There is a particular kind of optimism that surrounds open data: publish the dataset and the problems will follow it into the light. NASA’s Earthdata programme is a good test of that idea. It makes an enormous range of Earth observations and tools publicly available, at no cost. The openness is real. But a satellite image on a server and a decision in a village are separated by a long chain, and almost all the hard work happens along it.
It helps to think of that chain as a pipeline with six stages. Data only becomes useful when a community can travel the whole length of it.
Stage one: availability
Someone has to observe and archive. Earthdata does this at scale, drawing on NASA missions and partner networks to produce imagery and measurements of land, oceans, atmosphere and ice. This stage is the one most often celebrated, and rightly, without it nothing else is possible.
Stage two: access
Availability is not the same as accessibility. Can a district officer with a slow connection download the file? Is it in a format that ordinary software can open? Is there an interface for someone without remote-sensing training? Open licences remove legal barriers; they do not remove bandwidth, format or interface barriers.
Stage three: skills
Even a well-formatted dataset requires someone who can interpret it. Knowing that a band combination reveals flood extent, or that a vegetation index tracks crop stress, is a specialised skill. Where that skill is scarce, the data sits unused, not because it is hidden, but because no one nearby can read it.
Raw observation has to be turned into something a decision-maker can act on. A rainfall anomaly matters to a farmer only once it is expressed as a risk to this season’s harvest. Translation is the stage where science meets local knowledge, and it is usually done by intermediary organisations rather than by the space agency itself.
Stage five: decision
Finally, a choice is made: to release water, to move livestock, to pre-position supplies, to warn a community. The decision depends on institutions and authority as much as on information. Data can inform a decision; it cannot make one.
Stage six: feedback
The pipeline only improves if it loops. When a forecast is wrong, does that reach the people who build the model? Programmes such as famine early-warning systems and agricultural monitoring initiatives work precisely because they close this loop with the users of the information, not just the producers of it.
In practice, the break is rarely at stage one. It is between access and skills, or between translation and decision. A flood map that no one is trained to read, or that no authority is mandated to act on, changes nothing. The lesson is not that open data is overrated, it is that openness is necessary but not sufficient, and that the unglamorous middle of the pipeline deserves as much investment as the satellites.
An example: flood early warning
Flood early warning shows the pipeline in miniature. Satellites and rain gauges provide observations. Hydrological models turn them into forecasts. A national agency decides whether to issue a warning. Local radio, SMS and volunteers carry it to households. And after the event, the forecast is checked against what actually happened. Break any link and the warning fails: a good forecast that never reaches a village saves no one, and a warning no one is authorised to issue arrives too late.
Funding the middle
Space agencies are comparatively well funded; the training, translation and community engagement in the middle of the pipeline are not. This imbalance is the central problem of open data for development. A modest investment in local capacity, a university programme, a regional service, a network of trained intermediaries, often unlocks more value from existing data than a new satellite would. The return is not photogenic, which is exactly why it is under-supplied.
Open data removes one barrier and leaves several standing. That is an argument against treating openness as the finish line. The datasets are free. The work of turning them into decisions is not, and that work is where communities either benefit or do not.
A note on credit and licensing
Open data comes with terms. Most Earth-observation programmes require attribution, and some carry restrictions on redistribution or commercial use. For community organisations, understanding those terms is part of the access stage: a dataset that cannot legally be shared onward may be open to a researcher but closed to the local group that needs it. Checking the licence is unglamorous, but it decides whether the pipeline can legally reach its final stage.
This report draws on NASA‘s Earthdata programme and the Copernicus programme.
Image: NASA · Public domain · via Wikimedia Commons.