EUVIMEDEUVIMED entdecken
Crossref

MODULAR WEB PLATFORM ARCHITECTURE FOR MACHINE LEARNING-BASED FLOOD FORECASTING IN KAZAKHSTAN REGIONS

N. K. Nadirov, N. M. Taubakabyl

Bulletin of Toraighyrov University. Physics & Mathematics series · 30.06.2026 · (2, 2026) · 64-80

DOI: 10.48081/bgqf2034 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Bulletin of Toraighyrov University. Physics & Mathematics series, online erschienen am 30.06.2026. Verfasst von N. K. Nadirov und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Floods remain one of the most damaging natural hazards affecting the regions of Kazakhstan, with the spring events of 2024 highlighting the urgent need for digital early-warning tools that deliver region-specific forecasts directly to end users. Existing solutions are typically designed for scientific communities and require dedicated software, which limits their use by municipal authorities and the general public. This situation defines the relevance of developing an integrated, modular platform that combines data ingestion, machine learning, and notification delivery within a single system. This paper presents the design and prototype implementation of a flood forecasting web platform composed of five loosely coupled components: a Flask-based server core with a REST API, a gradient boosting classifier built on LightGBM, a web interface with an interactive SVG map of Kazakhstan, a Telegram bot for personalized alerts, and an RSS module that surfaces flood-related news from open sources. The system is trained on a 25-year archive of hydrometeorological observations (2000–2024) and classifies flood risk into three levels: safe, warning, and danger. The scientific novelty consists in the formalization of an integration-oriented architecture that connects a machine learning core with two parallel delivery channels web and messenger and an external RSS content layer, adapted to the Kazakhstani context of unevenly digitized hydrological data. The results confirm that a modular design combined with lightweight machine learning provides a practical foundation for accessible regional flood early-warning systems. Keywords: flood forecasting, modular architecture, machine learning, LightGBM, web platform, early warning system.

Zitieren

APA

Nadirov, N. K., & Taubakabyl, N. M. (2026). MODULAR WEB PLATFORM ARCHITECTURE FOR MACHINE LEARNING-BASED FLOOD FORECASTING IN KAZAKHSTAN REGIONS. Bulletin of Toraighyrov University. Physics & Mathematics series, 64-80. https://doi.org/10.48081/bgqf2034

Vancouver

Nadirov NK, Taubakabyl NM. MODULAR WEB PLATFORM ARCHITECTURE FOR MACHINE LEARNING-BASED FLOOD FORECASTING IN KAZAKHSTAN REGIONS. Bulletin of Toraighyrov University. Physics & Mathematics series. 2026:64-80. doi:10.48081/bgqf2034

BibTeX
@article{Nadirov2026,
  author = {Nadirov, N. K. and Taubakabyl, N. M.},
  title = {MODULAR WEB PLATFORM ARCHITECTURE FOR MACHINE LEARNING-BASED FLOOD FORECASTING IN KAZAKHSTAN REGIONS},
  journal = {Bulletin of Toraighyrov University. Physics & Mathematics series},
  year = {2026},
  number = {2, 2026},
  pages = {64--80},
  doi = {10.48081/bgqf2034}
}
Kostenlos starten →