Do cities receive more rain, and what are the satellites missing?
Sydney has experienced more than its fair share of flash flooding and heavy downpours in recent years, prompting an obvious question: does the city itself influence the rain that falls on it? Can concrete, roads and high-rise buildings actually change local rainfall? For anyone impacted by extreme precipitation in urban areas, it is more than an academic question.
More than half of the world’s population now lives in cities, so even small changes in rainfall can have important consequences for flood risk, stormwater systems, water security and transport infrastructure. Our new study, published in Environmental Research Letters (https://iopscience.iop.org/article/10.1088/1748-9326/ae7135), examined rainfall across 15 major cities using satellite observations. We found that cities generally experience rainfall more frequently than surrounding rural areas, but we also discovered that part of the apparent long-term trend reflects changes in the satellites observing system rather than changes in the atmosphere itself.
What satellites reveal
Measuring rainfall over cities is surprisingly difficult. Rain gauges provide accurate observations but are too sparse to capture how rainfall varies across an entire metropolitan area, while high-resolution weather models require enormous computing resources to simulate urban weather over many cities and decades. Satellite observations help bridge this gap.
We analysed NASA’s Integrated Multi-satellite Retrievals for GPM (IMERG), one of the world’s most widely used satellite rainfall datasets, across 15 major cities, including Sydney and Melbourne. A clear pattern emerged: cities generally experienced rainfall more frequently than surrounding rural landscapes, even though individual storms were not necessarily more intense. In Sydney, rainfall occurred more often over the urban area than over nearby bushland and coastal waters, while Melbourne showed a similar contrast with its drier inland surroundings. The satellite record therefore suggests that urbanisation primarily influences how often rainfall occurs, rather than making every storm heavier.
The hidden influence of the satellites
To understand whether this pattern reflected the atmosphere or the observing system, we examined how IMERG measures rainfall. The dataset combines information from two types of satellite sensors: infrared instruments, which estimate rainfall indirectly from cloud-top temperatures, and microwave sensors, which detect signals more directly related to raindrops and ice particles within clouds. When we separated the observations by sensor type, the urban rainfall signal was found almost entirely in the microwave observations. This did not mean the microwave measurements were unreliable. Instead, it highlighted an important issue: the number of microwave satellites has increased substantially since 2001. More satellites passing overhead means more frequent observations, increasing the chance of detecting short-lived rainfall events that earlier satellite constellations may have missed.
A real signal, but a smaller trend
To quantify this effect, we developed synthetic rainfall records that kept the underlying weather unchanged while allowing the satellite sampling pattern to evolve exactly as it had in reality. Any trend appearing in these synthetic records could therefore only arise from changes in the observing system rather than changes in rainfall itself. The analysis showed that increasing microwave sampling explained up to about 20% of the long-term rainfall trends across the 15 cities, with some urban centres, including Melbourne, showing areas where more than 40% of the apparent increase in rainfall frequency could be attributed to improved satellite coverage. Importantly, once this effect was removed, the urban rainfall signal remained. Cities still appeared to experience rainfall more frequently than surrounding regions, but the magnitude of the long-term trend became smaller. We also compared the Sydney results with another satellite rainfall product (CMORPH) and observations from Bureau of Meteorology rain gauges. Both supported the same overall pattern, although the limited number of gauges outside urban areas means the precise magnitude of the urban–rural difference remains difficult to determine.
Why it matters
Satellite rainfall datasets underpin research and decision-making in flood forecasting, agriculture, water resources, insurance and climate science, and in many parts of the world they provide the only long-term, consistent rainfall record. Our findings demonstrate that these datasets remain extremely valuable, but they also show the importance of understanding how they are constructed and how observing systems evolve over time. The study provides evidence that cities do influence rainfall, particularly by increasing how often rainfall occurs, while also showing that part of the apparent long-term trend can reflect improvements in satellite observations rather than changes in the atmosphere itself. More broadly, the research highlights an important finding for urban climate science: interpreting long-term environmental change requires understanding not only the physical processes, but also the instruments used to observe them.
This article was written by Shankar Sharma, a PhD candidate at the UNSW Climate Change Research Centre. The research was conducted with Professor Jason Evans and Professor Andy Pitman (UNSW Climate Change Research Centre) and Associate Professor Ali Behrangi (University of Arizona).