How Can Satellite Shoreline Data Inform a Bali Coastal Property Decision?

A property listing may show a generous strip of sand in front of the house, while a newer overhead image seems to put the sea almost against the same frontage. It is tempting to interpret the difference as years of erosion. Yet neither image tells you how high the tide was, what the surf had been doing or whether the camera captured the same part of the beach. A convincing coastal decision starts by resolving those differences, not by measuring a gap on two photographs.

This stretch of southwest Bali should not be treated as one beach with one trend. Batu Bolong sits in Canggu, Badung Regency; Nyanyi is farther northwest in Tabanan. River mouths, rocky sections, sand beaches and areas influenced by coastal structures interrupt the shoreline between them. A channel migration can change one short reach while the neighbouring sand beach moves differently. Even within Canggu, a rate derived for one transect may be irrelevant a few hundred metres away.

Evidence becomes stronger in stages. An undated marketing photograph provides a snapshot. Georeferenced images with known acquisition dates allow a preliminary comparison. Annual shoreline positions produced using a consistent sea-level reference can indicate a longer-term direction, provided that their quality and coverage are checked. A site survey and coastal engineering assessment then deal with elevation, beach profile, drainage, foundations and past shoreline interventions. Those last questions are not resolved by finding a higher-resolution basemap.

Published research gives buyers a useful warning against simplistic conclusions. A 2016–2024 study of Canggu Beach found predominantly seaward movement in parts of its study area, alongside local erosion near river mouths and structures. Its spatial scope includes the Batu Bolong area, not a ready-made assessment of Nyanyi. The point of reading the history is therefore to identify a credible reason for further examination of a particular frontage, not to obtain a universal Bali erosion score.

A shore has several lines that look similar on a map but answer different questions. The instantaneous waterline separates sea from visible land when a satellite passes overhead. Its position moves with the tide and breaking waves. Ina Coastline, developed on Indonesia's Piksel platform, instead estimates a representative annual shoreline close to mean sea level, or MSL, using Landsat imagery and tidal modelling. That median line is useful for examining long-term change. It does not show the highest reach of the sea during a storm, and it is not a surveyed boundary of a property.

The platform describes annual vector shorelines covering 1987–2025 nationally. Piksel's July 2026 article also mentions a Landsat source archive extending back to 1984; the image archive and the published temporal extent of a derived product are different things. In the annual layer, year, certainty and tide_datum describe the observation. The separate rate-of-change points have fields for usable observations, temporal span, change rate, statistical significance and estimated error. Those fields would make a local measurement auditable. They should not be presented as values for Bali merely because the national specification lists them: an accessible, georeferenced export for the selected Batu Bolong–Nyanyi reach was not established in this review.

BIG also operates a different resource, the Garis Pantai mapping service at 1:25,000 scale. Its 2022 update combines materials of varying dates and origins, including earlier surveys and imagery. TIPGPN identifies the category of shoreline, while DTMVER refers to its vertical datum and THNSBDATA to the source year. The service can represent high-water, mean-sea-level or indicative lines. These should not be overlaid as if they were two comparable years of the same physical measurement. Neither the indicative line nor the national mapping scale settles title boundaries, coastal setback obligations or permission to build on an individual parcel.

For a meaningful local example, start with the sandy frontage of Batu Bolong and distinguish it from rocks, structures and nearby channels. The beach can be found at approximately 8.66015° S, 115.13011° E in WGS84 according to a geographic gazetteer. These are search coordinates only: they are not survey vertices, an extracted shoreline or a basis for measuring metres to a building. The 2025 academic study identifies change around Batu Bolong and Nelayan within its Canggu study area, but it does not provide a verified parcel-specific transect or positional uncertainty for a house under consideration. Its mapped slices are evidence about the studied area, not an address certificate.

Nyanyi requires a separate geometry. It lies northwest of Batu Bolong in Tabanan rather than Badung, beyond the scope that can safely be assigned to the Canggu transects. A reproducible Nyanyi analysis would lock the actual beachfront into one coordinate system, identify a common shoreline reference, and compare observations at the same cross-shore location. Without that site-defined line, the correct outcome is a narrower conclusion: the techniques and nearby studies are informative, but they do not establish how far Nyanyi's coast has moved or where the legal edge of a particular lot lies.

The strongest geographically relevant public example in this review comes from Damayanthi, Pujianiki and Tonyes, published in 2025. Their Canggu Beach study uses Sentinel-2 scenes from 2016, 2018, 2020, 2022 and 2024, with 247 cross-shore transects analysed using shoreline-change methods. The paper includes the successive extracted coastlines and maps distinguishing erosion from accretion. Its imagery was chosen with attention to cloud conditions and a stable sea-surface elevation range in HYCOM. These are years shown in an academic investigation, not five independently downloaded Ina Coastline MSL vectors for a house on the market. Exact scene product identifiers and tide readings for each illustrated year are not established in the available material.

The findings make one thing clear about Batu Bolong: the direction of coastal change varies locally. Across the paper's accreting transects, mean net shoreline movement for 2016–2024 was +16.61 metres and mean end-point rate was +2.26 metres per year. The eroding group had corresponding means of −8.8 metres and −1.2 metres per year. These are summaries of different parts of the defined Canggu study area, not simultaneous growth and retreat at the same point. The authors describe much of the Batu Bolong–Nelayan area as accreting while also reporting small erosion segments around Batu Bolong, river mouths and affected structures. That apparent tension is precisely why a buyer needs a transect linked to the actual frontage. The article does not supply a defensible position error or statistical confidence interval for the property being considered.

A second useful record is the 2019 study by Astiti, Osawa and Nuarsa. Its geography is broader, running from Mengening near the Badung–Tabanan boundary through the Canggu coastline to Seminyak. It uses 100 transects and Sentinel-2A imagery captured on several dates in 2015–2019. Four acquisition dates are explicit: 25 October 2016, 2 July 2017, 18 April 2018 and 2 February 2019. The dates span different parts of the seasonal cycle, so simply putting those waterlines next to one another does not create a controlled seasonal comparison. The paper is also internally inconsistent about the December 2015 acquisition date, which should be resolved against the original product before treating that scene as precisely dated evidence.

For its own extensive study area, the 2019 paper reports average erosion of 1.42 metres per year. It describes erosion in its first two annual intervals and accretion in the next two. The authors also set out a tide correction example for the 2 February 2019 scene, including an acquisition time of 10:19 and a water-height input obtained from a tide application. That illustrates how much more information is required than an image timestamp alone. It does not establish a tide height in an explicitly matched vertical datum for a modern property survey. Nor can its regional 1.42-metre result be averaged with the positive rate for one group in the 2025 paper; the periods, shoreline definitions and spatial samples differ.

This leaves a deliberately limited but practical finding. There is credible research documenting both advance and retreat within the Canggu stretch studied around Batu Bolong from 2016 to 2024. There is no verified cross-shore movement or uncertainty value for a particular villa, and the study's results should not be carried northwest to Pantai Nyanyi. An absence of a Nyanyi measurement is a data gap, not evidence that the coast there has been stable. Likewise, an accretion label near Batu Bolong does not guarantee that nearby foundations are free from erosion or wave exposure.

The route to a stronger conclusion is documented by Ina Coastline, even though local vector access could not be confirmed here. A property-level GIS review would need a common transect, annual shoreline quality flags, a usable observation count, a valid year span, and the rate's error and significance alongside the plotted lines. Piksel labels fields such as certainty, valid_obs, valid_span, rate_time, se_time and sig_time for that purpose. Until those are obtained at the relevant Bali coordinates, a figure drawn from the platform's quantitative demonstration on Java must remain a Java result. It cannot be converted into a Canggu–Nyanyi erosion rate by changing a location name.

A beach can appear to shrink between two satellite passes without losing a comparable amount of sediment. At low tide, much of the shore platform may be exposed. At high tide, the visible edge of the sea sits farther inland; waves and foam can make the boundary look closer still. The effect is especially pronounced on a gently sloping beach. An annual median shoreline referenced to mean sea level reduces dependence on a single water level, but it cannot tell you how far storm surf reaches or how the beach behaves on the day you inspect a house.

Season is another source of misleading contrasts. Sand is exchanged between the visible beach, nearshore bars and neighbouring segments as wave and current conditions change. A channel at a river mouth may move at the same time, altering the shape of one small stretch. The 2019 Canggu–Seminyak study itself illustrates the problem: its verified Sentinel-2A image dates fall in October, July, April and February. Those scenes were selected with the benefit of low cloud cover; they are not controlled repeats of a single tidal and seasonal state. Even two low-tide views can disagree because the beach has recently experienced different wave conditions.

Shoreline definitions must also match. The five Sentinel-2 study slices from 2016–2024 are extracted observations under the study's cloud and HYCOM selection method. They are not the same data product as Piksel's Landsat-derived annual median MSL coastlines. A HYCOM sea-surface elevation selection range does not establish a verified tide value at an individual property. In the 2019 work, the authors calculate a correction using an external tide application for one scene; the resulting number should not be mixed with another service's coastline unless the local time, vertical datum and beach slope are all compatible. Removing a tidal effect does not remove wave run-up or seasonal redistribution.

Resolution adds a further trap. Copernicus lists native Sentinel-2 spectral bands at 10, 20 and 60 metres. Resampling a coarser band to a finer grid makes its pixels smaller on screen, not its underlying information more precise. Water-detection algorithms may interpolate a boundary between pixels; that output is not a certified survey line. Wet dark sand, foam, turbid estuary water, reef and bright coastal structures can all interfere with classification. Image alignment matters just as much: viewing two basemaps at the same zoom does not prove they share a precise reference geometry.

A defensible comparison therefore documents the scene identifier and acquisition time, the surface or waterline definition, tidal context, season, cloud quality and the exact coastline segment. If the detectable difference is no larger than the data's positional uncertainty, reporting a measurable retreat would be premature. For a buyer, that outcome is a reason to acquire better evidence rather than an instruction to believe either the most reassuring or most alarming photograph.

Historical imagery helps determine the next investigation, not the property's fitness for purchase. Consider three outcomes. First, well-matched shoreline observations show no convincing landward trend at the selected frontage. That reduces the evidence for sustained retreat over the observed period, but it cannot rule out storm-wave exposure, hidden scour or future changes. A movement smaller than the uncertainty of the mapped line is not a reliable finding of stability. A quiet-looking time series is one piece of the decision, not a coastal safety certificate.

Second, several comparable shorelines consistently move inland near the property, possibly accompanied by exposed foundations, fresh fill, damaged access steps or a changing river channel. That is a reason to pause an irreversible commitment until an independent coastal engineer can assess the site. There is no defensible universal buffer in metres that makes every Bali house safe or unsafe. The significance of a mapped shift changes with beach slope, land elevation, ground conditions, coastal structures and the uncertainty of the images. A buyer needs those physical facts before interpreting a trend as a consequence for the building.

Third, the evidence may simply be incomplete. Imagery comes from mismatched months, a historic scene lacks a trustworthy acquisition record, the tidal reference is unknown or the relevant vector layer cannot be retrieved. The appropriate conclusion is that the trend has not been measured, not that the beach is stable. Additional appropriately timed imagery and any local coastal monitoring records are more valuable than a precise-looking extrapolation. Where a deposit would become non-refundable, unresolved physical evidence is relevant to the timing and conditions of the commitment itself.

Different frontages demand different engineering questions. On an open sand beach, the assessment may need a measured beach profile, ground elevations, evidence of past high-water events and the position of the building relative to erosional features. At an estuary, the channel alignment and movement of sediment can be at least as important as the beach edge facing the sea. Beside a seawall, the engineer needs its construction and repair history, signs of undermining and the condition of adjoining beach sections. A satellite sequence can show that a structure appeared; proximity alone cannot establish whether it caused the subsequent change.

Land evidence must follow its own path. A competent surveyor establishes the parcel's registered geometry and surveyed features. Applicable planning restrictions, land rights, building approvals and the status of any shore protection works require address-specific records and, where necessary, independent legal advice. BIG's mapping layers are valuable context, but neither their scale nor the appearance of a line makes them a title survey. Equally, a positive engineering view of the beach cannot confirm a seller's legal entitlement or the permitted use of the house.

The published Canggu studies support a clear due-diligence threshold: whenever the evidence is locally contradictory, repeated retreat appears near a structure, or the quality of the imagery is inadequate for the requested measurement, the decision moves beyond a desktop comparison. That applies even if a nearby section around Batu Bolong is described as accreting. Nyanyi is farther away and needs its own evidence, not a borrowed Canggu average. A reasoned purchase assessment combines comparable historical observations, a site visit, topographic and boundary surveys, and documents tied to the property being acquired.