Cities are increasingly influencing rainfall patterns, but not in the way scientists initially hypothesized. Despite the common belief that urbanization intensifies storms, our research reveals a different story. We analyzed satellite data from 15 major cities, including Sydney and Melbourne, and discovered that the primary urban effect is not stronger rainfall but more frequent rain events. This finding challenges the conventional understanding and highlights the need for a more nuanced approach to studying urban rainfall.
The study, published in Environmental Research Letters, utilized NASA's Integrated Multi-Satellite Retrievals for GPM (IMERG) data, which provides high-resolution rainfall estimates globally. By examining 15 cities across diverse climates and geographic settings, we found that urban areas experienced more frequent rain events compared to nearby rural regions. However, individual storms over cities often released less water than those in surrounding areas.
A key insight emerged when we separated the data by observation type. Microwave observations, which directly detect raindrops and ice inside clouds, consistently showed the urban signal. In contrast, infrared sensors, which estimate rainfall from cloud top temperatures, failed to detect light, shallow, or warm rain, leading to no discernible urban pattern. This discrepancy highlights the importance of considering the limitations of different satellite sensors.
One critical factor is the changing frequency of microwave satellite observations over time. As new satellites are launched and older ones retired, the sampling frequency has doubled between 2001 and 2023. This increase in sampling frequency means that more rain events are detected, potentially inflating the observed trend. Our analysis revealed that changes in microwave sampling explained up to 20% of the long-term rainfall trends across the cities studied.
To address this issue, we compared microwave and non-microwave observations with long-term averages, effectively separating the impact of changing satellite sampling from actual weather changes. The results showed that the urban signal remained even after accounting for sampling effects, but the long-term trend was reduced. This suggests that cities indeed experience more frequent rain, but not as much as previously thought.
Furthermore, we compared IMERG with another satellite product, CMORPH, and ground-based rain gauges. CMORPH exhibited a similar urban pattern, but the independence of these products is limited due to overlapping microwave observations. The rain gauges, while providing a more independent check, have insufficient stations outside the urban core, making it challenging to confirm the true magnitude of the urban rainfall trend.
The implications of our findings are far-reaching. Satellite rainfall data is now integral to climate science, flood risk assessment, agriculture, insurance, and water planning. However, our research serves as a cautionary tale, emphasizing that part of the observed urban rainfall trend may be attributed to changing observing systems rather than genuine change. As we continue to rely on satellite data, it is crucial to account for these factors to ensure accurate measurements and informed decision-making.
In conclusion, our study challenges the conventional understanding of urban rainfall. While the urban heat, rough surfaces, and aerosols do play a role, the primary effect is more frequent rain events. As we move forward, it is essential to refine our measurement techniques and consider the complexities of urban environments to better understand and manage rainfall in our cities.
