JFRCalifornia
Keeper of San Juan Secrets
Water Year 2025 ended on September 30, and now USBR has published the final numbers to be able to compare it to other years in Lake Powell’s history. For some perspective, the average unregulated inflow to Lake Powell from 1991-2020 was 9.6 maf, which is what USBR uses as a benchmark. (From 2000-25, it's only been 8.4 maf.) Now, “unregulated” just means the inflow that would have made it to Lake Powell had none of the reservoirs existed farther upstream. Of course, users in the upper basin states divert water for their own use, so the “unregulated” amount is essentially what’s left over after those diversions (and evaporation, plus whatever soaks into the ground).
By that measure, WY2025 was far below average. The unregulated inflow to Lake Powell was about 4.7 maf, which is about 48% of the 1991-2020 average. That’s a huge drop off from even the below average year of 2024, which had an inflow of 8.0 maf. The unregulated inflow for WY2025 was the 5th lowest in the 62-year history of Lake Powell. The other four that were worse? 1977, 2002, 2018 and 2021… which means we’ve had 3 of the worst 5 years in the lake’s history since 2018. (A trend?) Not surprisingly, the lake fell a net 34 feet when you compare October 1, 2024 to exactly one year later (3578 vs. 3544). This was in spite of the fact that USBR minimized outflows to 7.48 maf, the second lowest release volume in the history of Lake Powell. Only 2022 was less (7.0 maf), and only 2014 and 2024 matched it.
So what will happen in 2026? Every October, USBR issues its forecast for the coming year, and this time, they predict an unregulated inflow of 7.6 maf, slightly better than in 2025—but still below average. And yet, they admit there’s a huge range of possible outcomes, anywhere from 4.1-14.5 maf, because just like the rest of us, they don’t have a fully functioning crystal ball. What does this translate to in terms of lake levels? USBR’s “most probable” projection predicts we end WY2026 (9-30-26) at 3538 after a summer peak close to 3547. In other words, grim. At the same time, they say the possible range of outcomes is huge—anywhere from a 9-30-26 lake level of 3492 to 3610. Something to please all the optimists and pessimists among us. But the alarming part of this is that a year ago, the “minimum probable” prediction in October 2024 for WY2025 was an unregulated inflow of 5.8 maf. Well, we ended up at 4.7 maf, far below even the “minimum probable.” If we do that again in 2026, we're below minimum power pool by early 2027, unless outflows are greatly reduced even beyond the low outflows in the current forecast. Of course, this emergency approach to propping up Powell would have substantial impacts on Lake Mead. Stating the obvious, all this underscores the imperative that the states get their collective act together as soon as possible.
At this point, you might ask: how accurate have their predictions been in the past? To figure that out, all you have to do is compare their prediction at the beginning of each water year to what actually happened. And here’s what you find:
They aren’t very accurate.
If you go back as far as USBR has published its 24-month forecasts (2010), we’ve got 16 years of data to examine. And in all those years, only twice has USBR’s pre-season prediction come anywhere close to the actual outcome, in 2014 and 2015:
Unregulated Inflow
2014
Predicted – 9.6 maf
Actual – 10.4 maf
2015
Predicted – 10.6 maf
Actual – 10.2 maf
Both years, USBR predicted an average inflow, and that’s pretty much what happened. And that’s a point worth making. In general, USBR tends to predict inflows that hover around average, which is likely an intentional (and reasonable) bias in their modeling. In all the years from 2011-2025, USBR’s predictions fell within a narrow range, mostly between 7.4 and 9.8 maf. Only in 2012 did it go out on a limb and project big—11.6 maf. And how did that work out? Well, 2012 turned out to be one of the worst years on record, with an inflow less than 5 maf. Go figure.
But USBR is if nothing else an equal opportunity poor prognosticator, missing the mark in both directions, often by a wide margin. Let’s start with the years where they guessed way too high (based on their prediction from the previous October):
2012
Predicted – 11.6 maf
Actual – 4.9 maf
2018
Predicted – 9.7 maf
Actual – 4.6 maf
2021
Predicted – 7.9 maf
Actual – 3.5 maf
2025
Predicted – 8.5 maf
Actual – 4.7 maf
In each case, the Bureau grossly overestimated inflow for the year. And in each case, the low flows were among the lowest ever recorded in the lake’s history. That is, nobody ever expects to encounter anomalously low flows, not even USBR. Unfortunately, we’ve had 6 of the lowest inflows in the history of Lake Powell since 2012. In addition to the four mentioned above, 2013, 2020 and 2022 were not far behind.
But before you despair too much, there’s several (but fewer) examples of where USBR underestimated inflow by about the same margin:
2011
Predicted – 9.6 maf
Actual – 16.7 maf
2019
Predicted – 7.6 maf
Actual – 12.9 maf
2023
Predicted – 8.1 maf
Actual – 13.4 maf
While those years were all cause for celebration, it’s worth noting that only 2011 was anywhere close to what in historic terms would be called a “big” inflow year. Still, with USBR’s tendency to predict numbers close to the 30-year average, even 2019 and 2023 looked comparatively great. Both years were also helped significantly by the fact that USBR reduced releases through the dam to historic lows, which really propped up the lake level and volume.
But back to USBR’s prognostication skills. The bottom line is this: in 6 of the past 8 years, USBR’s October prediction for the coming year was too optimistic. In recent years, they have tended to fall short, sometimes by a lot. Why? That’s for another post.
Here’s another interesting point worth noting: although there is great variation year to year in the accuracy of their projections, if you take the average of USBR’s projections over the long run, they come closer to the mark. No surprise there. Nevertheless, in the aggregate of those 15 years, they overestimated inflow by about 450,000 af. Doesn’t seem huge over that span, but it does suggest that they are slightly overoptimistic in the big picture—their estimating approach apparently just doesn’t quite capture what’s going on in the long-term world of precipitation, upper basin water use, and evaporation/seepage, especially recently.
So that leads us back to looking ahead to WY2026. What should we expect? Based on USBR’s track record, we have no idea. But we do know that USBR is planning to limit releases through Glen Canyon Dam to only 7.48 maf, the same as they did last year (and the year before). That’s potentially good news for Lake Powell (and bad for Lake Mead). And what does the unexpectedly rainy October mean for the coming year? Well, we’ll need to see the final numbers, but it looks like total inflow to Lake Powell for October 2025 will be just over 500,000 af. Unregulated inflow will likely be even higher, because a lot was captured in Navajo Reservoir during the month. This exceeds USBR prediction of 320,000 af of unregulated inflow to Powell for the month. A good start, and very helpful with regard to improving regional soil moisture. But it’s still all going to boil down to how the winter goes…
By that measure, WY2025 was far below average. The unregulated inflow to Lake Powell was about 4.7 maf, which is about 48% of the 1991-2020 average. That’s a huge drop off from even the below average year of 2024, which had an inflow of 8.0 maf. The unregulated inflow for WY2025 was the 5th lowest in the 62-year history of Lake Powell. The other four that were worse? 1977, 2002, 2018 and 2021… which means we’ve had 3 of the worst 5 years in the lake’s history since 2018. (A trend?) Not surprisingly, the lake fell a net 34 feet when you compare October 1, 2024 to exactly one year later (3578 vs. 3544). This was in spite of the fact that USBR minimized outflows to 7.48 maf, the second lowest release volume in the history of Lake Powell. Only 2022 was less (7.0 maf), and only 2014 and 2024 matched it.
So what will happen in 2026? Every October, USBR issues its forecast for the coming year, and this time, they predict an unregulated inflow of 7.6 maf, slightly better than in 2025—but still below average. And yet, they admit there’s a huge range of possible outcomes, anywhere from 4.1-14.5 maf, because just like the rest of us, they don’t have a fully functioning crystal ball. What does this translate to in terms of lake levels? USBR’s “most probable” projection predicts we end WY2026 (9-30-26) at 3538 after a summer peak close to 3547. In other words, grim. At the same time, they say the possible range of outcomes is huge—anywhere from a 9-30-26 lake level of 3492 to 3610. Something to please all the optimists and pessimists among us. But the alarming part of this is that a year ago, the “minimum probable” prediction in October 2024 for WY2025 was an unregulated inflow of 5.8 maf. Well, we ended up at 4.7 maf, far below even the “minimum probable.” If we do that again in 2026, we're below minimum power pool by early 2027, unless outflows are greatly reduced even beyond the low outflows in the current forecast. Of course, this emergency approach to propping up Powell would have substantial impacts on Lake Mead. Stating the obvious, all this underscores the imperative that the states get their collective act together as soon as possible.
At this point, you might ask: how accurate have their predictions been in the past? To figure that out, all you have to do is compare their prediction at the beginning of each water year to what actually happened. And here’s what you find:
They aren’t very accurate.
If you go back as far as USBR has published its 24-month forecasts (2010), we’ve got 16 years of data to examine. And in all those years, only twice has USBR’s pre-season prediction come anywhere close to the actual outcome, in 2014 and 2015:
Unregulated Inflow
2014
Predicted – 9.6 maf
Actual – 10.4 maf
2015
Predicted – 10.6 maf
Actual – 10.2 maf
Both years, USBR predicted an average inflow, and that’s pretty much what happened. And that’s a point worth making. In general, USBR tends to predict inflows that hover around average, which is likely an intentional (and reasonable) bias in their modeling. In all the years from 2011-2025, USBR’s predictions fell within a narrow range, mostly between 7.4 and 9.8 maf. Only in 2012 did it go out on a limb and project big—11.6 maf. And how did that work out? Well, 2012 turned out to be one of the worst years on record, with an inflow less than 5 maf. Go figure.
But USBR is if nothing else an equal opportunity poor prognosticator, missing the mark in both directions, often by a wide margin. Let’s start with the years where they guessed way too high (based on their prediction from the previous October):
2012
Predicted – 11.6 maf
Actual – 4.9 maf
2018
Predicted – 9.7 maf
Actual – 4.6 maf
2021
Predicted – 7.9 maf
Actual – 3.5 maf
2025
Predicted – 8.5 maf
Actual – 4.7 maf
In each case, the Bureau grossly overestimated inflow for the year. And in each case, the low flows were among the lowest ever recorded in the lake’s history. That is, nobody ever expects to encounter anomalously low flows, not even USBR. Unfortunately, we’ve had 6 of the lowest inflows in the history of Lake Powell since 2012. In addition to the four mentioned above, 2013, 2020 and 2022 were not far behind.
But before you despair too much, there’s several (but fewer) examples of where USBR underestimated inflow by about the same margin:
2011
Predicted – 9.6 maf
Actual – 16.7 maf
2019
Predicted – 7.6 maf
Actual – 12.9 maf
2023
Predicted – 8.1 maf
Actual – 13.4 maf
While those years were all cause for celebration, it’s worth noting that only 2011 was anywhere close to what in historic terms would be called a “big” inflow year. Still, with USBR’s tendency to predict numbers close to the 30-year average, even 2019 and 2023 looked comparatively great. Both years were also helped significantly by the fact that USBR reduced releases through the dam to historic lows, which really propped up the lake level and volume.
But back to USBR’s prognostication skills. The bottom line is this: in 6 of the past 8 years, USBR’s October prediction for the coming year was too optimistic. In recent years, they have tended to fall short, sometimes by a lot. Why? That’s for another post.
Here’s another interesting point worth noting: although there is great variation year to year in the accuracy of their projections, if you take the average of USBR’s projections over the long run, they come closer to the mark. No surprise there. Nevertheless, in the aggregate of those 15 years, they overestimated inflow by about 450,000 af. Doesn’t seem huge over that span, but it does suggest that they are slightly overoptimistic in the big picture—their estimating approach apparently just doesn’t quite capture what’s going on in the long-term world of precipitation, upper basin water use, and evaporation/seepage, especially recently.
So that leads us back to looking ahead to WY2026. What should we expect? Based on USBR’s track record, we have no idea. But we do know that USBR is planning to limit releases through Glen Canyon Dam to only 7.48 maf, the same as they did last year (and the year before). That’s potentially good news for Lake Powell (and bad for Lake Mead). And what does the unexpectedly rainy October mean for the coming year? Well, we’ll need to see the final numbers, but it looks like total inflow to Lake Powell for October 2025 will be just over 500,000 af. Unregulated inflow will likely be even higher, because a lot was captured in Navajo Reservoir during the month. This exceeds USBR prediction of 320,000 af of unregulated inflow to Powell for the month. A good start, and very helpful with regard to improving regional soil moisture. But it’s still all going to boil down to how the winter goes…
Last edited:
