An 81-location indicative flood-risk view using live Open-Meteo precipitation forecasts. River-gauge, satellite and official-warning feeds are not connected.
Study label: Madisuk Crypto User-defined project label; not a verified geographic location.
An illustrative view of 81 provincial capital locations. These are not sensor sites or verified flood observations. Click a province to inspect a scenario; values change automatically but do not reflect real conditions.
81 approximate provincial capital coordinates (WGS-84); these are not gauge locations or administrative centroids. Hover and click to explore the simulation →
The table uses the selected location’s live Open‑Meteo forecast window to calculate rolling rainfall depths and intensity. Return-period multipliers are an indicative screening proxy, not an official Turkish IDF curve or a design storm.
Select a province to calculate from its live forecast.
| Duration | Live max | 2 yr | 10 yr | 50 yr | Intensity |
|---|---|---|---|---|---|
| Live data bekleniyor… | |||||
Yağış ve debi verisi geldiğinde süre bazlı uyarı burada gösterilir.
Holt linear trend and Ridge autoregression AR(6) learn from the recent hourly Open‑Meteo window. Four horizons are compared and model disagreement is reported as uncertainty.
Schematic forecast surface around the selected city. It is not a satellite raster or an official inundation map.
Holt and Ridge AR results will be compared here.
A five-step walkthrough for emergency managers, municipal teams and analysts. No prior training required.
Click “Open the demo console” at the top of the page. The illustrative national view loads automatically. It is not a public flood-monitoring service.
The four cards at the top show the national risk index, number of active alerts, how many sensors are reporting, and the current peak rainfall. A rising risk index or a jump in active alerts is your first signal to look closer.
On the simulated regional risk map, each node represents a provincial capital location, not a monitoring station. Colours and inspector figures are fictional scenario values. Click a node to explore the interface; do not use these figures for decisions.
The simulated event feed generates fictional examples, newest at the top. The 72-hour rainfall chart is an illustrative scenario, not a forecast; do not use it for safety decisions.
High and Severe colours here are generated by the simulation and are not real warnings. Never brief responders, issue public warnings or make evacuation decisions from this page. Follow official meteorological and emergency-management authorities instead.
Low, Moderate, High and Severe are only four colours for fictional scenario scores. They do not describe current conditions or official alert levels.
This is a demonstration build. All sensor readings, alerts and forecasts shown are simulated for illustration and do not represent real conditions.
The four-source pipeline below is a design proposal, not implemented in this demonstration. This build does not ingest observations, run a hydrological model or issue genuine alerts.
Proposed: poll verified sources and normalise them to a common time base.
Proposed: match observations to stations, basins and provinces; check data quality.
Proposed: calibrate and validate a rainfall-runoff and routing model.
Proposed: validate any risk index and alert thresholds against independent events.
Observed and forecast precipitation, temperature and wind that drive the rainfall input to the model.
Water level, flow velocity and discharge measured at river stations along each basin.
Synthetic-aperture radar maps standing water through cloud, day or night, to confirm flooding on the ground.
Soil-moisture, land cover and a digital elevation model determine how much rain runs off versus soaks in.
The equation below is a proposed design only. The page's displayed scores are generated from random regional baselines and drift, not from these four measured factors:
Risk = 0.40·(river level ÷ flood stage) + 0.30·(forecast rainfall) + 0.20·(soil saturation) + 0.10·(upstream trend)
The simulated score is bucketed into Low (0–24), Moderate (25–49), High (50–74) and Severe (75–100). The displayed confidence is illustrative, not based on sensor coverage or model validation.
app.js using an unseeded random simulation — province baselines are weighted by region (Black Sea & Southeastern Anatolia run higher) and then drift over time on a 3-second tick. No network requests are made and no data leaves your device. To make it operational, each source above would be wired to a verified real API or sensor stream in the ingest layer.
A future operational system would need verified gauge, rainfall and soil-moisture feeds. None are connected in this demo.
A future system could use validated synthetic-aperture radar imagery. No radar imagery is used in this demo.
No weather ensemble or calibrated hydrological model runs in this demo. A future model would need independent validation.
Current events are fictional interface examples and are not sent to municipalities, AFAD or the public.
Every province in Türkiye, from the Eastern Black Sea to the Mediterranean basins, in a single console.
Values visibly change, but only because of a local random simulation; no fresh external data is received.