Impact of Urbanization on Precipitation from Meta-analysis and Causal Discovery

Jie Liu, Purdue University

Abstract

Recent decades have witnessed rapid urbanization across the globe. The temperature impacts of urbanization, referred to as urban heat island, in which the cities are typically warmer than the surrounding nonurban region are well known. A growing number of studies have recently also reported an interesting dynamical feedback on the regional weather and climate asserting that cities can modify thunderstorms and affect the rainfall in and around urban areas. However, unlike the temperature effects, the impact on rainfall are still unclear. A typical feature emerging from different studies is that there is some notable modification downwind of the cities - yet how much is this modification is unclear, with some studies even suggesting that there is no effect. One way to address this apparent inconsistency in the findings is by conducting a systematic meta-analysis that combines the results of urban impacts on precipitation change from prior published studies. After reviewing over 489, 49 unique papers were identified that had the quantitative assessment that can be used in an objective manner for the meta-analysis. Results from meta-analysis lead to the conclusion that urbanization indeed modifies rainfall, such that on average it is enhanced by 18% downwind of the city, 17% over the city, 2% on the left while 4% on the right with respect to the prevailing storm track. The distance over this occurs is approximately 20-50km. Results were further analyzed for summer vs winter, day vs night, case studies vs climatological studies, and observation vs modeling studies. In summer, city’s downwind locale experiences increase in precipitation as well as lower variance comparing to that in winter. In the day time,there is a significant increase in the rainfall over city about 30% and 10% increase in the downwind, however, in the night time, it shows a different story with 10% decreased rainfall in urban and 25% percent rainfall increase in downwind. The observational studies are more aggressive than models. Observational analysis show an increase on rainfall in both downwind and upwind (by about 30% and 20% approximately), and model analysis show a smaller increase, 15% rainfall increase downwind and about 5% upwind. The case studies indicates urban precipitation increase by about 20% in downwind, and about 15% increase in left side. The climatology analysis while also indicating downwind increase by about 15% as well as an increase over urban area by 25%, however, with 6% precipitation deduction on the left side. Results also highlight the need for standardizing the manner in which the findings are presented for urban rainfall modified studies that can aid a broader generalization of results. Building on these results, the second part of this thesis research explored data-driven graphical causal models to identify the relation between urban heat island (UHI) intensity and rainfall change from gridded climatological datasets. Eight different graphical causal discovery models were tested and results show a proof of concept for their use in understanding the urbanization impacts on rainfall changes. This models can be used in future studies without resorting to complex dynamic climate/weather models. Study finding highlights that urbanization has a significant, detectable and causal impact on precipitation changes around cities.

Degree

M.S.

Advisors

Zhang, Purdue University.

Subject Area

Meteorology|Atmospheric sciences

COinS