Machine Learning Based Environmental Weather Forecasting Frameworks for Urban Longevity: A Comprehensive 35 Years of Analysis Using ERA5 Data Over Pune Region, India

DOI

10.2991/978-94-6239-797-2_36How to use a DOI?

Keywords

Machine learning; Deep Learning; ERA5 data reanalysis; LSTM; Transformer Models; Weather Forecasting; aging population; environmental management; climate resilience; Pune; urban longevity

Abstract

Weather Forecasting is an intrinsically chaotic process and a critical component of disaster management, agriculture, defense and city planning and for communities to care for elderly residents. In cities like Pune in India, the population above the age of 60 years old is growing rapidly and for such age group, health and safety is the major concern. The ability to forecast the various weather parameters, extreme events, heavy rainfall even a day in advance can make a difference during the public health crisis and disaster management. In this paper the author explains a machine learning-based weather forecasting system which was implemented based on 35 years of ERA5 hourly reanalysis data [1] (1990–2024) on the Pune Metropolitan Region. Our dataset consisted of 4,602,240 observations on a 15-point spatial grid and comprised six atmospheric variables, including: 2-m temperature, dewpoint temperature, wind components, total precipitation, and surface solar radiation. They trained and compared four models, including LSTM [8], Transformer [7], CNN-LSTM [9] and XGBoost [17] against Prophet Model, a statistical baseline. We found that Pune was becoming increasingly warmer (+0.032 ℃ each year, R2=0.72), more humid, and less windy over the years, which directly aggravates the situation of older adults outside [25]. The Transformer had the lowest temperature deviation (RMSE=1.61 ℃, NSE=0.87), whereas CNN-LSTM was the most effective at monsoon rainfall (RMSE=2.18 mm). We mention how these predictions can be fed directly to heat alerts, flood early warning, and care facility management, which are useful tools that can significantly shield the aging urban population against the damage and destruction by weather [21].

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Anup Dakre
AU  - Chaya Jadhav
PY  - 2026
DA  - 2026/10/14
TI  - Machine Learning Based Environmental Weather Forecasting Frameworks for Urban Longevity: A Comprehensive 35 Years of Analysis Using ERA5 Data Over Pune Region, India
BT  - Proceedings of the 2nd International Conference on Lifespan Innovation (ICLI 2026) national Conference on Lifespan Innovation
PB  - Atlantis Press
SP  - 367
EP  - 375
SN  - 2352-5398
UR  - https://doi.org/10.2991/978-94-6239-797-2_36
DO  - 10.2991/978-94-6239-797-2_36
ID  - Dakre2026
ER  -