Visualizing "Invisible Surface Water" in Rice Paddies: Creatura Doubles Surface Water Detection Accuracy with Satellite Data x AI

Creatura Inc. (Headquarters: Chuo-ku, Tokyo; CEO: Michiyasu Hattori; hereinafter "Creatura"), a climate tech company, has announced new research findings in the peer-reviewed academic journal "IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (JSTARS)," which has a high citation record in the field of satellite Earth observation. This research demonstrates that an AI framework integrating meteorological and topographical data with satellite imagery can significantly improve the accuracy of surface water detection in rice paddies.

According to the research, it was confirmed that surface water detection performance could be improved more than twofold compared to conventional methods, even under conditions with thick and dense crop cover.

Alexis Declaro, Head of Data Science & GIS at the company, stated:

"The key is not which satellite data is best on its own, but to reproduce what is happening on the ground by combining multiple elements. We hope this research will serve as an opportunity to reconsider the premise of SAR*1 dependency in rice paddy water management monitoring and to re-evaluate the optimal use of optical data according to its application and context."

Features of this Framework

(Figure 1) Overview of the framework proposed in this research

In conditions where rice plants have fully grown, it has been difficult to determine the presence or absence of water using conventional satellite analysis, forcing reliance on estimation. This research has shown that by integrating meteorological and topographical data in addition to satellite imagery, it is possible to detect inundation events that were previously difficult to detect with high accuracy.

The feature of this framework is not merely data integration, but rather the explicit handling of how each input data reflects the physical conditions on the ground. It is based on observable elements such as the interaction between vegetation and surface water, atmospheric and soil moisture conditions, and water flow due to topography, yielding predictions with high logical transparency and explainability.

This research forms the scientific basis for the company's operational system, the "SWAP (Surface Water Absence and Presence)" model. SWAP is a platform that integrates multiple satellite data sources to enable near-daily water management monitoring at the rice paddy level, supporting the large-scale adoption of water-saving and emission-reduction techniques such as Alternate Wetting and Drying (AWD).

The full paper can be viewed here.

*1 SAR (Synthetic Aperture Radar) is a satellite technology that uses radio waves, not light, for observation, enabling the understanding of ground surface conditions by penetrating clouds and vegetation.

Paper Information

Title: Data-Driven Assessment of Climate and Topographic Integration With Satellite Imagery for Improved Surface Water Detection Under Agricultural Vegetation

Authors: Alexis Declaro, Jin Xiao, Robert Galla, Shinjiro Kanae (Tokyo University of Science/University of Yamanashi)

Journal: IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (JSTARS)

DOI: https://doi.org/10.1109/JSTARS.2026.3699594