Risiko Banjir N. Sembilan NS Flood Risk · indeks HAND / HAND index
Kaedah / Method
DEM: Copernicus GLO-30, 30 m. It is a surface model (canopy and roofs included); vertical error on flat land is 2–4 m, which is the same size as the HAND thresholds used here.
HAND (height above nearest drainage): DEM minus the elevation of the drainage cell each cell flows to. Pits and depressions filled, D8 flow direction (pysheds). Drainage = cells with upstream area ≥ 5 km² (a 10 km² run gives 84,600 people <2 m and 173,800 <5 m). Small streams under 5 km² are not drainage, so low spots beside them are not flagged.
HAND is a near-drainage index, not a flood extent. It does not model rainfall, river stage, drains, culverts, dams, tides or defences. Use it to rank places, not to say a house will flood. About 4% of land (61 kampung points) drains off the DEM edge and has no HAND value.
Bands: High <2 m, Medium 2–5 m, Low >5 m, from HAND at the kampung point. Exposed people are counted within 500 m of the point.
Population: WorldPop 2020 constrained 100 m (1.19 million on NS land) for exposure; DOSM Census 2020 for district and mukim totals. The two differ (WorldPop is modelled). The 65+ and under-15 shares are district-level OpenDOSM modelled 2020 estimates; no mukim or kampung ages exist.
Buildings: Overture Maps release 2026-09-23. Schools, clinics: OpenStreetMap. Relief centres (PPS): JKM list, all 241 (capacity 36,247 as listed by JKM; 8 flagged not active and left out of the gap and nearest-PPS). Locations are geocoded; 128 are at mukim-centre (zone) level only, so distance is approximate. A listed PPS is not necessarily open. PPS gap per mukim: people in HAND <2 m (WorldPop) minus listed capacity of active PPS in that mukim (JKM mukim field; boundary units split as Pekan/Bandar/Mukim X are summed to the parent mukim; PPS whose mukim label matches no polygon were placed by point-in-polygon). Mukims where no PPS carries the label are marked *: JKM may label nearby PPS to another mukim. Capacity is nominal.
Past floods: 135 geocoded incidents from news and agency reports, 2011–2026. Thin before 2016, so zero does not mean no floods. A kampung counts distinct dates of incidents within 1 km. Dec 2021 satellite extent: EOS-RS Sentinel-1 proxy (9 vs 21 Dec 2021); unreliable in urban and vegetated areas.
Official hotspots: 81 JPS/state flood hotspots, data.gov.my “Senarai Hotspot Banjir di Negeri Sembilan” (2022-era list). Located by matching names to our kampung points and flood incidents in the same daerah, then Nominatim (cached, results outside the stated daerah rejected). Hollow markers are area-level matches (town, mukim or river name), shown for reference and not used for the 1 km card test. The unlocated hotspots could cover some kampungs counted as “not on the list”, so that count is an upper bound. The list is 2022 and may be out of date.
Sirens: 17 JPS NS telemetry sirens (positions only, no status). Low bridges / roads at risk (v1): OpenStreetMap bridges tagged trunk, primary, secondary or tertiary, with HAND <2 m sampled at the bridge point. A bridge crosses the drainage, so HAND is low by construction and about half qualify; OSM has no deck height, so this does not say a road will be cut. Colour = live status of the nearest JPS NS station (grey if no reading within 2 h).
Gauges: JPS Negeri Sembilan and the national infobanjir feed, live.