages worldwide caused by floods, the most common natural disaster
Over 1.5bn people or 19% of the world’s population, are exposed to substantial risks from severe flood events
The rate of flood related disasters has more than doubled since the year 2000, partly due to climate change
From zero to five days is how much reliable river flood forecasts have been extended using AI-based technologies\
While the machine learning models have shown significant improvement in forecasting, the system still relies heavily on publicly available weather data and physical watershed information. This could pose a challenge in areas that lack necessary infrastructure for data collection, which often correlates with lower GDP and increased vulnerability to flood risks.
The models also need further development to cover other types of flood-related events, such as flash floods and urban floods. Accurate and timely warnings are critical for mitigating flood risks, but hydrological simulation models typically must be calibrated to long data records in each watershed.
Source: asiainsurancereview.com