Documented cluster · Thon Buri
Talat Phlu railway-market rows
ตึกแถวย่านตลาดพลู
Documented cluster · Thon Buri
ตึกแถวย่านตลาดพลู
Frontage and depth are the short and long sides of each footprint's minimum-area bounding rectangle, computed by scripts/build-shophouse-spine.py over the Overture/OSM screen. A footprint in this set has not been visited: the shape says shophouse, and nothing here says it is one.
Computed per footprint from depth, plot and neighbours — not merely listed. A clause counts as binding when this building's own geometry triggers it.
MR55 ข้อ 223 of 51
Each คูหา: width ≥ 4 m (column centreline to centreline); depth ≥ 4 m and ≤ 24 m; ground-floor area ≥ 30 m². If depth exceeds 16 m, an uncovered void of ≥ 10% of ground-floor area between 12 and 16 m depth.
The 4 m module is a floor set by law. Tirapas's surveyed 4 m module is the regulation made visible.
MR55 ข้อ 4121 of 51
Setback from public road centreline: 3 m; 6 m where the road is under 10 m wide, for buildings over 2 storeys or 8 m — including ตึกแถว.
THE decisive clause. An old shophouse occupies its whole plot because it predates this. Demolish it and the replacement must step back onto ground that, on a 12 m plot, is most of the site — so on many plots you cannot lawfully rebuild what you tear down.
MR55 ข้อ 41 of 51
Continuous construction ≤ 10 คูหา and total row length ≤ 40 m, regardless of ownership.
Why long shophouse streets break into readable blocks with gaps. The rhythm you see from the pavement is a fire rule.
The interesting number is the third. Where a row exceeds today's cap, the building standing there could not be built again — which means demolition is not a reversible decision, whatever replaces it.
Each footprint is placed on two axes: the appraised land value underneath it, split at the citywide median of ฿3,568,750, and its plot depth, split at 13.6 m — below which a 2 m setback takes a serious share of any compliant replacement.
Genuinely exposed12
High land value, deeper plot.
The land is worth enough to justify clearing and the plot is deep enough to absorb a setback. On present rules these are the ones that go, and an argument for keeping them has to be made on carbon, tenure or social capital rather than on the code.
Reuse is the easy answer21
Lower land value, shallow plot.
Little pressure to clear, and the setback would bite hard if anyone tried. Nothing needs defending here yet — which makes it the cheapest place to test what adaptive reuse actually costs.
Not under pressure16
Lower land value, deeper plot.
Neither the market nor the code is currently forcing a decision. Worth surveying precisely because there is time to do it properly.
2 of the 51 footprints here carry no quadrant, so the split above sums to 49 rather than 51. That is a join artefact, not a gap in the screen: the pressure layer classifies all 2,433 footprints, but the spine joins the two sets by centroid rounded to five decimal places, and 241 of the 2,433 miss that key by a metre or so. Those buildings have a quadrant; this page cannot currently name it.
Appraised value is the base for transfer tax and sits below market, so every figure here is a floor, not an estimate.
Storeys are known for 1 of the 51 footprints here — 50 carry none, and are left null rather than filled with a plausible number. Across the whole spine the same is true of storeys (for two-thirds of the set), tenure, structural condition, road width, build year.
A useful correction to palace-centred heritage: rail, canal, food trade and ordinary rowhouses made a second old Bangkok on the Thonburi side. The network matters more than a single trophy building.
Rail, road and canal market shophouses
Documentation: Bangkok News / Around Online neighbourhood guide ↗. Every computed figure on this page comes from scripts/build-shophouse-spine.py, which states each rule in its header.
Arkara, N. (2026). Talat Phlu railway-market rows — documented shophouse cluster (51 screened footprints) [Data set]. Shophouse Metropolis, BKKx. https://shophouses.nonarkara.org/shophouses/cluster/talat-phlu
All 32 clusters · the pressure map · the datasets behind them