Why does Control calculate saturation differently?
To account for the change in the slab's water capacity during an irrigation day.
This has a direct effect on the calculated water uptake and the stop time needed to achieve a the dry-down target. Therefore, a detailed calculation is essential.
How does the saturation calculation work for Control?
Source receives unfiltered slab weight data (in kg) from the climate computer. Every time the drain stops for a specific irrigation shot, the corresponding slab weight (in kgs) at that moment is recorded as a 'slab capacity point' at that time. This means that the slab could contain that amount of water at that moment.
However, these 'slab capacity points' are not constant and vary throughout the day. Some examples of why the 'slab capacity point' varies:
Early drain (or 'false drain'): At the beginning of the day, drain often starts coming out of the slab at a lower weight. This happens because the 'cone' (the water distribution) in the slab has not yet fully formed.
Sinking saturation: This is a known problem in the middle of the day. If the drain percentage becomes low, the weight of the slab visibly drops (in the graph). Also, at the end of the day, when a preference is given to shots with a lower drain percentage, the slab weight drops. In both cases, drain is still present at lower slab weights, causing the 'slab capacity point' to change throughout the day.
It is a known misconception that the slab capacity would be constant throughout the day; for example, Ridder assumes this in its saturation calculation. The reality is that the slab capacity is directly influenced by giving shots with a higher drain percentage per shot, because the water penetrates further into the slab. This is often referred to as 'the cone'. It is also a misconception that if early drain is achieved, it is impossible to increase the slab capacity during the day. Substrate suppliers, following this logic, always advise smaller shots in Phase 1 (up to drain) to prevent early drain, but we have numerous examples where the data shows that this can be easily solved with sufficient drain percentage in the first shots of Phase 2.
In the graph above (a day in June), you can see that the weight per shot shifts significantly, even though there is drain with every shot of the day. Early drain is also visible, but also that the slab weight returns to a good level with the high drain percentage in the subsequent shots.
What does >100% saturation mean?
This can mean two things:
An irrigation shot was given, but the drain has yet to come.
As explained above, the slab could hold more water at that moment than the average 'slab capacity level' of the day. This usually happens because large and/or frequent shots have been given compared to the average. With new data throughout the day, the data is recalibrated to be relative to the average 'slab capacity level'.
What is the difference with the Ridder and Priva calibration?
For the Ridder calibration:
When the drain stops, it is always reset to a new 100% point, instead of taking an average of all reference points where the drain stopped. This means that the 100% saturation point (and the kgs belonging to it) jumps throughout the day when the end of the drain is measured, instead of taking the average slab capacity.
The effect of this is that it is not possible to precisely calculate how much needs to be irrigated before drain occurs, because this slab capacity point jumps unrealistically throughout the day.
For example: if the shots toward the end of the day are smaller, there is less drain, and the 100% saturation reference point is set at a lower weight. The consequence of this would be that Control would recalculate the necessary irrigation volume for full saturation to be lower than what is actually needed.
Ridder does not handle shocks on scales, it fully relies on full reset after drain. This results that in practice the grower needs to take back manual control of shots, especially night shots on a regular basis when using undersaturation trigger. As a result, either night shots have to be planned manually or when done using undersaturation, it needs to be checked daily whether a shock on scale might have happened.
For the Priva calibration:
Priva takes the maximum weight on a day (until it is reset) as a reference, meaning you always stay below 100%. It does applies shocks on scale correction but in a bad way, for example resulting in the situation where irrigation is registered that never arrives in the slab (for example because of flushing the pipes), it will still show a saturation increase in the data while no such thing happened in reality.
The Control calibration:
Control takes the daily average and uses logic to manage the shocks on the scales. Generally, this ensures that you can still see trends in reference points (trends in 100% saturation weight throughout the day); the model also does not ignore these effects. However, if there is an unrealistic shock on the scale, we apply logic to filter it out (most importantly during the dry-down period). There is thus a balance between preserving real trends in the reference point (100% saturation levels) and removing shocks on the scales, which in the Ridder and Priva calibrations of undersaturation does not work nuanced enough to autonomously run irrigation.
Shock on Scale Anomaly Detection
The goal of this anomaly detection is to ensure that only relevant saturation adjustments are included in the calculations. It is specifically designed to have as accurate saturation data during the dry-down period (phase 3) as possible.
The model checks whether a shock on the scale is associated with irrigation.
The system uses specific upper and lower limits (thresholds) for this.
If the measured weight change does not coincide with irrigation, and falls outside these thresholds, the data is being corrected. If the shock happens during irrigation, it is a bit harder to detect, the increase or decrease in weight must be sufficiently greater than the shot size.
Data Correction:
The saturation data is then adjusted to 'continue' on the same trajectory as if the weight deviation had not occurred. This is done by correcting the weight data with the detected shock on scale. After the shock on scale, when drain has been reached with 3 shocks, the system will recalibrate both measurements to 100%. It is designed to do this only after 3 shots with drain, because with a single shot of drain it visually might look better, but it caused problems for example with early drain in the first few shots of the day.
Below you can see a shock on scale at the beginning of the graph. The shock is removed and after three shots with drain calibration continues as normal. What happens, is that the saturation graph from the moment of the shock until the 3 shots with drain, is shifted downwards to correct for the shot.
