Open AccessHow significant is sub‐daily variability of rainfall for hydrological modelling of floods? A satellite based approach to sub‐daily downscaling of gauged rainfall
Author(s) -
Shrestha Pallav K.,
Shrestha Sangam,
Ninsawat Sarawut
Publication year - 2019
Publication title -
meteorological applications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.672
H-Index - 59
eISSN - 1469-8080
pISSN - 1350-4827
DOI - 10.1002/met.1762
Subject(s) - downscaling , environmental science , hydrograph , flood myth , climatology , storm , satellite , meteorology , forcing (mathematics) , rain gauge , precipitation , geography , geology , archaeology , aerospace engineering , engineering
Floods exhibit pronounced sub‐daily variations originating from the rainfall variability during storm events. Accurate flood simulations require sub‐daily gauged rainfall which may not be available in all cases. This study evaluates an approach wherein daily gauged rainfall was downscaled to 3 hr resolution using the Tropical Rainfall Measurement Mission's (TRMM) 3B42 level 3 satellite rainfall product ( R sat ). Hydrological performance of the TRMM downscaled rainfall ( R ds ) was compared with R sat and daily gauged rainfall downscaled uniformly at 3 hr intervals ( R uni ) by forcing them individually into a pre‐calibrated distributed hydrological model. Satellite based sub‐daily downscaling of daily gauged rainfall ( R ds ) was witnessed to be the best alternative for flood analyses in the absence of sub‐daily measurements. Visual and statistical evaluation showed the superiority of R ds in flood hydrograph performance (hourly median values of Nash–Sutcliffe efficiency [NSE] 0.89 and R 2 [co‐efficient of determination] 0.96) as well as in full‐length hydrograph performance (hourly overall values of NSE 0.72 and R 2 0.69). Between R uni and R sat , the choice rests on the purpose as the latter was more valuable for flood analysis while R uni was the better choice for achieving water balance and full‐length hydrograph match. The performance of R ds (good), R sat (moderate) and R uni (poor) shows that a combination of rain gauge data downscaled with sub‐daily information from satellites provides the best option for accurate hydrological modelling of floods. Interestingly, R ds induced an additional lead error of 3.5 hr (median value) in flood peak timing which is recommended as a subject for future research.