Identification of linkages between urban heat Island magnitude and urban rainfall modification by use of causal discovery algorithms

This study is the first, purely data-driven technique to assess the classical urban climatological problem of the relation between urban heat island (UHI) and urban rainfall modification. Urban areas affect regional climate. This knowledge is based on studies involving observations or process-scale models. These studies, while important for understanding the dynamic processes, come at a cost (time, resources). In this paper, we explore the potential of using statistical algorithms for conducing such urban climate studies. Taking an example of the UHI and rainfall dataset, we test six different causal discovery algorithms in their search for a relation between UHI and urban rainfall. This study aims to highlight that a causal algorithm can provide us computationally efficient results that can be useful for urban rainfall studies. In terms of the causal techniques, the constraint versus score-based models provided different causal links, and the score-based models were found to be more realistic (IMaGES). The results provide a promise for the simple, data-centric models to be used as a screening mechanism for identifying cities and cases to study the influence of urban rainfall modification; and, can prove to be useful for further in-depth analysis using computationally intense numerical models, or field studies.