Selection of gridded precipitation data for Iraq using compromise programming
Appropriate selection of gridded precipitation data is very important for the region where long-term precipitation observations are not available. An approach based on compromise programing (CP) is proposed to select the gridded precipitation data for Iraq, where precipitation gauges are very sparse...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Published: |
Elsevier Ltd.
2019
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/87430/ http://dx.doi.org/10.1016/j.measurement.2018.09.047 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Universiti Teknologi Malaysia |
| Summary: | Appropriate selection of gridded precipitation data is very important for the region where long-term precipitation observations are not available. An approach based on compromise programing (CP) is proposed to select the gridded precipitation data for Iraq, where precipitation gauges are very sparse. The performance of seven widely used gauge-based, reanalysis and remote sensing-based gridded monthly precipitation data were assessed using observed precipitation records of 41 stations located within and surrounding Iraq. Obtained results showed the superiority of different datasets in term of different statistical metrics, namely percentage of bias, normalized root-mean-square error, Nash-Sutcliff efficiency, modified index of agreement, volumetric efficiency, and skill score. However, combining all the indices using CP, the gauge-based Global Precipitation Climatology Centre (GPCC) precipitation data was found best in term of replicating observed precipitation at 21 out of 41 stations. Validation of results using different approaches revealed that the GPCC precipitation was able to replicate annual and seasonal mean and variability and probability distribution of observed precipitation. The study concludes that CP can be used to select appropriate gridded precipitation data by avoiding confusion arise from contradictory results obtained using different statistical metrics. |
|---|