The problem with predicting the impact of floods using the older stochastic tools simply doesn't work accurately. It assumes that what has happened in the past can be translated into the future with extrapolation for probabilistic flood events such as 100-year, 500-year, 1000-year floods. And increasingly, we see that these methods fail to accurately predict impacts. The problem is with the underlying assumptions of stationarity. But land use changes render previous rainfall/run-off relationships to be less than accurate. They are useful, but not accurate enough to use with impunity. This brings us to the question as to why more sophisticated modeling using AI is not used. Decision support systems can be developed which use real-time, online data from monitoring and remote sensing. That is my challenge to the submerged community.
Taxonomy
- Remote Sensing & Data Analysis
- Flood
- Monitoring