What is the critical rainfall of an element?

The accumulated rain at which the element is first reached by a likely slope failure on its own ground, by the path of a debris flow, or by more than a set depth of runoff. It is a threshold, not a forecast: it says how much rain the place can take, not when that rain will fall.

Which models does it use?

Runoff with the SCS curve number method (USDA TR-55), slope stability with the infinite slope model and a Monte Carlo estimate of the probability that the factor of safety drops below 1, and debris-flow runout with a travel angle (Fahrböschung) along the steepest descent.

Does it replace a risk assessment?

No. It is an indicative tool to communicate the hazard to authorities and to the people who live there. An official assessment (in Peru, the EVAR under the CENEPRED method) needs field data, local soil parameters and calibration against past events.

Does my data leave the browser?

The KMZ and the results stay on your device. The map's tiles, the relief and OpenStreetMap come from public providers, which see the area and your IP address, as with any web map; the map asks for consent first.

Back to the BIM & IFC blog

Tools & comparisons · 2026-10-27 · 8 min

Critical Rainfall for Landslides From a DEM and a KMZ, in the Browser

How much rain does it take for a slope above a school to fail? From a terrain model and a KMZ of the study area, the browser computes runoff, the probability of shallow landslides and the likely path of a debris flow as a storm goes on, and gives every home, school, health centre and road its critical rainfall.

A 3D maquette of Quebrada Quirio in Chosica, Peru, cut out of the relief with earth walls: slopes coloured from green to dark red by landslide probability, water down the gullies in blue, and labels for the exposed homes, schools, health post and road

Where the rain comes down steep, bare slopes above a town, the hazard is not only the water. The ground itself gives way, and what starts as a small slide on a hillside can come down a gully as a debris flow and reach the houses at its mouth. Explaining that to the people who live there, and to whoever decides when to warn them, is hard with a static map.

This tool shows it as it happens. You give it the study area as a KMZ, it takes the terrain from the map, and it runs a storm step by step: how much of the rain runs off, which slopes become unstable as the soil soaks up the rest, where a debris flow from them would travel, and at what accumulated rain each home, school, health centre and road is first reached.

Key takeaways

Runoff by the SCS curve number, slope failure by the infinite slope model with Monte Carlo, debris flow by a travel angle: standard, documented methods.
Every exposed element in the KMZ gets a critical rainfall and the cause: landslide, debris flow or runoff. Exported as CSV and GeoJSON.
Slopes that fail even without rain are shown apart: no rain threshold describes them.
It runs in the browser in a few seconds and needs no IFC model. It is indicative: it does not replace an official risk assessment.

The maquette of Quebrada Quirio at the end of a 40 mm storm: green stable ground, yellow to dark red slopes as their landslide probability rises, blue water in the gullies, and labels on the exposed elements — Quebrada Quirio, Chosica, at the end of a 40 mm storm. The relief is real; the placemarks are made up for the example.

What goes in

The study area as a KMZ or KML: one polygon for its boundary (named "study area", "boundary", "catchment"… or simply the largest one) and placemarks for what is exposed. Points, lines and polygons are read, in folders or not.
The terrain, from the map's relief or from a DEM you import (GeoTIFF or ASCII grid). With no IFC model loaded, the KMZ itself places the scene, so the map opens on the study area.
The storm, from the presets of the flood panel or drawn by hand as a hyetograph.
The soil: hydrologic soil group, prior moisture, cohesion, friction angle, depth and unit weight, with editable defaults for residual soils on steep slopes.

Exposed elements are recognised by their folder, name, description or data: "vivienda", "I.E.", "colegio", "posta", "centro de salud", "carretera", "trocha" and their equivalents in other languages. A line with no name is taken as a road. With no KMZ, the schools, health centres, homes and named roads come from OpenStreetMap.

How it computes each step

Step	Method	What it gives
Runoff	SCS curve number (TR-55), CN from OpenStreetMap land cover and the soil group	How much of the rain fallen so far runs off, and how much soaks in
Wetness	Infiltrated water over the soil's drainable porosity, more in hollows that collect water from upslope	The saturated fraction of the soil layer, cell by cell
Slope failure	Infinite slope factor of safety, 200 Monte Carlo samples of c′, φ′, γ and soil depth	The probability that the slope fails, P(FS < 1)
Debris flow	Steepest descent from the likely failures, stopped by a travel angle (12° by default)	Where the flows go and where they stop
Critical rainfall	The first step at which an element's cells fail, are crossed by debris or exceed a runoff depth	A rain threshold per element, with its cause

The curve number gives the depth of runoff for the rain accumulated so far^[tr55]. What does not run off enters the soil and raises its water table, faster in hollows that drain a large area above them. The infinite slope model then weighs the soil's strength against its weight on the slope, with the water reducing the friction^[hammond]. Soil properties are never known exactly, so each is drawn 200 times around its value and the share of draws in which the slope fails is its probability.

Debris flows start where the rain made the slope fail and follow the steepest way down. They stop where the line from their source to their front becomes flatter than the travel angle: the empirical ratio of drop to distance that debris flows reach^[corominas].

Watching the storm

The timeline replays the storm one step at a time. At the first millimetres only a few patches stand out; by the end of the storm the steep faces above the gully are orange and red, and the paths of the debris flows run from them to the valley floor. The labels of the exposed elements turn red at the step that reaches them.

The same maquette at the first step of the storm: mostly green, with a few purple patches that fail without rain and yellow spots on the steepest faces — First step of the storm. Purple: ground that fails even without rain.

The maquette after the full rain tested, 250 mm: the steep faces above the gully in orange and dark red, debris paths in brown down to the valley floor, every label red — After 250 mm accumulated. The labels turn red once the rain reaches their critical value.

Two more layers show the parts of the chain. Runoff colours the ground by how much has run off so far and makes the drainage network visible. Debris flow shows where flows from the likely failures travel (light brown) and stop (dark brown).

The runoff layer at about 120 mm of rain: the whole area in light blue, with the gullies and the drainage network in darker blue — Runoff at about 120 mm accumulated: the drainage network stands out.

The debris-flow layer, close up: brown paths from the unstable slopes down the gullies, the exposed elements labelled among them — Debris-flow paths and deposits after the full rain tested.

Slopes that fail without rain

On the example, with the default soil, 13.6 ha fail before a drop has fallen: faces too steep for a soil of that strength, usually rock or a cliff in reality. The model cannot give them a critical rainfall, since no amount of rain is needed, and counting them would give every element below a cliff a threshold of nearly zero. They are shown in purple, set no threshold and start no debris flow, and an element standing on them is flagged. This is the "unconditionally unstable" class of SHALSTAB^[shalstab]. Where they matter, the soil parameters should come from the field.

The critical-rainfall table

The table lists every exposed element, the most exposed first: the rain at which it is reached and by what. On the example, with the default parameters:

Element	Critical rainfall	Cause
Carretera Central (stretch)	1 mm	Debris flow
I.E. Inicial 520	7 mm	Landslide
I.E. N° 1201	70 mm	Landslide
Puesto de salud Quirio	90 mm	Runoff
Homes (three groups)	140 mm	Runoff

A road reached at 1 mm is not a forecast that one millimetre will cut it: it says the cells the road crosses lie below slopes at the edge of equilibrium in this model, the most susceptible class. That is the value of the table: an order of what to watch first. Calibrated against past events and local soils, values like these are what early-warning thresholds are built from. The table exports as CSV and as GeoJSON, which opens in QGIS or Google Earth.

What it does not do

It does not forecast rain or say when a slope will fail. It says how much rain each place can take under the parameters given.
It uses one soil for the whole area, with uncertainty, not a geological map. Rock outcrops, colluvium and fill all look the same to it.
Infiltration and groundwater are simplified: no lateral flow over time, no drainage between storms.
Global relief (about 30 m) smooths small gullies and steep faces. A local DEM gives better slopes.
It does not replace an official risk assessment, a field investigation, or an early-warning system's own calibration.

References

Urban Hydrology for Small Watersheds (Technical Release 55) — USDA Natural Resources Conservation Service. The curve number method and its tables of CN by land cover and hydrologic soil group.
Level I Stability Analysis (LISA) Documentation for Version 2.0 — Hammond, Hall, Miller & Swetik, USDA Forest Service. The infinite slope factor of safety with uncertain soil parameters, assessed by Monte Carlo.
A physically based model for the topographic control on shallow landsliding — Montgomery & Dietrich, Water Resources Research 30(4). Critical rainfall from topography and the infinite slope; the "unconditionally unstable" class.
The angle of reach as a mobility index for small and large landslides — Corominas, Canadian Geotechnical Journal 33(2). Travel angles (H/L) of debris flows and other landslides.

Critical Rainfall for Landslides From a DEM and a KMZ, in the Browser