Performance Evaluation of CHIRPS, ERA5-Land and PERSIANN-CDR Precipitation Products over the Black Sea Region of Türkiye (1984-2024)
International Symposium on Applied Geoinformatics 2026, Ostrava, Çek Cumhuriyeti, 13 - 15 Mayıs 2026, ss.1-8, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.31490/9788024848693-14
- Basıldığı Şehir: Ostrava
- Basıldığı Ülke: Çek Cumhuriyeti
- Sayfa Sayıları: ss.1-8
- Karadeniz Teknik Üniversitesi Adresli: Evet
Özet
Precise monitoring of precipitation in the Black Sea region of Türkiye is difficult due to rugged topography and coastal dynamics. In this study, the performance of three structurally distinct precipitation products, CHIRPS (V2.0 Final), ERA5-Land, and PERSIANN-CDR, was evaluated using a network of 135 homogenized stations between 1984 and 2024. Data extraction was conducted through a point-to-pixel matching approach, and performance was tested using continuous metrics (CC, KGE, RB, RMSE) along with categorical scores (POD, FAR, CSI) focused on the 95th percentile threshold. Additionally, performance variations were examined with respect to mean precipitation intensity, distance to the coastline, and elevation. According to the performance metrics, CHIRPS demonstrates the highest statistical consistency with station-based observations, yielding a CC of 0.755, a KGE of 0.654, and a negligible bias (-0.007). ERA5-Land shows high correlation (CC: 0.71) and the best detection for extreme precipitation (POD: 0.50), but exhibits a significant positive bias (RB: 0.34) and high FAR (0.65). On the other hand, PERSIANN-CDR shows the weakest performance, with the lowest correlation (CC: 0.538) and very low detection scores (POD: 0.029). In conclusion, CHIRPS provides a more balanced representation, whereas ERA5-Land exhibits systematic overestimation, and PERSIANN-CDR shows limited capability in capturing rainfall timing and magnitude.