Impact of trends in river discharge and ocean tides on water level dynamics in the Pearl River Delta

Highlights

• Response of tidal amplitudes and phases to trends in river discharge varies spatially in a delta channel network.

• Increased ocean tides amplify tides in the seaward part of a delta and elevate the mean surface levels at upstream stations.

• Channel deepening reduces flood dominance in delta channels.

Abstract

The spectrum of tidal and subtidal water level variations in river deltas responds to river discharge variation, ocean tides, and human activities of many kinds. It remains a contemporary challenge to identify the main sources of changes in tidal dynamics in deltas, because of nonlinear interactions between tides and the river discharge in a changing setting. Understanding the main forcing factors controlling the evolution of mean water levels and the associated amplitudes and phases of tidal constituents can help to understand the causes of floods and the occurrence of low flows hindering navigation. Here, a nonstationary harmonic analysis tool (NS_TIDE) is applied to hydrological data from 14 stations in the Pearl River Delta (PRD) spanning the period 1961–2012. The water levels and main tidal constituent properties are decomposed into contributions of external forcing by river discharges and ocean tides, providing insight into the nonstationary tidal-fluvial processes. Significant temporal trends in mean water levels and tidal properties are observed in the PRD. Results indicate that there is spatial variability in the response of mean water levels and tidal properties to river discharge variation in the delta. The abrupt changes in bathymetry in the delta due to intensive sand excavation are likely responsible for the observed spatial variations in tidal response, which reduce the flood-dominant tidal asymmetry in this area.

Introduction

Deltas are the most densely populated areas of the world subject to fast economic development (Syvitski et al., 2009). Located where the river flows enter the ocean, many deltas face the combined risk of sea level rise, land subsidence, and storm surges (Hoitink et al., 2017), which may increase the risk of flood hazards worldwide (Syvitski et al., 2009). In some deltas, human interventions have a stronger influence on tidal river environments than sea level rise and land subsidence, and may overwhelm the gradual changes caused by the latter two factors (Vellinga et al., 2014). Intensive human activities during a short time period, such as sand mining (Templeton and Jay, 2013), dredging (Jewell et al., 2012), reservoir construction (Jay et al., 2011), channel deepening (Guo et al., 2014) and closure of a channel (Vellinga et al., 2014), cause significant variations of water levels and tidal properties in terms of constituent amplitudes and phases, which may exacerbate the problem of flooding and navigational safety (Hoitink and Jay, 2016).

Numerous studies on river tides in deltas are available (Jędrasik et al., 2008; Buschman et al., 2009; Arns et al., 2013; Dean and Houston, 2013; Weisse et al., 2014). Water level dynamics within deltas are subject to variations of many factors, such as wind, ice cover, channel deepening, river discharge and oceanic tidal forcing, making the tidal river environments highly nonstationary (Sassi et al., 2012; Buschman et al., 2013; Guo et al., 2016; Hoitink and Jay, 2016). Since ample data and basic theory of river tides are readily available, the response of low frequency and higher frequency (quarterdiurnal, semidiurnal, diurnal etc.) tides to changing of external forcing by river discharges and ocean tidal ranges are frequently addressed (Jay and Flinchem, 1997; Godin, 1999; Kukulka and Jay, 2003; Guo et al., 2016). In river deltas and estuaries where the river discharge is strong and exhibits evident seasonal patterns (e.g., the Yangtze Estuary, the Amazon Estuary, the Pearl River Delta), the modulation of tidal propagation by the river discharge is reflected by damped tidal amplitudes and amplified water levels in the landward direction (Jay and Flinchem, 1997), especially during the flood season due to increased impacts of friction. Extremely high water levels generally occur when a peak of river discharge and extreme tidal amplitudes are in phase (Godin, 1991). The river discharge-tide interaction will cause mean water levels to be higher during spring tide than during neap tide, explaining fortnightly oscillations in subtidal water levels (Buschman et al., 2009; Matte et al., 2013). The combined effects of river discharge and tides make the tidal signal strongly nonstationary. It is a nontrivial exercise to untangle the contributions of river discharge and tidal range variation to trends in mean water levels.

The study of nonstationary tidal signals reflecting river discharge-tide interactions remains a contemporary challenge. The classical tidal harmonic analysis is based on the assumption that the water levels are stationary, which neglects the impacts of non-tidal forcings (e.g., wind and river discharges) on water level variations (Darwin, 1893; Doodson, 1921; Flinchem and Jay, 2000). Numerical modeling can be instrumental to study nonstationary tides. For example, Gallo and Vinzon (2005) have built a 2D hydrodynamic model to investigate the effects of river discharge on tides in the Amazon, and to provide insights into the river discharge-tide interactions. However, the effect of monthly variations of river discharges are not accounted for in the work of Gallo and Vinzon, and hence the model performance is unsatisfactory in predicting the semimonthly and monthly tides in the upper part of the study region. With the application of horizontal acoustic Doppler current profilers (H-ADCPs), the time-series of both flow velocities and water levels become available (Hoitink et al., 2009; Sassi et al., 2011; Kästner et al., 2018). Buschman et al. (2009) studied the sources of subtidal water level variations with H-ADCP velocity and water level data by decomposing the friction term into contributions of river discharge, tides and river discharge-tide interactions. Since the long term (10 years or more) flow velocity data is quite difficult to collect, there have been no attempts to investigate the sources of long term changes in water levels using H-ADCP data to date.

Although the continuous wavelet transforms (CWT) can consider the effect of nontidal forcings on water levels (Jay and Flinchem, 1997; Sassi and Hoitink, 2013), limited information of nonstationary tidal-fluvial processes can be obtained from CWT because of its inability to resolve tidal constituents. To address this shortcoming, Matte et al. (2013, 2014) have developed a nonstationary tidal harmonic analysis method (NS_TIDE) and applied it in the lower Columbia River and the St. Lawrence Estuary, building the external forcing factors (river discharge and ocean tides) directly into the tidal basis functions. NS_TIDE performs much better than classical tidal harmonic analysis in predicting water levels in deltas, especially at upstream stations. It can calculate the time series of water levels and tidal properties (amplitude and phase) as functions of discharges and ocean tidal range, providing possibilities to investigate how much the water levels and tidal properties vary in time, as functions of the input time series of river discharges and tidal ranges. Although tides are typically considered to be of secondary importance to floods, relative to storm surges and pluvial flooding in deltas, they can play a decisive role. Through application of NS_TIDE, the water levels can be much better predicted, especially in the upper part of the delta that faces the highest flood risk. Using NS_TIDE, the contributions of various forcing factors to peak water levels can be established, which helps to better understand the extremes.

Due to climate change, the global mean sea level is expected to increase by 2 m in the 21st century (Nicholls and Cazenave, 2010). A response of tidal range may be expected, which depends on the delta bathymetry (Du et al., 2018). Land reclamation (Kuang et al., 2013; Zhang et al., 2015) and channel deepening for navigation (Jiang et al., 2012) change the bathymetry around the mouth of a delta and modulate the tidal forcing at seaward stations. Consequently, dramatic variations in the tidal range at the mouth of deltas may occur (Zhang et al., 2010). Due to the construction of reservoirs, the spring season flows in deltas have weakened, causing the number of low flow days to decrease (Kukulka and Jay, 2003). The reduced seasonal flow variability will limit the seasonal variations of water levels since the varying friction levels in response to river flow are the main source of these temporal oscillations. The substantial effects of sand excavation and dredging that took place in delta channels have reduced the frictional effects, making it easier for river flows to enter the delta and affect water levels in deltas. Due to a lack of long-period observation data, few studies have revealed the river tide response to varying landward and seaward boundaries.

The overall aim of this study is to evaluate the response of mean water levels and tidal amplitudes to trends in external forcing by river discharge and ocean tides. The Pearl River Delta (PRD) in China is set as the subject of our study. Hallegatte et al. (2013) provided a quantification of present and future flood losses in the 136 largest coastal cities around the globe, which indicated that the Pearl River Delta (PRD, Fig. 1) experiences the most frequent and severe events of compound flooding. The urban “agglomeration” of cities around the PRD, including Hong Kong, Macao, Guangzhou and Shenzhen, indeed make the PRD the world's largest metropolis (Van Mead, 2015). Improved predictions of water levels are essential to the welfare of millions of PRD residents. The obvious changes in upstream and downstream boundaries under intensive human interventions have motivated our study.