File(s) not publicly available
Evaluation of a model for disaggregating point daily rainfall into hourly rainfields: The Albert catchment case study
conference contributionposted on 21.12.2017, 00:00 by Yeboah Gyasi-AgyeiYeboah Gyasi-Agyei
The Albert catchment (688 km2) is located in south east Queensland, Australia. There are no automatic weather stations (AWS) that record fine timescale rainfall located within the catchment, but there are 32 AWS located within a square region (150 km x 150 km) encompassing the catchment. Also, the catchment is under the Mt. Stapylton rainfall radar. For daily rain gauges, 11 are located within the catchment and 266 within the square region. These point datasets were used to generate hourly rainfields over the catchment at a grid resolution of 1 km2. The daily spatial rainfall model is based on Kriging interpolation with the parameters estimated using point daily rainfall data. Disaggregation of the spatial daily rainfields into hourly rainfields is achieved using the scaled hourly storm profile of the nearest AWS station. In order to test the model, the full hourly radar rainfields were used as input into a GIS rainfall-runoff model, and the generated discharges were considered as the 'TRUTH'. Then, the collocated point daily radar rainfall values were selected to run the spatial rainfall model to generate the daily rainfields. These daily rainfields were then disaggregated into hourly rainfields using the nearest collocated AWS station's radar scale hourly storm profile, which were used as input into the GIS rainfall-runoff model. Discharges at 10 sub-catchments' outlets were compared, assuming the same parameters, topographic, soil and initial conditions. Of the 3 days of different rainfall patterns over the 10 sub-catchments examined, 17% have NSE values between 0.84 and 0.90, the rest having values being greater than 0.9, indicating a perfect match of simulated radar rainfall modelled runoff to the observed radar rainfall data. With these results, the over 100 years of observed point daily rainfall and limited AWS data can, therefore, be confidently used to generate corresponding hourly rainfields for meaningful hydrological modelling.