Data Mining and Remote Sensing Techniques in Hydrological Modeling

The project was initiated to address the critical challenges in water resource management, underscored by the urgency to tackle water shortages, data inaccessibility, and the impacts of climate change on hydrological processes. It aimed to exploit novel data mining and remote sensing techniques to provide comprehensive solutions, rooted in collaborative efforts among hydrology, remote sensing, and data science experts. These experts identified an essential need for advanced technological approaches to understand and mitigate global-scale challenges.

Regarding the achievements of the project's original goals, we have made notable progress. The substantial funding received enhanced our team's expertise in machine learning, data mining, and remote sensing, allowing us to apply our methodologies worldwide, including in Finland, Norway, Turkey, and Iran. The project also supported the salaries of two doctoral researchers at the University of Oulu, further contributing to our team's capacity.

Innovative data mining techniques led to new insights into water shortages, improving effective water resource management strategies. Key accomplishments include:

- Developing an approach for mapping suitable sites for Rainwater Harvesting (RWH) structures using socio-environmental variables and artificial intelligence algorithms, addressing water scarcity.

- Utilizing hybrid evolutionary algorithms to estimate Groundwater Level (GWL) fluctuations more accurately, aiding in integrated water resource management.

- Introducing a methodology, in collaboration with Turkish scholars, for predicting sea level changes, crucial for coastal structure design and harbor operations.

- Proposing a new hybrid Random Forest model for drought forecasting, enhancing the accuracy of drought index predictions.

- Advancing flood and flow prediction with a new evolutionary time series model and a hybridized model based on neural networks and swarm intelligence for flood-inundation mapping.

- Addressing ecohydrological modeling challenges by proposing jittered binary genetic programming to identify dominant parameters affecting Arctic charr occurrence.

The application of remote sensing technology and satellite imagery has been pivotal in overcoming data limitations, enhancing our understanding of hydrological processes worldwide. Notably, we utilized the Google Earth Engine extensively for water resource engineering applications, producing detailed analyses of surface water distribution, evapotranspiration, and land cover, among others. This approach enabled us to assess vegetation health, sea surface salinity, and river ice cover with high spatial resolution, offering new perspectives on river hydrology and the effects of extreme flow events.

In summary, the project has not only met but exceeded many of its original objectives, providing innovative solutions to pressing water management challenges and fostering significant advancements in hydrological modeling and environmental monitoring.