Satellite Data for Asia: A Field Picker's Guide

How do I choose satellite imagery for a project in Asia?

Start with the question your map has to answer, not with the sensor. If you need to see a monsoon flood moving this week, you want a high revisit mission with a coarse pixel; if you need to count rice paddies, you want a finer pixel and you can wait. Match resolution, revisit and licence to the decision the map supports, then pick the archive that holds the scene. The rest of this piece walks through that choice for Asian projects, the free places to download from, and the national programmes that put hardware over the continent. Write down three numbers before you open any browser. The first is the smallest object you must resolve: a city block, a single field, a flooded road. The second is how often you must see it: once, monthly, or every few days during the monsoon. The third is the licence you can live with, because a scene you cannot legally publish is not data, it is a screenshot. Those three numbers eliminate most of the catalogue. A 10 m pixel suits regional flood mapping and broad land cover; a 30 m pixel suits multi-decade change; a 1 m or finer pixel suits building footprints and damage assessment but arrives with a commercial licence and a smaller footprint per scene. Revisit is the harder constraint in Asia, because cloud is the real enemy. A sensor that passes every 16 days is useless over the Mekong in July if the pass lands under cumulus; a sensor that passes daily at 250 m or 500 m will give you a usable time series even when individual frames are partly obscured. One more filter that saves weeks: check whether someone has already processed the scene. Surface reflectance products, cloud masks and mosaics from the same provider will beat your own atmospheric correction on a deadline.

Where can I download free satellite images of Asia?

Free does not mean one archive. It means four or five archives with different rules, and you should know which one you are standing in. Landsat is the long memory. The USGS EarthExplorer archive carries the full Asian record back to the 1970s at 30 m, and it is the reference dataset for decadal change: delta retreat, urban growth, glacier margins. Sentinel-2, from the European Copernicus programme, adds 10 m optical imagery with a five-day revisit at the equator, which in practice means several usable passes a month over most of South and Southeast Asia. Sentinel-1 radar is the one to remember when the sky will not cooperate: it sees through cloud and works at night, which is why flood responders in Bangladesh and the Philippines lean on it during the monsoon. For weather and fire at continental scale, the Japanese Himawari geostationary series delivers full-disk imagery every ten minutes. It is not a mapping sensor, but for tracking a typhoon track or a smoke plume across the South China Sea it is the fastest free view available. NASA's FIRMS and MODIS products cover active fire and thermal anomalies at 250 m to 1 km, useful for haze season in Indonesia and for crop residue burning. Three practical habits make these archives work. Download the metadata with the scene, because you will need the acquisition time and the sun angle later. Keep a local copy of anything you might publish, since archive interfaces change. And read the licence page once per project, not once per download.

Which Asian countries fly their own Earth observation satellites?

Several, and the list has grown steadily since the 1990s. Japan operates the ALOS series for radar and optical land observation, alongside the Himawari weather satellites and the GOSAT greenhouse gas missions. India runs the IRS programme, one of the oldest civilian Earth observation lines in the world, with Cartosat for high-resolution mapping and Resourcesat for wider multispectral coverage; the data are distributed through the national remote sensing centre and are widely used across South Asia. China flies the Gaofen series and the Fengyun weather satellites, with Gaofen providing sub-metre to metre-class optical data and radar. South Korea operates KOMPSAT, including high-resolution optical and SAR payloads. Thailand has THEOS, an optical satellite built with international partners and used for agriculture and disaster work. Indonesia, Vietnam and the Philippines have all launched smaller national or university-built satellites, often as technology demonstrators that still return usable imagery. The practical consequence for a project is that national missions often have better coverage of their own territory than the global free archives, and their data policies vary widely. Some distribute freely to domestic researchers, some sell commercially, some require a request. If your study area is inside one of these countries, it is worth a direct enquiry before you assume the global archive is your only option.

What does ground truth actually change in a map?

Everything that involves a number. A classification can look convincing and still be wrong by twenty percentage points in the class you care about, and the only way to know is to visit points on the ground or to compare against independently collected reference data. In rice mapping, that means field boundaries and crop calendars from extension officers. In flood mapping, it means high-water marks and local reports. In urban heat work, it means station temperatures, not just surface brightness. Plan the validation before you classify. Decide how many points you can realistically check, stratify them by class rather than scattering them randomly, and record the date of each observation, because a field visited in February does not describe August. The published accuracy figure is only as good as that log.

Where does the regional community publish its work?

The Asian Conference on Remote Sensing has run annually for more than four decades, organised by the Asian Association on Remote Sensing, an NGO founded in 1981 with members across more than 29 countries. Recent editions give a sense of scale: the 45th conference was held in Colombo in November 2024 with over 500 participants, and the 46th in Makassar in October 2025 drew 450 participants from 25 countries and 244 papers, split between oral and poster sessions. The proceedings are archived, which makes them a useful place to look for applied work on exactly the landscapes you are mapping, often with local validation you would otherwise have to repeat.

A short checklist before you commit

Name the decision the map supports. Fix the minimum object size, the required revisit and the licence. Pick the archive that satisfies all three, and prefer radar when cloud is the limiting factor. Check whether a national mission covers your area better than the global archive. Budget time for validation, and write down the accuracy you can defend. Then start processing, because the scene you downloaded today will be superseded by a better one next month, and the project still has to ship.

Then check the geometry of your area. Archipelagos and coastlines reward wide swaths and off-nadir agility; mountain terrain in Nepal or the Tibetan Plateau rewards a mission with good orthorectification and a DEM you trust. If your project crosses borders, confirm the licence covers every country in your study area, not just the one where your institution sits. A practical shortcut is to read how other practitioners frame the same trade-offs; the editorial team at Ground Truth Asia has a running guide to choosing satellite data asia projects actually use, organised by sensor family and by country.