DATOS: making satellite information useful and checkable

The Remote Sensing and Data Science Help Desk, known as DATOS, applies geospatial analysis and data-science methods to practical questions such as flooding and agriculture. DOST-ASTI describes satellite observations supplied through PEDRO and computing and archiving support from COARE. These are complementary roles: receiving data, preserving it and interpreting it are connected activities, but they are not the same task.

A 2022 transition account states that the original DATOS project period had ended while its activities continued as services. This matters when reading an older project page. A project’s funding dates do not automatically tell you whether every associated capability disappeared. Equally, a continuing service name does not prove that its methods or institutional responsibilities remain unchanged forever.

Consider a fictional flood classifier evaluated against one hundred independently checked locations. It correctly labels ninety overall, but misses nine of the ten truly flooded locations. Its overall accuracy is ninety percent, yet it detects only one tenth of the floods. A single summary percentage can therefore hide exactly the error that matters most for the intended use.

The example is not a reported DATOS result. It demonstrates why validation should distinguish missed events from false alarms and use reference information appropriate to the question. DOST-ASTI’s published example flood products explicitly require further validation and interpretation. A colored map is an analysis output with assumptions and uncertainty, rather than a direct photograph proving that every marked location has a particular condition.

One percentage can hide a problem

A useful product should retain acquisition time, source sensor, geographic reference, processing method and relevant limitations. If an algorithm changes, version information helps explain why a new map differs from an earlier one. Preserving the underlying observations allows analysts to revisit the question instead of relying only on a final image whose origin is unclear.

DATOS also described crop analysis using changes through time. A time series can provide information that a single scene misses, but dates, missing observations and seasonal differences must be considered. For both crops and floods, the final product should answer a defined user question and communicate what it cannot establish. Such information supports the work of responsible agencies; it does not by itself replace their official alerts or decisions. The value of a space-data program lies in making its results useful, traceable and open to checking.