India and the Middle East · live
Climate and water hazards, mapped from open data.
This map shows climate and water hazards in India and the Middle East. India has seven hazards, including street flooding, heat, groundwater loss and landslides. The Middle East has five, including dust storms, wet-bulb heat and flash floods. All data comes from open satellite and weather sources. We test each model on cities that we did not use to train it.
In India the map covers Delhi, Chandigarh, and the Punjab, Haryana and Himalaya area. It also covers Mumbai and Bengaluru for flood and heat. In the Middle East it covers nine Gulf cities. The map is free to use and read-only.
Two regions
Two regions, one system
India
Seven hazards and about 16,000 data points. We test the flood, heat and landslide models on cities that we did not use to train them. We publish each score, including the low ones. The map also shows dams, rivers, groundwater loss, drought and possible forest loss. Most of this data covers North India.
The Middle East
Five hazards that apply to the Gulf. They are dust storms, wet-bulb heat, wadi flash floods, cyclones and air quality. Wet-bulb heat measures conditions that the human body cannot survive. The data comes from Open-Meteo and GDACS and covers nine cities. Each hazard uses a calculated index. We have not yet trained a model for them.
Use the switch at the top left of the map to change region. Both regions use the same system. If a hazard has no data for a place, the map shows nothing. It does not show an estimate. The Coverage page lists every place and hazard. It also shows which hazards use a trained model and which use a calculated index.
India
What the map shows
Flood
The map shows which streets hold water after rain. It uses terrain and drainage data at about 670 m resolution. Click a point to see the reasons for its score. A typical reason is low ground that is far from a drain. This hazard also shows recorded flood points, live rainfall against past floods, and known flood points along a route.
Heat
The map shows the daytime surface temperature of each area, measured against the least built-up ground in the same map extent. It uses satellite surface temperature, which is not the air temperature people feel. In the pre-monsoon season dry open ground is often hotter at the surface than the built-up city, so many urban areas read below their reference. The model explains the difference by how built-up, green or reflective each area is. Every city uses the same colour scale. A 2030 view adds a climate outlook. It shows four extreme-weather measures under two warming paths, from four climate models, across nine areas.
Dams & rivers
The map shows large reservoirs and the river network of North India. This includes more than 5,400 river sections and 4,000 canals. Click any dam or river section to trace the water downstream. These are public records, not a forecast. This hazard also shows four Himalayan pilgrim valleys that heavy rain can affect: Kedarnath, Hemkund, Manimahesh and Haridwar. It combines IMERG rainfall data with the number of visitors each month.
Groundwater
The map shows how fast the water table falls in Punjab and Haryana. It uses 22 years of CGWB well records. Each district has a measured specific yield. This value converts the drop in water level into the volume of water lost. The water table falls fastest in the cotton area of south-west Punjab. It is more stable in districts that canals supply. A check against GRACE satellite data runs every week.
Forest loss
The map shows 148 points where Sentinel-1 radar detected a possible loss of tree cover. They cover 8 areas of North India. We have not yet checked these points against Hansen or GLAD records. Each point is therefore a possible loss, not a confirmed loss.
Drought
The map shows how far this month's rainfall is below normal. It uses the SPI-3 index over 34 years of CHIRPS rainfall data. Ludhiana is live now. We will add more districts in Punjab and Haryana. We have not yet tested this index against recorded IMD drought declarations.
Landslide
The map shows 517 recorded landslides from NASA COOLR in Himachal and Uttarakhand. Each one has a date and a size. Next to them is a model that ranks where landslides can occur. The model uses terrain, monsoon rainfall and distance to road cuts. It scores 0.85 in Himachal and 0.81 in Uttarakhand on data it did not train on. It shows where, not when. A test of a rainfall trigger found no skill, so there is no daily landslide forecast. Use the switch to change between the recorded landslides and the model.
The map also shows reference data. This includes 289 Delhi wards, 63 Chandigarh sectors, 25,956 roads and 7,436 rail lines. It also includes 8,577 hospitals, schools, police stations and fire stations in Delhi, Mumbai, Bengaluru and Chandigarh. Ward and sector boundaries cover Delhi and Chandigarh only.
Accuracy
Accuracy on data the model did not see
A model always scores well on the data used to train it. A useful test uses a different city, or a different part of the map. The table below gives that score for each hazard. It includes the low scores.
The evaluation page gives the full method, the score for each city, and the confidence intervals.
Method
How we build the map
Open data sources
We use satellite data for terrain, surface temperature and built-up cover. We also use public flood reports and dam records. All sources are free and you can trace each one. You can download the data behind any view.
Models you can examine
We use simple models that a person can examine. The map divides each prediction into the factors that caused it. You can therefore see the reason for a score.
Limits shown on the map
Scores compare places inside one city. They are not probabilities. A place with no mark has no data. It is not necessarily safe. The map shows the uncertainty and the gaps in the data.
The data
The data behind the map
The map reads from a structured hazard database. Every record states where the data came from. The same records are published as static JSON files that anyone can download. Insurers, city planners and infrastructure operators can use this data to build a risk score for a street or a ward. The data is free and open today.
Open the map
Click a street to see its hazard score and the reasons for that score.