After another spell of wet weather along Australia's east coast, with storms, heavy rain and flash flooding across Sydney and parts of New South Wales, it is natural to ask whether our cities are shaping the rainfall that descends upon them.
This matters because most people now live in cities. If urbanisation changes rainfall, even slightly, the effects can reach large populations through flooding, stormwater design, water supply and infrastructure planning.
Satellite data have consistently shown that many cities experience more rain events than the countryside around them. The usual explanation is that cities themselves are involved: urban heat, rougher surfaces, aerosols and changed land cover can all affect how storms develop and where rain falls.
Our new study , published in Environmental Research Letters, asks a related question: how much of this data reflects real changes in rainfall, and how much depends on how we observe it?
Why we need satellites
Understanding rainfall over cities is hard.
Rain gauges accurately measure rainfall at a specific location, but are irregularly distributed and cannot fully capture how rain varies across a large city. Climate models can simulate urban weather in detail, but kilometre-scale simulations across many cities and decades remain computationally expensive.
Satellite observations help fill this gap.
NASA's Integrated Multi satellite Retrievals for GPM , known as IMERG, provides near-global rainfall estimates at high resolution, and is now widely used for studying rainfall over cities.
What the satellite data shows
We examined IMERG rainfall data across 15 of the world's largest cities, including Sydney and Melbourne. The cities span different climates and geographic settings, including both coastal and inland regions.
A clear pattern emerged. Rain events occurred more often over urban areas than over nearby rural ones. The strongest signal was not that every storm became stronger, but that satellites counted more hours in which it was raining over cities. Individual events over urban centres often dropped less water than those in surrounding areas.
In other words, the main urban signal in IMERG is more frequent rain, not heavier rain.
Different sensors, different stories
Modern satellite rainfall data combines both infrared and microwave observations.
Infrared sensors estimate rainfall indirectly from the temperature at the top of clouds. They provide broad coverage, but can miss light, shallow or warm rain because these can occur even when the tops of the clouds are not very cold.
Microwave satellites fly in low orbit and detect signals more directly linked to raindrops and ice inside clouds, making them particularly useful for identifying whether rain is actually occurring.
When we separated the IMERG data by observation type, the urban signal mainly came from microwave observations, while infrared estimates showed no urban pattern.
This does not mean the microwave signal is wrong, but it raises a potential problem for long-term studies: microwave observations have changed over time. New satellites have been launched and older ones retired, and across the cities we studied, microwave sampling frequency happened almost twice as often by 2023 as it had in 2001.
This matters because the more often a microwave sensor passes overhead, the more rain events it can detect. A light shower missed in 2002 could now be caught by one of several satellites passing within the hour.
Testing the artefact
To test whether this changing sampling affects observed rainfall trends, we compared the microwave and non-microwave with long-term averages. This meant we could separate out the result of changing satellite sampling from the actual changes in weather.
Changes in microwave sampling explained up to about 20% of the long-term rainfall trends across the 15 cities. For rainfall frequency, cities such as Lagos, London, Melbourne, Beijing, Berlin, Mexico City and Paris showed areas where more than 40% of the apparent trend could be linked to the changing observing system.
The satellites did not create the whole urban rainfall pattern. After accounting for sampling effects, the urban signal remained, but the long-term trend became smaller. So we think it really is raining more often over cities, but perhaps not as much as we thought.
Moving forward
For Sydney, we also compared IMERG with CMORPH , another satellite product, and with Bureau of Meteorology rain gauges. CMORPH showed a similar urban pattern, though the two products are not fully independent because they use overlapping microwave observations.
The gauges are a more independent check, but with too few stations outside the urban core, in Sydney and most cities, the true magnitude cannot yet be confirmed on the ground.
Satellite rainfall data is now used everywhere, in climate science, flood risk, agriculture, insurance and water planning. In many regions it is the only consistent rainfall record over large areas. Our results are a caution: part of an apparent trend can come from the changing observing system rather than real change.
As for why cities get more frequent rain, the likeliest explanations are familiar: urban heat that lifts air, rougher surfaces that nudge winds upward, and aerosols that alter cloud droplets. The signal is real. The task now is measuring it properly.
Shankar Sharma receives funding from the Australian Research Council.
Andy Pitman receives funding from the Australian Research Council.
Jason Evans receives funding from the Australian Research Council.
