Satellites and Data
Free Satellite Imagery of Asia
Landsat and Sentinel-2 cover all of Asia every few days at no cost. What each archive offers, how to search by place and date, and what the scenes cannot do.
Two satellite programmes photograph every square kilometre of Asia, several times a month, and give the images away. The American Landsat series and the European Sentinel-2 fleet together form the largest free record of the continent's surface, and they are the first stop of almost every mapping project in the region. This guide explains what each archive offers, how to search them by place and date, and where their limits sit so that a reader can judge what a free image can and cannot support.
What Landsat has watched since 1972
The Landsat programme, run jointly by NASA and the United States Geological Survey, launched its first satellite in 1972 and has kept an unbroken watch ever since. Its modern satellites, Landsat 8 from 2013 and Landsat 9 from 2021, carry the Operational Land Imager, which records visible and infrared light in tiles of 185 kilometres wide at a ground resolution of thirty metres. Each satellite revisits the same point every sixteen days, and the pair flies offset so the effective revisit is closer to eight. Since 2008 the entire archive, including scenes gathered over Asia in the 1970s, can be downloaded at no cost, which turned a generation of historical comparison from a luxury into a routine. A district office tracing how a delta has moved since the Vietnam War era works from the same shelf as a university laboratory.
What Sentinel-2 adds from 2015
The European Union's Copernicus programme launched Sentinel-2A in 2015 and its twins after it, each carrying a multispectral imager that resolves ten metres in the visible bands and twenty in the red edge and shortwave infrared. The swath is wide, 290 kilometres, and the revisit for the full constellation is five days at the equator, tighter at Asian latitudes. The sensor was designed with vegetation in mind, and its thirteen bands make crop and forest work markedly easier than with earlier free data. Where Landsat earns its place through depth of history, Sentinel-2 earns it through detail and frequency, and the two archives overlap enough that a team can cross-check one against the other. Scenes arrive as ready-corrected products through the Copernicus Data Space Ecosystem, with a map browser that previews any Asian province before login.
How a first search actually goes
A practical search starts from a place, not from a satellite. An analyst draws a rectangle around the study area, sets a date window that brackets the season of interest, and asks the catalogue for every scene that touches the box. The result is a grid of thumbnails, each stamped with its capture date and its cloud score. Over most of tropical Asia the cloud score does the sorting: scenes under ten percent cloud are few, scenes above sixty are common, and the working set assembles itself from the clear minority. Downloading a full scene, which covers a region rather than a village, usually serves several projects at once, and teams share the download folders the way offices share reference books.
What can be read at ten and thirty metres?
At ten metres, a Sentinel-2 image separates paddy fields, fish ponds, orchards, roads and built-up blocks with confidence, and a trained reader follows a monsoon flood across them. At thirty metres, Landsat still separates broad land cover classes, rice from forest from urban, and its long record makes it the tool for change: mangrove retreat, lake shrinkage, city growth over decades. Neither sensor counts individual houses reliably, reads vehicle traffic, or resolves the narrow bunds that divide smallholdings. When that level of detail matters, projects either move to commercial tasking or combine free scenes with field photographs and administrative records. The guide on choosing satellite data for work in Asia develops this matching of pixel to task.
How the monsoon gap is handled
Free optical archives have one structural weakness across South and Southeast Asia: the seasons when the land changes fastest, the wet months, are exactly the seasons when cloud closes the view. Teams answer with composites, which merge every acceptable scene of a month or a season into one synthetic clear image, and with radar, which clouds do not block. The flood mapping practice described in the guide to mapping Asian floods leans on radar for exactly this reason, while crop calendars in the guide to rice paddies seen from orbit are built from gap-filled optical time series. Neither trick is exotic anymore; both are buttons in the free platforms.
What the licences permit
Landsat scenes are public domain under United States policy, with no restriction on reuse or commercial products beyond correct attribution of the source. Copernicus Sentinel data carries a free and open licence that likewise allows published maps, reports and derived products, commercial ones included. Both licences expect the user to keep the attribution line, and neither allows claiming the raw imagery as one's own product. National Asian missions, surveyed in Asia's own Earth observation satellites, often sit on more restrictive terms, which is one reason international projects default to the two open fleets.
A reasonable first hour
A reader new to the subject can do worse than spend one hour in the USGS search portal: draw a box around a home city, request Landsat scenes from the last two dry seasons, and open the oldest scene the archive returns for the same box. The comparison between that scene and a current one explains, faster than any definition, what forty years of thirty-metre pixels can show. The portal, EarthExplorer, needs only a free registration, and its download buttons deliver the same files that professional flood and crop services build on.