- Title
- Testing a remote sensed rainfall product (PERSIANN) against ground based rainfall data for three sites in NSW
- Creator
- Senevirathne, Nalim; Willgoose, Garry
- Relation
- Hydrology and Water Resources Symposium 2012. Proceedings of the 34th Hydrology & Water Resources Symposium (Sydney, Australia 19-22 November, 2012) p. 485-492
- Relation
- http://search.informit.com.au/documentSummary;dn=903951150789261;res=IELENG
- Publisher
- Engineers Australia
- Resource Type
- conference paper
- Date
- 2012
- Description
- PERSIANN (Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks) is a satellite derived product used for rainfall estimation at a resolution of 0.25 x0.25 (27.8km x 24.1km for NSW). It has been successfully used for estimation of rainfall rates over the regions of United States. In this study, the PERSIANN rainfall product is evaluated against rain gauge measurements of areal average rainfall for the period of 2000-2011. The study area includes three sites (Merriwa, Wollongong and Louth) of different sizes, all located in South Eastern Australia (New South Wales). The Wollongong site encompasses the coastal escarpment behind the city, and tests the ability of PERSIANN to estimate rainfall for coastal eastern Australia. The Louth site is near Bourke and tests the ability to estimate rainfall for the Australian outback. The Merriwa site is 200km west of Newcastle and tests the ability to model rainfall for temperate inland Australia. In order to compare the datasets, a correlation analysis is performed at three different time scales (daily, weekly and monthly average rainfalls). This analysis also includes a sensitivity analysis that is done to understand the influence of quantity of rain gauges (i.e. how well is the ground truth rainfall estimated) and pixels (i.e. how important is remote sensing noise) on the correlation coefficient. Results of the correlation analysis show that PERSIANN rainfall product is able to estimate the rainfall recorded by rain gauges at daily level (correlation coefficient 0.30 to 0.49). The correlation of the two datasets is improved for the weekly and monthly averaged data (correlation coefficient 0.43 to 0.60 for weekly, 0.46 to 0.65 for monthly). However, the correlation between the datasets varies significantly from site to site. For instance, the correlation for Wollongong is significantly worse than for the other sites as a result of the orographic effect of the escarpment. Results of the sensitivity analysis indicate that the correlation coefficient is not affected by the size of study area and the number of rain gauges unless the study area or number of rain gauges is very small.
- Subject
- PERSIANN; rainfall data; rain gauge data; precipitation estimation from remotely sensed information
- Identifier
- http://hdl.handle.net/1959.13/1308723
- Identifier
- uon:21708
- Identifier
- ISBN:9781922107626
- Language
- eng
- Reviewed
- Hits: 1131
- Visitors: 1105
- Downloads: 0
| Thumbnail | File | Description | Size | Format |
|---|