Earth Through Time: Every Year Since 1972
Every year of Landsat since 1972 over any place on Earth — the marshes of Iraq, the Aral Sea, Lake Urmia, Dubai or your own area: measured, compared and played as a film.
Pick a place in Iraq (the marshes, Lake Sawa, Razzaza, the Mosul, Haditha or Dukan lakes, Baghdad, Erbil, Basra) or in the world (the Aral Sea, Lake Urmia, the Dead Sea, the Atatürk Dam lake, Dubai, Cairo, Lake Mead, the Columbia Glacier, the Amazon in Rondônia, Shanghai, Lake Poopó), search any place by name, or draw any area on Earth. For every year from 1972 to today the tool takes the clearest Landsat scene of the same season, crops it to the area, measures open water and green vegetation on every pixel with its true ground area, and builds framed map frames with coordinates, north arrow, scale and legend. It draws the curves of water and vegetation through the years, maps what was lost and gained between any two years, and records the whole story as a video in landscape, square or vertical reel format.
Overview
Iraq has changed more in forty years than in centuries: the marshes were drained in the 1990s and partly restored after 2003, Lake Sawa dried out in 2022, Razzaza shrank, new reservoirs filled behind the Mosul, Haditha and Dukan dams, and Baghdad, Erbil and Basra spread over their farmland. The Landsat archive recorded all of it every sixteen days; this tool turns it into numbers and pictures that anyone can read. The same is true across the world: the Aral Sea that mostly vanished, Lake Urmia that shrank, the artificial islands of Dubai, the cleared Amazon forest and the retreating glaciers, all since the first Landsat image in 1972.
Every frame is a map, not a screenshot: true ground area, coordinates, scale and legend make it usable in a thesis, a report, a news story or a reel, and every number can be checked in the table and the CSV file.
What you enter · What you get
What you enter
- A place from the lists, a place found by name, or an area drawn on the map anywhere on Earth (up to about 150 × 150 km)
- First and last year (from 1972), step and season, cloud limit
- Thresholds of the water and vegetation indices, for modern and for MSS years
What you get
- Framed map frames for every year, played as a time-lapse
- Water and green vegetation area through the years
- Change map between any two years with areas lost and gained
- Video (16:9, 1:1, 9:16), frames as PNG, JPG and PDF, table as CSV
How it works in the field
Method and equations
Scenes
From 1982: Landsat Collection 2 Level-2 surface reflectance from Landsat 4–5 TM, 7 ETM+ and 8–9 OLI (30 m). From 1972 to 1981, and in later years up to 2012 where no such scene exists: Landsat 1–5 MSS Level-1 (60 m, green, red and two near-infrared bands). From 2017 Sentinel-2 Level-2A surface reflectance at 10 m is used by default, three times sharper than Landsat (it can be switched off). For small areas the clearest of the candidate scenes is chosen by its local contrast, so hazy or dusty days are avoided, and a joint contrast stretch of the three colour channels removes the remaining grey haze. Clouds and cloud shadows are masked with the quality band of each scene (Landsat QA_PIXEL, Sentinel-2 scene classification) widened by two pixels, so they are never counted as water; years with 5 % or more local cloud are kept out of the curves. Years before 1982 are left out for areas narrower than 15 km, which their 60 m pixels cannot show. When one scene does not cover the whole area, up to four scenes of the same season from neighbouring paths and rows are joined into a mosaic, the closest in date first; years that still do not cover the area are left out of the film unless every year is asked for. For each year the scenes of the chosen season below the cloud limit are ranked by cloud cover, coverage of the area and sensor condition (Landsat 7 after the scan-line-corrector failure of May 2003 is used only when nothing else exists).
Water and vegetation
Open water by the modified normalised difference water index of Xu (2006) from the green and short-wave infrared bands; green vegetation by NDVI among the remaining pixels. Thresholds are 0 and 0.3 by default and can be changed. MSS has no short-wave infrared band, so for its years water is found with the NDWI of McFeeters (1996) from the green and near-infrared bands, with its own thresholds (0.1 and 0.2 by default).
Area
Each pixel of the geographic crop is given its true ground area from the cosine of its latitude; years whose scene covers less than 90 % of the area are listed but left out of the curves.
Accuracy and error limits
- MSS years (before 1982) are coarser and less certain: 60 m pixels, no blue band so the colour is approximated, and uncalibrated digital numbers, so their areas can differ from the later years by 5–15 % for large lakes and more for small ones; read them as the starting point of the trend and check each early frame by eye.
- Landsat pixels are 30 m: water bodies are measured to about one pixel along their shores, so areas of large lakes are within 1–3 % and those of small ponds and canals within 10–30 %. MNDWI water maps reach 90–95 % overall accuracy in published tests.
- One scene per year is a snapshot: a flood or a dry month moves it by much more than the measurement error. Water under dense reeds counts as vegetation; cloud, cloud shadow, dark lava or wet soil can be taken as water; Landsat 7 scenes after 2003 have striped gaps; the archive has missing years in many regions, especially in the 1970s and 1990s.
- To reduce the error: keep the same season every year, look at each frame before trusting its number, raise the water threshold to 0.1–0.2 where shadows or wet soil appear, and read the trend over several years rather than a single year.
Who uses it
Standards and references
- Drusch, M. et al. (2012). Sentinel-2: ESA’s optical high-resolution mission for GMES operational services. Remote Sensing of Environment, 120, 25–36.
- McFeeters, S.K. (1996). The use of the Normalized Difference Water Index (NDWI) in the delineation of open water features. International Journal of Remote Sensing, 17(7), 1425–1432.
- U.S. Geological Survey (2021). Landsat Collection 2 Level-1 MSS data products. USGS.
- U.S. Geological Survey (2020). Landsat Collection 2 Level-2 Science Products. USGS Fact Sheet 2021-3055.
- Xu, H. (2006). Modification of normalised difference water index (NDWI) to enhance open water features in remotely sensed imagery. International Journal of Remote Sensing, 27, 3025–3033.
- Tucker, C.J. (1979). Red and photographic infrared linear combinations for monitoring vegetation. Remote Sensing of Environment, 8, 127–150.
- Partow, H. (2001). The Mesopotamian Marshlands: Demise of an Ecosystem. UNEP Early Warning and Assessment Technical Report.
- Pekel, J.-F., Cottam, A., Gorelick, N. & Belward, A.S. (2016). High-resolution mapping of global surface water and its long-term changes. Nature, 540, 418–422.
- Microsoft Planetary Computer, Landsat Collection 2 Level-2.
At a glance
- Category
- Remote sensing
- Runs on
- Phone, tablet and desktop browser
- Export
- Time-lapse video (MP4 or WebM, 16:9, 1:1 or 9:16), frames and change map as PNG, JPG and PDF, chart as PNG, JPG and PDF, table as CSV
- Access
- Free
Run this survey on your next site
All 64 tools, report export and the AI assistant come with one subscription.