Details
| Original language | English |
|---|---|
| Pages (from-to) | 88-101 |
| Number of pages | 14 |
| Journal | Journal of hydrology |
| Volume | 508 |
| Early online date | 26 Oct 2013 |
| Publication status | Published - 16 Jan 2014 |
Abstract
This study investigates the performance of merging radar and rain gauge data for different high temporal resolutions and rain gauge network densities.Three different geostatistical interpolation techniques: Kriging with external drift, indicator kriging with external drift and conditional merging were compared and evaluated by cross validation. Ordinary kriging was considered as the reference method without using radar data. The study area is located in Lower Saxony, Germany, and covers the measuring range of the radar station Hanover. The data used in this study comprise continuous time series from 90 rain gauges and the weather radar that is located near Hanover over the period from 2008 until 2010. Seven different temporal resolutions from 10. min to 6. h and five different rain gauge network density scenarios were investigated regarding the interpolation performance of each method. Additionally, the influence of several temporal and spatial smoothing-techniques on radar data was evaluated and the effect of radar data quality on the interpolation performance was analyzed for each method.It was observed that smoothing of the gridded radar data improves the performance in merging rain gauge and radar data significantly. Conditional merging outperformed kriging with an external drift and indicator kriging with an external drift for all combinations of station density and temporal resolution, whereas kriging with an external drift performed similarly well for low station densities and rather coarse temporal resolutions. The results of indicator kriging with an external drift almost reached those of conditional merging for very high temporal resolutions. Kriging with an external drift appeared to be more sensitive in regard to radar data quality than the other two methods. Even for 10. min temporal resolutions, conditional merging performed better than ordinary kriging without radar information. This illustrates the benefit of merging rain gauge and radar data even for very high temporal resolutions.
Keywords
- Geostatistics, Kriging, Merging, Radar, Rainfall
ASJC Scopus subject areas
- Environmental Science(all)
- Water Science and Technology
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
In: Journal of hydrology, Vol. 508, 16.01.2014, p. 88-101.
Research output: Contribution to journal › Article › Research › peer review
}
TY - JOUR
T1 - Geostatistical merging of rain gauge and radar data for high temporal resolutions and various station density scenarios
AU - Berndt, Christian
AU - Rabiei, Ehsan
AU - Haberlandt, Uwe
PY - 2014/1/16
Y1 - 2014/1/16
N2 - This study investigates the performance of merging radar and rain gauge data for different high temporal resolutions and rain gauge network densities.Three different geostatistical interpolation techniques: Kriging with external drift, indicator kriging with external drift and conditional merging were compared and evaluated by cross validation. Ordinary kriging was considered as the reference method without using radar data. The study area is located in Lower Saxony, Germany, and covers the measuring range of the radar station Hanover. The data used in this study comprise continuous time series from 90 rain gauges and the weather radar that is located near Hanover over the period from 2008 until 2010. Seven different temporal resolutions from 10. min to 6. h and five different rain gauge network density scenarios were investigated regarding the interpolation performance of each method. Additionally, the influence of several temporal and spatial smoothing-techniques on radar data was evaluated and the effect of radar data quality on the interpolation performance was analyzed for each method.It was observed that smoothing of the gridded radar data improves the performance in merging rain gauge and radar data significantly. Conditional merging outperformed kriging with an external drift and indicator kriging with an external drift for all combinations of station density and temporal resolution, whereas kriging with an external drift performed similarly well for low station densities and rather coarse temporal resolutions. The results of indicator kriging with an external drift almost reached those of conditional merging for very high temporal resolutions. Kriging with an external drift appeared to be more sensitive in regard to radar data quality than the other two methods. Even for 10. min temporal resolutions, conditional merging performed better than ordinary kriging without radar information. This illustrates the benefit of merging rain gauge and radar data even for very high temporal resolutions.
AB - This study investigates the performance of merging radar and rain gauge data for different high temporal resolutions and rain gauge network densities.Three different geostatistical interpolation techniques: Kriging with external drift, indicator kriging with external drift and conditional merging were compared and evaluated by cross validation. Ordinary kriging was considered as the reference method without using radar data. The study area is located in Lower Saxony, Germany, and covers the measuring range of the radar station Hanover. The data used in this study comprise continuous time series from 90 rain gauges and the weather radar that is located near Hanover over the period from 2008 until 2010. Seven different temporal resolutions from 10. min to 6. h and five different rain gauge network density scenarios were investigated regarding the interpolation performance of each method. Additionally, the influence of several temporal and spatial smoothing-techniques on radar data was evaluated and the effect of radar data quality on the interpolation performance was analyzed for each method.It was observed that smoothing of the gridded radar data improves the performance in merging rain gauge and radar data significantly. Conditional merging outperformed kriging with an external drift and indicator kriging with an external drift for all combinations of station density and temporal resolution, whereas kriging with an external drift performed similarly well for low station densities and rather coarse temporal resolutions. The results of indicator kriging with an external drift almost reached those of conditional merging for very high temporal resolutions. Kriging with an external drift appeared to be more sensitive in regard to radar data quality than the other two methods. Even for 10. min temporal resolutions, conditional merging performed better than ordinary kriging without radar information. This illustrates the benefit of merging rain gauge and radar data even for very high temporal resolutions.
KW - Geostatistics
KW - Kriging
KW - Merging
KW - Radar
KW - Rainfall
UR - http://www.scopus.com/inward/record.url?scp=84887864697&partnerID=8YFLogxK
U2 - 10.1016/j.jhydrol.2013.10.028
DO - 10.1016/j.jhydrol.2013.10.028
M3 - Article
AN - SCOPUS:84887864697
VL - 508
SP - 88
EP - 101
JO - Journal of hydrology
JF - Journal of hydrology
SN - 0022-1694
ER -