A parameter allocation approach for flow simulation using the WetSpa‐Python model

Abstract

Hydrologic models are simplified representations of natural hydrologic systems. Since these models rely on assumptions and simplifications to capture some aspects of hydrological processes, calibration of parameters is unavoidable. However, utilizing the philosophy of a recent modelling framework proposed by Bahremand (2016), we show how calibration of most model parameters can be avoided by allocating or presetting these parameters utilizing knowledge gained from sensitivity analyses, field observations and a priori specifications as a part of a parameter allocation procedure. This paper details the simulation of daily river flow of the Shemshak‐Roudak watershed performed using the Python version of the WetSpa model. The WetSpa‐Python model is a distributed model of hydrological processes applied at the watershed scale. The model was applied to the Shemshak‐Roudak watershed of Iran with parameter allocation. Model calibration involved only two parameters. Straightforward methods were proposed for allocating model parameters, including three baseflow‐related parameters and the determination of maximum active groundwater storage using a mass curve technique. Also, the Budyko curve was used to constrain a correction factor for potential evapotranspiration. The WetSpa‐Python model was extended to include the influence of snowmelt. A failure to include snow in the hydrological processes of the WetSpa‐Python model creates a significant discrepancy between the observed and simulated hydrographs during the spring. The results of daily simulations for 12 years (2002–2014) are in good agreement with observations of discharge (Kling‐Gupta Efficiency = 0.84). These results demonstrate that it is feasible to simulate hydrographs with limited calibration given a knowledge of hydrological processes and an understanding of relationships between catchment characteristics and model parameters.

Parameter allocation approach for runoff simulation in an arid catchment using the KINEROS2 hydrological model

“…A "parameter regionalization" approach was used to estimate model parameters in a spatial fashion using kriging and regression; this approach was deemed to be useful for the prediction of discharge. Bahremand et al (2021) presented an example of how the WetSpa-Python (Rezazadeh Helmi et al, 2019;Wang et al, 1996) model parameters can be allocated as per the conceptual framework introduced by Bahremand (2016). As per the conceptual framework, an initial sensitivity analysis is conducted, model parameters are allocated, and only a few parameters are adjusted for calibration.…”

“…Viviroli et al (2009) used parameter regionalization with kriging and regression methods and concluded that this approach was deemed to be useful for the prediction of discharge. Bahremand et al (2021) used parameter allocation for WetSpa-Python model with an initial sensitivity analysis, allocation of the model parameters, and adjusting only a few parameters for calibration, and Zarghi et al (2021) applied this method using the MIKESHE model. The approach used in this research is more similar to the method used by Bahremand et al (2021) and Zarghi et al (2021).…”

“…Bahremand et al (2021) used parameter allocation for WetSpa-Python model with an initial sensitivity analysis, allocation of the model parameters, and adjusting only a few parameters for calibration, and Zarghi et al (2021) applied this method using the MIKESHE model. The approach used in this research is more similar to the method used by Bahremand et al (2021) and Zarghi et al (2021). Goodrich et al (2012) identified conditions for adequate parameter calibration using the KINEROS2 model.…”

“…Bahremand et al (2021) presented an example of how the WetSpa‐Python (Rezazadeh Helmi et al, 2019; Wang et al, 1996) model parameters can be allocated as per the conceptual framework introduced by Bahremand (2016). As per the conceptual framework, an initial sensitivity analysis is conducted, model parameters are allocated, and only a few parameters are adjusted for calibration.…”

“…As an example of parameter determination, the mass‐curve technique for dam reservoir sizing via the Rippl Method (Mass Diagram Analysis) was applied to obtain the maximum active groundwater storage in a watershed. Bahremand et al (2021) explained how numerical methods of calibration involving optimization could be avoided by adjustment of only two parameters over a limited range of numerical values. A similar example of parameter allocation was reported by Zarghi et al (2021) for the MIKESHE model.…”