Derin öğrenme tabanlı yağış tahmini literatürünün tematik yapısı, evrimi ve yükselen yönelimleri

Aydin O.

I. Ulusal Samsun Coğrafya Sempozyumu, Samsun, Türkiye, 2 - 04 Nisan 2026, cilt.1, ss.22, (Özet Bildiri)

  • Yayın Türü: Bildiri / Özet Bildiri
  • Cilt numarası: 1
  • Basıldığı Şehir: Samsun
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.22
  • Ankara Üniversitesi Adresli: Evet

Özet

This study aims to bibliometrically reveal the high-impact core of the deep learning–based precipitation forecasting literature over the 2020–2025

period. Publication and research trends were analyzed based on the top 100 most-cited papers selected from Web of Science data. The analyses

were conducted in the RStudio environment using the bibliometrix package and the biblioShiny interface. Within this framework, thematic maps,

thematic evolution trees, word clouds, and trend topic visualizations were generated, enabling a joint assessment of the field’s thematic structure,

conceptual patterns, and temporal orientations. The findings indicate that the methodological core increasingly concentrates around “deep

learning” throughout the period. Results from the thematic map show that the “precipitation nowcasting” and “radar” axis constitutes a motor

theme of the field with high centrality and high density. The same map also indicates that the themes “bias correction” and “generative” are

positioned close to “deep learning”, suggesting a strengthening methodological diversification along this axis. In contrast, the theme “statistical

downscaling” is separated within a more limited connectivity space. The thematic evolution tree demonstrates the continuity of the terms

“precipitation”, “prediction”, and “model”, while “lstm” and “rainfall prediction” are represented with lower visibility. Overall, this study

contributes to identifying future research agendas by systematically making visible the dominant themes and emerging methodological

orientations in deep learning based precipitation forecasting.