Water Through the Decades: Surface Water Atlas
Where water was, stayed, appeared and disappeared since 1984, from every Landsat image: lakes, rivers, marshes and reservoirs of any place on Earth.
The global surface-water atlas classifies every Landsat image from March 1984 to 2020, more than three million scenes, into water and land. For any area up to 150 km across this tool shows how often each pixel was water, what happened to it between the first and the last years (permanent, new permanent, lost permanent, seasonal, new and lost seasonal, seasonal to permanent and the reverse), how water occurrence changed between 1984–1999 and 2000–2020, and how many months it was wet in 2020, with every area in km². It is also the reference against which the Earth Through Time tool can be checked.
Overview
Iraq is a country of rivers, marshes and lakes that are changing fast: Lake Sawa dried out, Razzaza and Tharthar shrank, the marshes were drained and partly restored, and new reservoirs filled. The atlas puts all of this on one map and in one table.
Unlike a single image, the atlas is built from every clear observation, so it is not misled by one hazy day or one flood.
What you enter · What you get
What you enter
- A water body from the list, a place found by name or an area drawn (up to 150 × 150 km)
What you get
- Map of water transitions 1984–2020
- Water occurrence map
- Change of occurrence between two periods
- Months of water in 2020
- Areas of every class in km², chart and CSV
How it works in the field
Method and equations
The atlas
JRC Global Surface Water (Pekel et al. 2016, extended to 2020): every Landsat 5, 7 and 8 scene classified into water, land or no observation by an expert system using multispectral and multitemporal rules and the visible-infrared transforms. Layers: occurrence, change, seasonality, recurrence and transitions at 30 m.
Areas
The atlas tiles that cover the area are joined, read without resampling, and every pixel is given its true ground area from its latitude. The net change weights each pixel by the change of its occurrence.
Accuracy and error limits
- In the published validation the atlas has less than 1 % false water and fewer than 5 % missed water pixels. Areas of large water bodies are therefore within a few per cent; narrow rivers and canals under two or three pixels wide are often missed.
- What is not covered: years after 2020, water under dense reeds or trees (counted as land), the depth and volume of water, and periods with few Landsat observations (some years of the 1980s and 1990s over Iraq), which make occurrence less certain.
- To reduce the error: read the transitions together with the occurrence map, check recent years with the Earth Through Time tool, and for marshes remember that water hidden under reeds is not counted.
Who uses it
Standards and references
- 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.
- European Commission Joint Research Centre (2021). Global Surface Water Explorer, data users guide (v4).
- 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.
At a glance
- Category
- Water & rivers
- Runs on
- Phone, tablet and desktop browser
- Export
- Four water maps as PNG, JPG and PDF; transition chart as PNG, JPG and PDF; areas as CSV; full PDF report
- Access
- Free
Run this survey on your next site
All 64 tools, report export and the AI assistant come with one subscription.