Analyzing Urban Heat Island in Moscow with Machine Learning

What is it about?

The paper focuses on approximating Urban Heat Island (UHI) dynamics in Moscow using machine learning models based on meteorological predictors. Two formulations of the machine learning problem are considered: one based on instant values of predictors and another accounting for delayed connections between UHI and predictors. Meteorological variables like air temperature, humidity, wind speed, and cloud cover are selected as predictors. The study uses Ridge Regression as a baseline model and compares it with advanced models like Random Forest Regression, Gradient Boosting Regression, CatBoost Regression, Support Vector Regression, and Multi-Layer Perceptron Regression. Feature importance analysis is conducted to understand weather-related controls of UHI magnitude in Moscow.

Featured Image

Why is it important?

The topic is important because understanding Urban Heat Island (UHI) dynamics in Moscow using machine learning models and meteorological predictors can help in analyzing weather-related controls of UHI magnitude. This knowledge is crucial for studying urban climate, assessing the impact of urbanization on temperature variations, and developing strategies to mitigate heat-related issues in urban areas. Understanding Urban Heat Island (UHI) dynamics is important as it helps in analyzing weather-related controls of UHI magnitude, studying urban climate, assessing the impact of urbanization on temperature variations, and developing strategies to mitigate heat-related issues in urban areas.

Perspectives