
White: waterline of the chosen summer. Gold: summer 1984. Open circle: the fixed measuring point.
Sfântu Gheorghe mouth · Danube Delta
Sacalin Island between river and sea: what 41 summers of satellite images show
In the 2025 image the western end of the island lies 4.4 km further west than in the 1984 image. The distance from a fixed point to the north-east grows less, from 6.99 to 10.74 km, because the point lies off to one side. Comparing the 1984 and 2025 Landsat frames: 4.4 km at threshold zero between water and land, 4.8 km at threshold −0.1.
Measure two yearsor scroll the summers, from 1984 to 20252025
What you can and cannot see
Sacalin is a strip of sand several kilometres long at the mouth of the Sfântu Gheorghe branch of the Danube, between the inner lagoon and the Black Sea, in the Danube Delta in eastern Romania. The picture is summer 2025, seen by the Landsat 9 satellite.
Gold is the outline from summer 1984. The gold line joins the two tips. The open circle is the fixed point the distance is measured from: a marker north-east of the island, in the lagoon area, which the images show as water in some years and land in others; it matters only as a marker.
From 1984 to 2025 the distance from the fixed point to the tip grows from 6.99 to 10.74 km. The growth is not smooth: in 1988, 1990 and 1991 the tip was read at about 5.55 km, then in 1992 it jumps to 8.52 km. From there to 2025 it climbs with small reversals of at most 0.4 km.
Each figure is the waterline of a single summer day, not the island's path over the whole year. The table at the end keeps all 41 readings, unsmoothed.
1897
How it formed
Gâștescu (2021) dates the island's birth to 1897, after a flood on the Sfântu Gheorghe branch. According to Giosan and colleagues (2005), in 1902 it was a mostly submerged bar about 4 km long; by 1909 it had emerged as a barrier island (citing Antipa, 1915), and by 1935 it had doubled in length.
Currents along the shore carry sand south-west and have kept lengthening the island that way. Giosan and colleagues cite lengthening of more than 100 m a year from earlier work (Giosan 1998; Vespremeanu et al. 2004), without saying which years. The more-than-200 m a year in the same study is about the underwater lobe, not the island.
Gâștescu gives 17 km for 1995. The straight-line distance measured here for 1995 is 8.42 km. The two cannot be compared: the source does not say from where to where it measures, and a straight line is shorter than the curve of a shore.
1970s and 1980s
Joining the mainland
The island's northern end attached itself to the shore. Giosan and colleagues (2005) date it three ways: in the 1970s, in the late 1970s and in the early 1980s. The source picks no year, so none appears here.
The same study links changes at the Sfântu Gheorghe mouth to a drop in sand supply after dams were built on the lower Danube in the 1970s and 1980s. That is an explanation for the whole river mouth; the 41 frames cannot show on their own what the dams did to Sacalin.
2000
How the frames are read
The series holds 41 Landsat summer frames (June to September) from 1984 to 2025, one for each year except 2012, which is missing. One pixel is 30 m. Land is separated from water with the MNDWI index (Xu 2006): a pixel is land if the index is below zero.
The threshold matters. If the threshold is lowered to −0.1, the 2025 area falls from 516 to 201 ha, and the 1984 area from 337 to 240 ha. That is why the area in the table does not prove, on its own, that the island grew; it is there to show how uncertain it is.
The distance to the tip moves less: 6.63 to 6.99 km in 1984, 10.70 to 10.74 km in 2025. But before 1992 the readings are the most fragile: in 1986 the same image gives 3.88 or 7.53 km, depending on the threshold.
2012 is missing. the TM instrument of Landsat 5 had stopped acquiring images in November 2011, and Landsat 7 is excluded after its scan defect of 2003. Landsat 8 enters the series in 2013. Frame alignment was checked against the 2025 frame: none exceeds the threshold, for six the estimated offset is one pixel (2003, 2006, 2008 to 2011), and a 2010 scene shifted by more than 5 km was replaced.
2010
What changes between years
The tips of 1984 and 2025 are 4.4 km apart, almost exactly westward; the north-south difference is about 43 m, roughly one pixel. Divided by 41 years that is 106.6 m a year. It is an average between two endpoints, not an annual rate: in between, the tip jumped back and forth by kilometres.
The figure from Giosan and colleagues (more than 100 m a year) is of the same size. It does not confirm this one, though: the study does not state that estimate's period or method.
2020
Why the area figures differ
Gâștescu (2021) describes deposition as the dominant process until 1990, erosion from 1990 to 2006 and deposition returning after 2006. He publishes 670 ha in 1990, 620 ha in 1996 and 450 ha in 2000, plus 350 m of shoreline retreat between 1990 and 2000 in the most active sector.
The series here gives 204 ha in 1990 and 333 ha in 2000. The gap with Gâștescu is large, and the source does not say what extent it measured. The sources consulted do not explain the gap; the two series are shown separately, without an average.
2025
Compare with your year
Pick a year, or two, and see how far the tips moved. For gaps of at least ten years the distance divided by the number of years also appears; it is not an annual shoreline rate.
2025
Method, limits and sources
Method. Landsat Collection 2, level 2 frames, June to September, 1984 to 2025; clouds are excluded with the quality mask. The window is the same every year: south of a line across the lagoon. Area is the count of connected land pixels at 0.09 ha each. The tip is the land pixel farthest from the fixed point (29.578° E, 44.818° N); the distance is the straight line between them.
Limits. The line is the water on one day; water level, wind and turbidity move it. At threshold +0.1 the method also counts the shallow lagoon floor as land; that threshold was excluded from the comparison. Area is land detected in the window, not cadastral area. The tip is an extreme, sensitive to a single reading. A few isolated drops (2014, 2023) are not trends.
Sources. Gâștescu (2021): 1897, 17 km in 1995, historical areas and rates, 350 m. Giosan and colleagues (2005), doi: 10.2110/pec.05.83.0393: 1909, 1935, transport, lengthening, joining the mainland. Xu (2006), doi: 10.1080/01431160600589179: MNDWI. Data from the Landsat 4, 5, 7, 8 and 9 satellites, images courtesy of the U.S. Geological Survey, public domain, retrieved through Microsoft Planetary Computer.
The measured series, year by year. Summer waterline. 2012 is missing.
Year Area (ha, uncertain) Tip distance (km) Tip (°E) 1984 336.6 6.99 29.509 1985 248.9 7.05 29.510 1986 272.1 7.53 29.501 1987 177.1 7.51 29.502 1988 116.8 5.55 29.528 1989 239.0 7.61 29.502 1990 204.1 5.58 29.530 1991 156.3 5.55 29.528 1992 324.1 8.52 29.485 1993 329.0 8.48 29.485 1994 310.0 8.42 29.487 1995 332.6 8.42 29.487 1996 385.3 8.68 29.484 1997 298.4 8.43 29.487 1998 322.3 8.59 29.486 1999 310.4 8.56 29.487 2000 332.6 8.59 29.486 2001 332.9 9.11 29.480 2002 328.1 8.76 29.486 2003 355.0 9.50 29.471 2004 315.7 9.14 29.480 2005 338.6 9.50 29.470 2006 372.4 9.56 29.470 2007 408.1 9.56 29.472 2008 389.0 9.73 29.469 2009 401.9 9.73 29.469 2010 341.7 9.86 29.466 2011 396.6 9.89 29.463 2013 360.2 9.98 29.464 2014 248.8 9.87 29.465 2015 305.2 9.90 29.464 2016 351.7 9.86 29.464 2017 407.1 10.31 29.462 2018 366.8 10.14 29.467 2019 467.3 10.15 29.467 2020 545.4 10.45 29.460 2021 456.8 10.48 29.459 2022 536.0 10.46 29.460 2023 348.3 10.66 29.455 2024 517.3 10.77 29.453 2025 515.6 10.74 29.454 2025
If you want to go there
The page measures the outline from space; it says nothing about current access or visiting rules. Ask the Danube Delta Biosphere Reserve Administration before planning a trip.