Understanding the Interannual Variation in the River Stage–Discharge Relationship and Its Impact on Flood Discharge Estimation

Abstract

Stage–discharge relationships or rating curves are essential for estimating river discharge when direct flow monitoring is challenging. Although various hydrometeorological factors affect streamflow, understanding the rating curve variability and its local and global drivers are crucial for gaining deeper insights into riverine processes. However, limited studies have addressed the uncertainties (such as parametric and observational uncertainties) associated with the stage–discharge relationship and further investigated the potential drivers of the uncertainties. The current study has three major objectives: (1) develop a robust process-based stage–discharge relationship that can account for interannual shifts in rating relationships, (2) estimate the influence of a robust rating curve in flood discharge measurement, and (3) understand the climate and morphological drivers that impact the interannual variation in the stage–discharge relationship. Toward this, a Bayesian rating framework based on on-site hydraulic knowledge was adapted for five gauging stations in the Mahanadi and Godavari basins of the Indian peninsula. Interannual variations in rating parameters and the differences in high and low estimations were obtained using rating curves with all gauging pooled and split yearly. Furthermore, a boosted regression forest model was employed to identify the most significant predictors of key rating parameters based on associations among rating drivers. The findings revealed substantial interannual variations in rating parameters, resulting in a high magnitude of difference in streamflow estimation, ranging from −69.1% to 94.3% when considering rating shifts. Additionally, the natural climate variability modes were identified as key predictors in developing stage–discharge relationships, explaining the significant variance in rating parameters. This study emphasizes the importance of developing holistic river monitoring approaches for accurate flow assessment and minimizing hydrological risks. Rating curves (stage–discharge relationships) are an essential part of the hydrometry process, commonly used to estimate river discharge by monitoring the river stage. Temporal instability in the rating curves can lead to substantial errors in flow estimation. This study explores the interannual variation in rating curves by developing a Bayesian rating framework based on river hydrogeomorphology. The study selected five gauging sites in two Indian river that experiences substantial sediment deposition in the riverbeds along with a seasonal variation in the discharge. We investigated how local and global drivers (such as precipitation and climate variability modes) influence the stage–discharge relationships. Our findings highlight that rating parameters experience nonstationarity due to shifts in these relationships from year to year, which can strongly impact high-flow estimates. By understanding the causes behind these variations, particularly the influence of climate drivers like global oceanic–atmospheric oscillations and local hydrometeorological factors, this research emphasizes the need to incorporate the interannual dynamics into rating curve models. Such an approach shall improve the accuracy of streamflow estimation and subsequent predictions, which is crucial for skilful flood and water resource management in a changing hydroenvironmental condition.