- Tytuł:
- Comparison and combination of interpolation methods for daily precipitation in Poland: evaluation using the correlation coefficient and correspondence ratio
- Autorzy:
-
Konca-Kędzierska, Krystyna
Wibig, Joanna
Gruszczyńska, Marta - Powiązania:
- https://bibliotekanauki.pl/articles/24085172.pdf
- Data publikacji:
- 2023
- Wydawca:
- Instytut Meteorologii i Gospodarki Wodnej - Państwowy Instytut Badawczy
- Tematy:
-
precipitation
interpolation methods
daily gridded data
validation of precipitation gridded data set
observational data - Opis:
- Interpolation of precipitation data is a common practice for generating continuous, spatially-distributed fields that can be used for a range of applications, including climate modeling, water resource management, and agricultural planning. To obtain the reference field, daily observation data from the measurement network of the Institute of Meteorology and Water Management – National Research Institute was used. In this study, we compared and combined six different interpolation methods for daily precipitation in Poland, including bilinear and bicubic interpolation, inverse distance weighting, distance-weighted average, nearest neighbor remapping, and thin plate spline regression. Implementations of these methods available in the R programming language (e.g., from packages akima, gstat, fields) and the Climate Data Operators (CDO) were applied. The performance of each method was evaluated using multiple metrics, including the Pearson correlation coefficient (RO) and the correspondence ratio (CR), but there was no clear optimal method. As an interpolated resulting field, a field consisting of the best interpolations for individual days was proposed. The assessment of daily fields was based on the CR and RO parameters. Our results showed that the combined approach outperformed individual methods with higher accuracy and reliability and allowed for generating more accurate and reliable precipitation fields. On a group of selected stations (data quality and no missing data), the precipitation result fields were compared with the fields obtained in other projects-CPLFD-GDPT5 (Berezowski et al. 2016) and G2DC-PLC (Piniewski et al. 2021). The variance inflation factor (VIF) was bigger for the resulting fields (~5), while for the compared fields, it was below 3. However, for the mean absolute error (MAE), the relationship was reversed - the MAE was approximately half as low for the fields obtained in this work.
- Źródło:
-
Meteorology Hydrology and Water Management. Research and Operational Applications; 2023, 11, 2; 1--27
2299-3835
2353-5652 - Pojawia się w:
- Meteorology Hydrology and Water Management. Research and Operational Applications
- Dostawca treści:
- Biblioteka Nauki
Artykuł