Rain Check: Using AI to Fill Gaps in South Korea’s Precipitation Data

Precipitation is crucial for understanding the water cycle, including runoff, soil moisture, and evaporation. However, missing data at weather stations can hinder the accuracy of hydrological analysis. To tackle this, researchers developed a tool using machine learning to detect and predict missing precipitation data. This study focused on 30 weather stations in South Korea, exploring how well artificial neural networks (ANN) and random forest (RF) algorithms could fill these data gaps using environmental factors like air pressure, temperature, humidity, and wind speed.

The ANN model proved to be more accurate, with an 80% success rate in detecting missing data and a correlation coefficient ranging from 0. 5 to 0. 7. While both models performed well, the ANN showed slightly better results in estimating daily precipitation. This research suggests that machine learning can significantly enhance the performance of hydrological models by recovering vital precipitation data.

Actions