Punjab's climate observatory will sit on an air monitoring network that has nearly doubled in a year
The Lahore facility promises real time tracking of air, water, soil and weather. The province already runs 59 air stations, and the observatory has no published budget or opening date.

Punjab is building what its chief minister calls Pakistan's first climate observatory, and the telling part is what it will stand on. The province already runs an air quality network of 59 stations, up from 30 when its public portal published its first reading in October 2025. The observatory is meant to add water, soil and forecasting to that base, with an AI system over the lot.
Chief Minister Maryam Nawaz Sharif announced the Lahore facility in a post on X on Sunday 15 November. It will monitor air quality, water, soil and weather in real time with an AI powered climate intelligence system, and will carry early warning and flood forecasting systems to support disaster response and water conservation. It will also hold research laboratories and environmental data for young scientists. The project was approved in May 2025 after a feasibility review that covered budget, equipment procurement and land, and the University of Engineering and Technology, Lahore, submitted a report on its technical aspects. The facility is still under construction, and no budget, completion date or equipment list has been given.
What already exists
The air side is the part Punjab can already measure. The Environment Protection and Climate Change Department launched its network on 1 October 2025 with 30 regulatory grade stations and has since reached 59, with 100 planned. In September the provincial dashboard added forecasts, satellite data and weather information, built with the International Centre for Integrated Mountain Development.
Two sums show the scale. The network grew by 29 stations in about 13 months, a rise of 97 percent. Spread across Punjab's area of roughly 205,000 square kilometres, 59 stations is one for every 3,500 square kilometres or so, and 100 would be one for every 2,050. Each station has to stand in for a very large stretch of land.
Why weather belongs with air
Putting air and weather in one building is sound engineering. An analysis by the monitoring project Hamari Hawa, using IQAir data, put Lahore's monthly average PM2.5 between 124 and 183 micrograms per cubic metre from October 2025 to February 2026, against 40 to 59 from April to September 2026. The lowest winter month was 2.1 times the highest summer month. The same analysis, using ERA5 reanalysis data, puts the average mixing height, the depth of air in which pollutants can spread, at about 830 to 860 metres in May and June, falling to about 260 metres in November and 190 metres in December. By November that is about a third of the summer depth. Part of the winter rise therefore comes from shallower, stiller air and not from emissions alone, which is why an air quality forecast is largely a forecast of mixing height and wind.
It also exposes a weak point. ERA5 is a modelled estimate, and the provincial department says it carries biases of 200 to 500 metres over the monsoon region. Suparco's ceilometer, an instrument that measures the layer directly, is the kind of ground measurement an observatory would need to anchor its forecasts, and it should name its reference instruments and publish their raw data.
Where the AI will struggle
Machine learning forecasts need long, clean records from instruments checked against a standard. The expanded network has been reporting for about a year, which means a single full smog season of local ground truth. A model can borrow satellite and global weather data to fill the gap, but its local errors will be hard to measure until a second winter has passed.
Flood forecasting is harder still. Much of the water in Punjab's rivers originates upstream, outside the province's own borders, so a Lahore centre can only forecast from rainfall and river flow data that other agencies collect and agree to share. Water and soil monitoring, the two items on the list that the air network does not cover, need new sensors in rivers, canals and fields and people to maintain them. None of the announcements names a data partner or a sensor count for either.
The winners, if it works, are researchers who get open data and officials who get a single forecast to act on. The risk is a building and a dashboard with no one responsible for the readings. December is the shallowest month in the Hamari Hawa analysis, at 190 metres, so the first test is whether the 59 stations already reporting can anchor a forecast for it.





