Automatic flood detection using sentinel-1 images on the google earth engine.

Flood events pose significant risks to human populations and infrastructure. Timely detection of flood-affected regions is essential for emergency response and damage mitigation. While traditional methods rely on field surveys and manual image interpretation, these are often time-consuming and limited in spatial coverage. Recent advances in satellite remote sensing have enabled more efficient flood monitoring. Prior research has shown that radar imagery, such as from Sentinel-1, can penetrate cloud cover and provide reliable flood extent data. However, the accuracy of automated flood detection methods remains variable due to differences in surface reflectivity and terrain. This gap motivated the need for a robust and scalable flood detection approach. No prior work had resolved the challenge of consistently applying thresholding techniques across diverse landscapes. The need for rapid and accurate flood mapping tools has driven innovation in this field.

The study aimed to evaluate the effectiveness of the Otsu thresholding algorithm for automatic flood detection using Sentinel-1 satellite imagery. The researchers focused on a specific flood event in northern Iran in March 2019, which lasted for two months. They sought to determine if a single threshold value could be applied consistently across multiple satellite images to delineate flood-affected areas. The study also aimed to assess the accuracy of the automated method compared to independent datasets. The motivation was to provide a scalable and efficient solution for flood monitoring in complex environments. The researchers proposed that an optimized threshold could reduce the need for manual interpretation. The study was designed to test the feasibility of using radar imagery and thresholding for real-time flood mapping.

The study used Sentinel-1 satellite images collected during a flood event in northern Iran in March 2019. The Otsu thresholding algorithm was applied to classify flooded areas based on backscatter intensity values. A threshold value of -14.9 dB was calculated and applied to each image scene. The algorithm segmented the images into flooded and non-flooded regions. The resulting flood maps were compared with independent datasets to validate accuracy. The researchers used a time series of images to monitor changes in flood extent over two months. The methodology included statistical analysis to assess the performance of the thresholding approach. The study did not rely on manual digitization or field surveys for validation.

The Otsu thresholding method successfully identified flood-affected areas with an overall accuracy exceeding 90%. The threshold value of -14.9 dB was found to be consistent across multiple image scenes. The method captured the spatial variability of the flood event over time. The flood extent varied significantly between image acquisitions, but the thresholding algorithm maintained high accuracy. The results showed that the automated approach could detect changes in flood extent with minimal user input. The study found that the thresholding method outperformed manual interpretation in terms of speed and coverage. The accuracy was validated using independent datasets, confirming the reliability of the method. The results suggest that the Otsu algorithm is suitable for rapid flood mapping in complex landscapes.

The study demonstrated that the Otsu thresholding algorithm is effective for automatic flood detection using Sentinel-1 imagery. The threshold value of -14.9 dB provided consistent results across multiple image scenes. The method was validated using independent datasets, confirming its accuracy. The researchers concluded that the approach is suitable for rapid flood mapping in complex environments. The study proposed that the thresholding method can be used for real-time flood monitoring. The results suggest that the Otsu algorithm is a reliable alternative to manual interpretation. The study did not claim that the method is universally applicable but highlighted its effectiveness in the tested region. The authors emphasized the need for further testing in different geographic contexts.