Identifying the Minimum Number of Observed Rainfall Events Required for Optimal TOPMODEL Parameters in Mid-sized Equatorial Catchments

抄録

TOPMODEL, a topography-based, semi-distributed hydrological model was applied to the 84 km² Atari catchment in Eastern Uganda. The study sought to identify the minimum number of rainfall events needed to optimally calibrate 5 unknown parameters for yearly hydrological simulation. Model input data was daily averaged precipitation, river discharge and evapotranspiration for the year 2015 with the output being simulated discharge. A rainfall event was defined as consecutive days of effective rainfall - effective rainfall being a daily rainfall ≥ 5.0 mm. Parameterization was done for Sequentially Accumulated Rainfall Events (SARE), beginning with 1 event and sequentially progressing until all 54 observed rainfall events in the year were used. All SARE had similar starting dates with the end dates being variable. The ‘true’ parameters were those derived from inputting all observed rainfall events while the other instances of the parameters from partial SARE were classified as ‘non-true’. Elimination criterion of ‘non-true’ parameters was set at an error of ±30%. Parameter values varied with the change in number of rainfall events, showing their dependence on rainfall characteristics. Downslope saturated transmissivity (T_e) and maximum root zone storage deficit (SR_max) were the most and least variable from their means respectively. Also, exponential decay parameter (m) and delay time constant (t_d) needed the least and the greatest number of rainfall events to stabilise within the ±30% error bounds respectively. Therefore, the minimum number of rainfall events required to calibrate TOPMODEL and to optimise t_d in mid-sized equatorial catchments in Eastern Uganda are equivalent. Consequently, it required at least 49 rainfall events to calibrate TOPMODEL in 2015.