Twenty-one Sacramento Valley almond blocks got drier every single week from mid-June to the second week of August. Then most of them turned around. Six did not.
Harvest is the one time of year an almond grower has a good reason to let the orchard dry down. The floor has to be firm enough to shake onto, and the nuts have to dry. Water goes off, and for a few weeks the tree is on its own.
What happens after the nuts come off is a different question, and the research on it is unusually one-sided. We read 153 sensors across 24 almond blocks in the Sacramento Valley. Here is what they measured through this year’s harvest, and where the blocks stand now.
Eight weeks down, then the turn
Of those 24 blocks, 21 have an unbroken weekly record going back to mid-June. Those 21 are the panel below — the same blocks in every row, so a change in the number is a change in the trees and not a change in which trees are counted.
Each block is a block average, not a single tree. A valve is all-or-nothing, so the block is the unit a grower can actually act on.
| Week starting | Block average (Δbar) | |
|---|---|---|
| June 15 | −4.4 | |
| June 22 | −4.8 | |
| June 29 | −5.4 | |
| July 6 | −5.7 | |
| July 13 | −6.6 | |
| July 20 | −6.7 | |
| July 27 | −7.6 | |
| August 3 | −8.1 | |
| August 10 | −9.3 | deepest week of the season |
| August 17 | −7.5 | |
| August 24 | −6.6 |
Eight consecutive weeks of decline, a bottom in the second week of August, and two weeks of recovery since. The shape is what you would expect from a harvest dry-down followed by water going back on.
What a Δbar is, briefly
The numbers above are not raw stem water potential. They are the distance between what the tree actually measured at midday and what a fully watered tree would have measured under that block’s own weather that same day. Zero means the tree is keeping up with the day. −9 means it is nine bars behind.
The reason for reporting it this way is that a raw number moves when the weather moves. A tree at −12 bars on a 105°F day and a tree at −12 bars on a 78°F day are not in the same condition. Dividing the weather out is what lets you compare a block to itself in July and in September, or to the block across the road.
The orchard average hides the orchard
The panel average in the last full week of August was −6.6 Δbar. On its own that number is close to useless, because the blocks are nowhere near each other.
The wettest tenth of the panel sat at −1.4 Δbar — all but caught up. The driest tenth sat at −14.5. That is a spread of about 13 bars between blocks in the same valley, in the same week, all of them almonds.
The recovery was just as uneven. Measured from the August 10 bottom, 15 of the 21 blocks are wetter than they were. Six are not. The median block gained about 1.8 bars back.
We report the count rather than the average recovery on purpose: the mean gain across blocks is 2.7 bars, well above the median, which means a handful of large recoveries are pulling it. Fifteen out of twenty-one is the honest version.
Why the post-harvest window is the one to watch
Almonds tolerate deliberate deficit better than most crops, but not equally at every stage, and the published work draws a sharp line at harvest.
On the pre-harvest side the evidence is reassuring. A five-year northern Sacramento Valley trial by Stewart and colleagues (2011) held midday stem water potential between −14 and −18 bars through hull split and saved roughly five inches of water a year with no significant yield loss, at a cost of slightly lighter kernels. UC ANR Publication 8515 reports the same band from kernel fill through hull split costing two to three percent of kernel weight while saving ten to fifteen percent of the season’s water.
After harvest the picture changes. Goldhamer and Viveros (2000) compared pre-harvest irrigation cutoffs of 8 to 57 days against post-harvest water deprivation in a mature Nonpareil block. The pre-harvest cutoffs were survivable. Withholding water after harvest was not: bloom density in the following season fell by as much as 52 percent and fruit set by as much as 94 percent. Their advice to growers working with limited water was to bias the water they have toward the post-harvest period.
One unit note, because it matters and it is easy to get wrong. The −14 to −18 bar figures in those trials are absolute midday stem water potential. The Δbar numbers in our table are distances from a fully watered baseline. They are different measurements and they should not be compared directly to each other.
What we are watching in September
The interesting blocks right now are the six that have not turned. Some of that is scheduling — a block that came off later has had less time. Some of it is that the water going back on is not reaching the trees the way the grower thinks it is, which is the kind of thing that shows up in the plant weeks before it shows up in a yield report the following August.
The trees still have leaves, and while they do they are building next season’s spurs and buds. That work is happening now, in September, on this year’s water. It is the part of the season with the least visible feedback and, on the evidence above, some of the highest stakes.
A block average tells you which of your blocks are in that position. The valley average does not.
FloraPulse measures stem water potential continuously from inside the trunk, block by block. See what California orchards are reading this week.
About the data: All readings come from the FloraPulse sensor network —
continuous microtensiometer measurements from orchards across California.
Figures here cover 21 Sacramento Valley almond blocks with a complete weekly record from June 15 to August 29, 2026.
Data is anonymized and aggregated by region.