Usefulness of rainfall estimated by satellite in a distributed hydrological model applied to an extratropical Mediterranean basin
Abstract
The aim of this study is to evaluate the usefulness of two satellite-based rainfall estimation products, by means of a distributed hydrological model in an extratropical Mediterranean basin, as an alternative to the estimation of precipitation in regions where conventional gauges are scarce or non-existent. The study area is the river Jucar basin, which is located east of the Iberian Peninsula. Satellite products have a daily temporal resolution and spatial resolution of 0.25 degree (PERSIANN) and 0.04 degree (PERSIANN-CCS). In general, the PERSIANN-CCS product overestimates, while PERSIANN underestimates at different basin aggregating scales. In addition, PERSIANN has a higher rate of rain detection, but also of false alarms. The rain detection is less in the Albaida subbasin (coastal area with torrential rains and likely SCM in fall) that in the Pajaroncilo subbasin (mountainous area with orographic rain). The volume error (Ev) referred to rainfall, for all levels of basin aggregation, underestimated with PERSIANN and overestimated with PERSIANN-CCS. Albaida basin (1301 km2) has better performance in terms of index Nash-Sutcliffe efficiency (E) referred to rainfall estimation with the two satellite products, possibly due to greater presence of convective rainfall that the satellite better identifies, which coincides with what reported by Ebert et al (2007). Whereas smaller basin as Pajaroncillo (861 km2) has better performance in Ev but only with the product PERSIANN-CCS. The parameter calibration of the hydrological model TETIS has allowed to improve the modelling performance. In addition, various authors performed a calibration of the hydrological model to improve performance with products from satellite-based rainfall estimation (Bitew y Gebremichael, 2011b; Bitew et al., 2011; Jiang et al., 2012; Moreno et al., 2012). Thus, in hydrological modelling, we obtained "unsatisfactory" performance with PERSIANN, whereas with PERSIANN-CCS become "successful" performance. The results are encouraging for PERSIANN-CCS and it seems that the error of overestimating rainfall volume is better corrected in hydrological modelling. Finally, we show that the higher spatial resolution satellite product has better performance. In order to improve the possibilities of practical use of satellite rain, we implemented a Bayesian model to combine information from gauges with PERSIANN-CCS rainfall at different gauge densities, in the mountainous subbasin Pajaroncillo. Results, specifically to the study area, indicate that the average value of the estimated rainfall with PERSIANN-CCS improves with densities of 100 km2/gauge or less. With regard to hydrological modelling using a Bayesian model, the performance obtained are "good" to "very good" with densities below 100 km2/gauge. As a final conclusion, we can say that the new product for satellite-based rainfall PERSIANN-CCS, in addition to increasing the spatial resolution of PERSIANN, also improve its reliability for its use in hydrological modelling, especially if combined with rain gauge data, becoming the starting point for further research.
- Publication:
-
AGU Spring Meeting Abstracts
- Pub Date:
- May 2013
- Bibcode:
- 2013AGUSM.H44A..01R
- Keywords:
-
- 1854 HYDROLOGY / Precipitation;
- 1855 HYDROLOGY / Remote sensing