Accessing and Reading Lake Sakakawea Water Level Data
The primary source for Lake Sakakawea water level information is the USGS National Water Information System. Go to waterdata.usgs.gov and search for Garrison, ND stream gauge (06325500). That gauge sits right at the dam and has been recording since 1953. The data comes back in daily mean stages measured in feet relative to the National Geodetic Vertical Datum of 1929. The pool elevation at full conservation storage is 1,684 feet above NGVD29. When you pull that dataset, you get a CSV or Excel export with columns for date and stage. The raw numbers don't always tell the whole story though. The USGS updates are typically delayed by a day or two for reviewed data. There's also an "estimated" flag on some readings when the instrument had issues. I learned this the hard way in 2019 when I was tracking spring runoff predictions for a hydrology project. The gauge showed a sudden 3.2-foot jump overnight that wasn't a real event. It was a sensor calibration hiccup. The workaround was cross-referencing with the nearby Williston gauge (06325000) and the Corps' own daily operation reports, which both showed normal conditions that day. Always triple-check anomalies before trusting a single reading. Beyond the basic gauge data, the Army Corps of Engineers publishes monthly reservoir operation reports for the Missouri River system. Those are available through the USACE Omaha Division website and include actual pool elevations, inflows, outflows, and spillway usage. The correlation between the two datasets is usually strong but not perfect. The Corps uses slightly different measurement references and their reporting cadence differs from the USGS daily means.
Here's something most people miss when they start working with this data. Lake Sakakawea's stage-discharge relationship isn't linear the way you might expect from a simple reservoir. The lake has a very large surface area relative to its volume, which means small changes in elevation can represent enormous volume shifts. A one-foot drop at low pool can mean hundreds of thousands of acre-feet lost. Conversely, near the top of the conservation pool, the same one-foot rise represents significantly less volume. If you're trying to estimate water volume from stage data alone, you need the Corps' stage-storage curve. Using a linear approximation will throw off your calculations substantially, especially during drawdown periods like the 2021-2022 drought when the pool hit near-record lows around 1,640 feet. That's roughly 44 feet below conservation pool and represented a serious shortage situation across the entire Missouri River Basin. The historical record shows some clear patterns. The late 1990s through early 2000s were notably wet, with the lake frequently at or above conservation pool. Then came the extended drought period starting around 2012 that pushed levels down significantly. The 2019-2022 period was particularly brutal. I remember pulling the data during the winter of 2020-2021 when the lake was sitting so low that the Corps had to use supplemental water deliveries from upstream reservoirs to maintain minimum downstream flow requirements. The institutional memory on the river runs deep but the raw numbers make it immediately obvious when you're looking at a stressed system versus a healthy one. Another practical thing to note. The USGS gauge at Garrison measures the water surface elevation right at the dam structure. This doesn't perfectly represent the entire lake surface, which can have slight gradients due to wind setup and momentum effects, particularly during high inflow events. During normal conditions the difference is negligible. During major spring runs when the Missouri is pushing millions of cubic feet per second into the reservoir, the upstream end of the lake can be a foot or so higher than the gauge reading. If your application requires accuracy along the full length of the reservoir, you should supplement the Garrison gauge with other monitoring points or aerial imagery where available.
For anyone doing longer-term trend analysis, the National Drought Monitor data and the Colorado River Basin precipitation reports aren't directly relevant here but the broader climate pattern data from NOAA's Climate Prediction Center does matter. Lake Sakakawea's replenishment depends almost entirely on Rocky Mountain snowpack and northern plains precipitation. The teleconnection patterns that drive Colorado River flows are different systems, though they sometimes correlate in weird ways during El Niño and La Niña years. I've found that checking the Pacific Decadal Oscillation phase alongside the lake data adds useful context for multi-year forecasts. The data is freely downloadable. USGS offers direct CSV, XML, and web feature service formats. The Corps also provides some summary datasets but they're less granular than what the USGS publishes. If you need hourly data instead of daily means, the USGS Real-Time Data portal at nwis.waterdata.usgs.gov has that, though it comes with the caveat that real-time data is preliminary and subject to revision. I typically pull both real-time and archived datasets, reconcile them after the revision cycle completes, and work from the finalized version for any published work. It adds maybe an extra hour of processing but saves you from having to correct mistakes later when the revised data drops.
Get the Full Details
