Comparing Satellite Imagery: Commercial vs. Public Data for Surface Water Detection

A recent study reveals that commercial satellite imagery often outperforms public data sets in general surface water detection, but public data excels at identifying water hidden by forest cover.

Satellite imagery plays a crucial role in mapping surface water. This is achieved through machine-learning algorithms that analyze color data across various spectral bands. Data for these analyses can be sourced from either commercial purchases or public availability, with commercial options typically offering higher-resolution images.

Study Comparison: PlanetBasemap vs. DSWE

Researchers specifically compared the commercial PlanetBasemap data set with the Dynamic Surface Water Extent (DSWE), a public data set from the United States Geological Survey Landsat program.

Lead author Mollie Gaines emphasized a key difference: "PlanetBasemap's four-meter resolution led to more detailed images compared to DSWE's 30-meter resolution, making it more capable of detecting small water bodies and river extents."

Spectral Range Advantage of Public Data

However, the public DSWE data demonstrated a significant advantage during seasons with high vegetation. This is because DSWE captures a wider portion of the electromagnetic spectrum, including the shortwave infrared band, which is particularly effective for detecting water concealed by vegetation. In contrast, PlanetBasemap, built on Planet Scope data, is limited to visible light and near-infrared.

Furthermore, including all three of DSWE's "confidence classes" substantially enhanced its ability to capture water in complex features such as winding streams and rivers.

Conclusion: Distinct Applications for Each Data Set

The study ultimately suggests that both commercial and public data sets have distinct and valuable applications.

Commercial data is more reliable for identifying very small water bodies, while public data is a suitable option for larger study areas, especially when vegetation obscures the water.

The findings were published in the paper "Impact of spatial scale on optical Earth observation-derived seasonal surface water extents" in Geophysical Research Letters.