Evaluation of the efficiency of satellite precipitation data in simulating river flow using the IHACRES Model (Case study: Toroq Dam Watershed)

Document Type : Research Paper

Authors

• Abolfazl Mosaedi ¹
• Sina Mohammadian ²
• Fereshteh Modaresi ²

¹ Department of Water Sciences and Engineering, Faculty of Agriculture, Ferdowsi University of Mashhad, Mashhad, Iran.  

² Department of Water Sciences and Engineering, Faculty of Agriculture, Ferdowsi University of Mashhad, Mashhad, Iran

Abstract

Keywords

• hydrological simulation
• IHACRES
• MERRA-2
• rainfall-runoff model
• Toroq watershed

Main Subjects

• هیدرولوژی

Evaluationof the efficiency of satellite precipitation data in simulating river flow using the IHACRES Model (Case study: Toroq Dam Watershed)

EXTENDED ABSTRACT

Introduction

In the realm of water resource management, accurate modeling and prediction of river flow rates are essential tasks, often relying on measured or estimated data, particularly precipitation data. In watersheds lacking sufficient precipitation measurement stations, the utilization of satellite data emerges as a viable alternative due to its extensive spatial and temporal coverage. This study focuses on the application of the IHACRES rainfall-runoff model to simulate river flow in the Toroq dam watershed using ground station and MERRA-2 satellite data over 9-year and 29-year periods.  The findings of this study can be useful in predicting the flow hydrograph due to predicted rainfall, especially in areas with limited data.

Methods

In this study, the IHACRES rainfall-runoff model has been used to simulate river flow in the Toroq dam watershed (with an area of 164.76 square kilometers). daily discharge amounts at the Kertian hydrometric station were used along with ground precipitation and temperature data from the evaporation station and MERRA-2 satellite data within the vicinity of the Kertian hydrometric station located in the Toroq dam watershed over a 29-year period. After ensuring the homogeneity of these data at a 95% confidence level, the performance of the IHACRES model was investigated using ground and satellite precipitation data with simulating watershed discharge in daily and monthly time steps over two time-periods of 9 years (from 23 September 2013 to 22 September 2022) and 29 years (from 1 January 1994 to 22 September 2022). Various model performance indicators, including Root Mean Square Error (RMSE), Normalized Root Mean Square Error (NRMSE), Nash-Sutcliffe Efficiency (NSE), Pearson correlation coefficient (P), Bias and Relative Bias, were employed. Additionally, a Taylor diagram was utilized to provide a comprehensive assessment of the IHACRES model's accuracy in simulating the discharge within the study area.

Results & Discussion

The results show that over the study period, the accuracy of estimated precipitation values based on the MERRA-2 satellite has gradually increased compared to measured values. In the 29-year period, the model validation results are weaker than the 9-year period, which could be due to conditions resulting from changes in land use and gradual temperature increase in the watershed. With a Nash-Sutcliffe Efficiency (NSE) coefficient of 0.8461 obtained in the validation stage of the 9-year period with daily time steps using satellite data, as well as a Pearson correlation coefficient of 0.9250, there is good correlation between the daily simulated flow with satellite data, and the daily measured flow. The IHACRES model has shown effectiveness in simulating low and moderate flows; however, it has demonstrated limited capability in simulating peak river flows. Notably, the model performed better in estimating peak flows on a daily basis over a 9-year period when utilizing MERRA-2 satellite data compared to using ground-based data. After determining the average annual rainfall over this 29-year period, years with rainfall above the period's average were considered as wet years, while years with below-average rainfall were classified as dry years. in wet years, the satellite model underestimated annual precipitation compared to actual rainfall, with the largest discrepancy between observed and satellite rainfall. Over time, the accuracy of the satellite model has improved, with the discrepancy between satellite precipitation values and observations decreasing as we approach the present. In dry years, the average simulated river flow values using MERRA-2 satellite data were closer to the observed river flow averages, whereas in wet years, ground-based data provided better results compared to satellite data. Additionally, in all scenarios, the model tended to overestimate the flow rates.

Conclusions

Although the satellite data showed almost similar results to the ground-based data over the 29-year medium-term period, overall, the model's performance in simulating flow during the 29-year medium-term period was weaker than the 9-year short-term period. In fact, the weaker performance of the IHACRES model in simulating flow over the 29-year period may be due to changes in land use, Climate changes caused by global warming, and changes in the accuracy of satellite precipitation estimates over time. Furthermore, the results showed that the IHACRES model performs better in monthly streamflow simulation compared to daily streamflow simulation, with the best performance of this model in simulating monthly streamflow for the 9-year period. The use of MERRA-2 satellite data on a daily scale for the 9-year period showed a higher Nash-Sutcliffe Efficiency coefficient compared to ground-based data for the same time period. Based on the Taylor diagram and other error evaluation parameters such as RMSE and Pearson correlation coefficient (R), which had better values compared to ground-based data for this period, it can be suggested that using estimated precipitation values based on MERRA-2 satellite data in recent years is a suitable approach for estimating river discharge in areas where station data are insufficient or statistically flawed.