Summary

Riverbend Beverages pledged to return more water to its watershed than its Oregon bottling plant uses each year — about 850 million litres. Instead of relying on engineering estimates, Terrapulse measured the effect of upstream restoration directly. Over two years, 22 stream gauges and a network of soil and groundwater sensors recorded 1.2 billion litres of replenishment: 1.4 times the plant’s annual use.

The watershed

The plant draws from an aquifer in the upper Willamette basin, fed by small tributaries that run through farmland and degraded floodplain. Decades of straightened channels and bare winter fields meant rain ran off quickly in winter and streams ran low in late summer — exactly when demand peaks.

The challenge

Riverbend’s replenishment claim rested on a desk model that converted restoration activities into estimated litres. No one had checked the estimate against flow in the river, and investors and local groups had started to ask.

Our water pledge is now backed by gauges in the river, not a spreadsheet.

— Dana Whitfield, VP Sustainability, Riverbend Beverages

What we measured

Rivers vary enormously from year to year. To separate the effect of restoration from the weather, we used a paired-catchment design: every restored tributary was matched with a similar, unrestored control stream.

• 22 stream gauges — above and below restored reaches, recording flow every 15 minutes

• 48 soil-moisture arrays — across farms adopting cover crops

• 9 groundwater wells — monitoring aquifer levels

• 18 acoustic units — recording birds, frogs and insects along the riparian corridor

A restored tributary in the upper Willamette basin.

What the data showed

The restoration — beaver dam analogues to slow the water, reconnected floodplain, and cover crops on 1,800 hectares of farmland — changed how the watershed holds water. Late-summer baseflow in restored tributaries rose 19% compared with the paired controls, and soil moisture under cover crops was 14% higher through the dry months.

Year two was wetter than average. Because the reported figure only counts the difference against control streams, a wet year raises both and doesn’t inflate the result.

What didn’t go to plan

Three gauges were buried by gravel in a February flood and had to be reinstalled, losing six weeks of data. Beaver dam analogues on one tributary washed out and were rebuilt with a sturdier design. We report both gaps and the wider uncertainty that comes with them.

The result

An independent hydrologist verified the replenishment total against the Volumetric Water Benefit Accounting (VWBA) method. Riverbend now reports its water balance season by season — dry years included.

What’s next

Riverbend is extending the programme to two more tributaries in 2026 and sharing gauge data with the local watershed council, so farmers and agencies can plan with the same numbers.