Sand mobility in a riffle-pool section of a regulated gravel bed river: a case study of a salmon spawning reach of the San Joaquin River, California
Abstract
The San Joaquin River extends almost 400 miles. It begins in the Sierra Nevada mountain range, flows through the low-land Central Valley, and drains into the San Francisco Bay. The San Joaquin River has a tortuous history of channel modifications, including in-channel and floodplain gravel mining, hydroelectric power operations, and water resources related diversions and dams. In 1942, Friant Dam was constructed near Fresno, California, by the U.S. Bureau of Reclamation. Friant Dam provided operators the ability to regulate flows in the San Joaquin River to provide societal benefits such as flood control, drought resilience, and recreation; however, it also altered the river's hydrologic regime, sediment conveyance, and ability to support the riverine ecosystem that existed before the dam. These changes, in part with other passage barriers downstream, resulted in the extirpation of Chinook salmon within the low-land gravel bedded reach. A lawsuit was settled in 2006 that created the San Joaquin River Restoration Program and required federal and state agencies to reintroduce a self-sustaining spring-run Chinook salmon population between Friant Dam and the Merced River confluence. That mission is plagued by significant challenges, such as the sediment conveyance barrier imposed by Friant Dam between the upper watershed and lowland reaches. Additionally, the current salmon spawning reach contains an abundance of sand that inhibits salmon egg incubation and fry emergence. The source of this sand is unknown, and it is not clear if the sand content is changing with time. These characteristics of the San Joaquin River provide ample opportunity for researching morphological and sediment transport processes.
This thesis encompasses three chapters that describe how sand (defined herein as sediment smaller than or equal to 2 mm) moves through California's San Joaquin River within a nine-mile study reach directly downstream Friant Dam. Chapter 1 introduces the problem statement and field area, states my research question, then describes methods, results, and a discussion of bedload sampling within the mainstem San Joaquin River at flows ranging between 220 and 6,900 cfs. Chapter two discusses bedload sampling in the ephemeral Cottonwood Creek, the upstream-most tributary downstream of Friant Dam, marking what is believed to be the upstream-most sediment source in the reach. The sand supplied to the mainstem San Joaquin River from a large storm in March 2023 is estimated and compared to the mainstem bedload transport rates discussed in chapter 1. This provides a basis to investigate if the sand is stored within or transported through the study reach. Chapter 3 describes in-channel sand presence and extent within the study reach and tracks the erosion of a bank at Ledger Island. Surficial sand storage volumes are estimated for fall of 2021, 2022, and 2023, which mark the baseline in a low water year, after several months of approximately bankfull flow releases (up to 1,800 cfs), and after several months of flood flow releases (up to ~10,400 cfs), respectively. Chapter 3 then ties each of the chapters together by providing a discussion and my conclusions on the bedload transport of sand through this river in three consecutive years of low flows, moderate flows, and high flows.
Bedload transport within the study reach varies spatially and with stream discharge. Wadable low flows produced negligible sand transport when measured with a Helley-Smith bedload sampler. Bankfull flows occur at approximately 1,500 cfs in this reach (2-year recurrence) and were measured for bedload transport at two sites in the reach (Ledger Island, 4.7 miles downstream of Friant Dam and Owl Hollow, 9 miles downstream of Friant Dam) when Friant Dam released two pulse flows of 1,500 cfs in February of 2022. Bedload transport rates were measured with a cataraft-based sampling platform and Tutle River - 2 bedload sampler, allowing the cataraft to collected equal-interval bedload samples laterally across the channel. Moderate flows transported trace amounts of sand in bedload, along with large amounts of organic debris. Sand transport at moderate flows is likely discontinuous throughout the reach due to low shear stress zones where the sand deposits on the bed, typically in pools.
Bedload transport rates were measured at the same sites with the same methods at high flows (about 6-to8-year-recurrence), which confirmed that a 6,000 to 7,000 cfs flow release is capable of mobilizing the size ranges of sand that we measured in storage on the bed.
Sand bedload transport rates at Ledger Island at 6,430 cfs ranged between 1.3 and 8.7 tons/day, with a mean of 5.1 tons/day and trace amounts of gravel present with a maximum grain size of 40 mm. At flows of 6,900 cfs at Owl Hollow, sand bedload transport rates ranged between 32.7 and 95.6 tons/day, with a mean of 64.4 tons/day. The largest particle found in transport was 21 mm in diameter through its intermediate axis. Bedload transport rates at these sites show that more sand is exiting the study reach than passing through the halfway point, thus suggesting increasing sediment supply downstream. Additionally, an existing HEC-RAS model predicts that bed shear stress is higher at Owl Hollow than at Ledger Island, such that it has the capacity to transport more sediment. Bedload transport sampling on ephemeral Cottonwood Creek during a storm confirmed that the creek is a source of sand to the San Joaquin River and showed that it delivers sand (and trace amounts of gravel) at the top of the study reach during infrequent flows as low as 160 cfs. We estimate that Cottonwood Creek supplied about 50 tons of sand to the San Joaquin River during a storm in March 2023, and about 450 tons throughout the study period. Average sand bedload transport rate was 20.1 tons/day with a maximum of 59 tons/day and minimum of 0.1 tons/day.
Between 2021 and 2023, a bank at Ledger Island eroded an average of 13.9 feet laterally and a volume of about 2,700 cubic yards (about 4,000 tons). Because more sediment eroded from the bank than was measured in storage in the pool below, it is evident that sand is being mobilized downstream.
The sand supplied by Cottonwood Creek, the eroding bank at Ledger Island, and other minor near-channel sediment sources was flushed through the study reach with a 37 percent decrease in sand content between 2021 and 2023. I estimate that in-channel surficial sand content was as high as 170,000 tons in August 2021, and then decreased after an extended bankfull flow event in 2022 and a 12.5-year recurrence high flow (10,400 cfs) to 107,000 tons in August 2023. The results of this thesis indicate from multiple methods that in-channel sand storage within the study reach is decreasing, despite episodic sand contributions from tributaries and near-channel sources.