· hatched: Indian Illegally Occupied Jammu Kashmir (disputed) · dashed: Line of Control · grey: none in footprint
HISAABHazard, Impact & Simulation Analysis for Asset-Based risk · AI Geo Navigators
What makes up the yearly loss
Cost by return period · approximate
Expected loss in an average year, by district
Ten districts with the highest expected yearly loss
Hardest-hit districts
Peril ledger · what stands inside each footprint
1-in-100 or the nearest modelled scenario. Counts from the source's own crossing tables.
People inside the 1-in-100 flood footprints, by province
Withheld, and why
CVE = exposure × district weight (High 100% · Medium 50% · Low 20%) × sector weight (High 100% · Medium 30% · Low 10%). Sector bands are the circular's Annexure-A Table 2. District band is a hazard proxy for the NDMP 2025 flood score the circular names: the share of the district inside the 1-in-100 riverine, flash and urban flood footprints — none is Low, the rest split at the median. Add an ndmp column (extremely low / low / medium / high) and it wins. Unknown districts score High. Scoring runs in your browser; nothing is uploaded.
District band · AGN flood-footprint proxy (not the NDMP score)
Type coordinates or click the map below. River flood and earthquake are worked out for the location's 250 m cell with the same rules as AGN's engines; landslide, flash flood, cyclone, heat and drought are the tehsil's layers. Needs the AGN server (serve.py).
Factories, warehouses, branches, schools or loan collateral: each location is placed in its district and gets that district's hazard profile and average loss rate. The file is read in your browser. The point check then sends each row's coordinates, building type and value to this platform's own server (not to any other party) and nothing is stored there.
Where they are
Pick one or more design events and an area. Footprints and what stands inside them come from the hazard layers; under a CMIP6 climate the same footprint keeps its size but returns more often. Loss comes only from the CLIMADA plugin, for the perils in its run.
Each simulated year draws an event for every chosen peril and district from its return-period curve. How far apart districts can be in one year is set by the dependence: 0 = every district independent, 1 = the same return period everywhere (how hazard maps are drawn). Exposure only; loss is in the loss engine panel below.
Loss figures (average annual loss, 1-in-200, VaR, TVaR) come from AGN's own engine, run outside the platform on CLIMADA. The platform does not install or import it: the engine drops one results file into the plugin slot, the build checks it against the contract, and every loss card fills. Until then each card is empty and marked "Needs plugin with CLIMADA". The source's scenario losses are not shown.
Plugin slot
Calibration · the run against recorded disasters
More tests · every recorded monsoon season, and earthquakes not used in the fit
Recent years · 2025 and 2026 (to 30 Sep) replayed through the run
Combined modelled hazards · Pakistan
Flood damage beyond buildings · crops, livestock, roads, railways, power, telecom, irrigation
Outdoor air pollution and health · model estimate, not recorded deaths
Heat and health · model estimate, not recorded deaths
Year loss table · distribution
Hazard inputs already in the platform
Footprint grows with return period
A larger event should cover at least as much ground. Where it does not, the source layers disagree with each other.
Registers · this bundle against the source
Schools in the riverine flood footprint · records and schools
The source writes one row per hazard band, so rows over-count. The platform counts distinct positions.
Unit and geometry faults found at source
What the platform shows and withholds
Data notes
Rain gauges against ERA5-Land · wettest day
GHCN-Daily gauges with 10 or more complete years in 1980-2025. Gauge days are multiplied by 1.13 to compare with ERA5-Land's rolling 24 hours (WMO-No. 1045). The factor is gauge over ERA5-Land; it scales the design rainfall of districts within 250 km.
Risk index · method
CMIP6 climate projections · what is behind every climate-change layer
Highest districts