Irrigation by Protocol: When Vineyards Delegate Decisions to Networks
When vineyards can speak through data, the question isn’t whether to listen but whether we still know how to read what the vines themselves are saying
Applications to Protocols for the Long Now, our participatory cross-disciplinary labs organised in collaboration with the Long Now Foundation, are closing on June 5. Apply now at Long Now Labs.
There’s an old Eastern European saying that a home and a vineyard don’t need a master, but a steward. There’s some truth to it, because houses and vineyards don’t operate according to simple ownership logic, but according to a logic of constant care. They teach that ownership isn’t power, but an obligation. If you don’t tend to your home, it will quickly begin to deteriorate until eventually it comes crashing down on your head.
The same applies to vineyards. Vineyards demand constant care, such as pruning, tying, protection and understanding the soil, the weather, and the timing to water the vines. The last is particularly important because vines don’t show their thirst clearly. Wilting is a reliable sign of thirst, but by then, the message has arrived too late. The first signals of thirst in a vineyard are much more subtle.
A thirsty vine might change the angle of its leaves or slow its growth. Stress, caused by a lack of water, appears as a pattern distributed across soil, root depth, grape variety, and a plethora of other things. Truly, to read a vineyard is to read a slow, uneven, living surface. All the signs of thirst are there, but they don’t appear in a way that’s easily legible. That’s why, more than anything else, vineyards need experienced growers.
An experienced grower usually walks the rows, looks at the leaves, touches the soil, and looks at the weather. The decision to water a vineyard is based on the timing, risk, terrain, plant stress, water availability, and crop quality. It’s not based on a few simple, quantifiable measurements, but on many subtle signals passed through the filter of the grower’s experience and their ability to read the vineyard, a type of literacy that has passed through generations for most of our agricultural history.
Irrigation is one of the oldest agricultural protocols. As seen in the vineyard example, it’s not just taking water from a source and dumping it where it is needed. No, water has to be released under certain conditions, following specific schedules, pipes, pumps, valves, habits, and rules of thumb. This means that vineyards are protocolized environments, and have been long before any sensor-driven automation was in charge of irrigation.
So, what happens when the stewards of vineyards delegate watering decisions to sensors, network architectures, and low-power, long-range communication protocols, such as LoRaWAN?
The Old Water Protocol
We’ve established that before the introduction of sensor-fed dashboards, viticulturists (people who cultivate grapevines) relied heavily on their experience in reading environmental signals. However, ancient irrigation protocols weren’t, and still aren’t, primitive. Quite the opposite; they’re actually highly sophisticated.
But, being mostly based on experience, much of these protocols can be very difficult to formalize, because you can’t put a numerical value on intuitions. Most growers know their vineyards and realize that they’re complex, living ecosystems. Most vineyards have that one terrace that dries faster than the rest, or that looks absolutely horrible in the afternoon, but quite vibrant in the morning. Growers also know that weather forecasts don’t always hold true.
Gaetano “Guy” Virone, founder of Environmental Designers Irrigation, makes a similar point from the perspective of irrigation system design and field performance:
The best operators use threshold-based protocols as a decision framework, not a blind rulebook. The strongest results come when smart controls are paired with regular audits, monthly checks, and field verification, because intuition without data can overwater, but data without field context can miss what the plants are telling you.
The main difference between modern irrigation protocols and those based on viticultural experience is that the latter are often embodied and local. This is where LoRaWAN comes into play, as a means of rendering informal, intuitive protocols formal and measurable.
When Vineyards Speak
For those who aren’t in the know, LoRaWAN (Long Range Wide Area Network) is a networking protocol designed to connect battery-operated devices to the internet or local networks over distances greater than your typical home WiFi. It allows small low-power devices, such as sensors, to send small data packets (readings) to an application that can store and process said data, and display it in a way that is easily legible to humans.
The fact that it’s low-powered and able to cover a wide area makes LoRaWAN perfect for monitoring the environment, plants, and soil in vineyards, which might occupy 50 or even 100 hectares (about the area of 140 soccer fields). A team of researchers installed environmental, plant, and soil sensors in a vineyard located at Quinta dos Aciprestes in Douro, Portugal, and connected them through LoRaWan for data transmission.
The main goal of the project was to solve the practical problem of monitoring vine water in remote or difficult-to-access areas where WiFi, Bluetooth, and cellular coverage might be limited. And it worked. The system provided a superior range at a fraction of the cost, which has allowed growers to manage their vineyards and water resources more efficiently.
It has also introduced a cultural shift, because the vineyard isn’t simply being monitored by the grower post-implementation. Through sensor readings, the vineyard itself becomes an active participant in the decision loop that controls its irrigation, because the conditions in the field become data points, which then turn into dashboard patterns that could suggest, or even trigger, action.
Adequacy as a Superpower
Compared to Bluetooth, WiFi, and cellular, the LoRaWAN communication protocol is actually quite limited in terms of data size and two-way communications. However, those limitations aren’t incidental. They’re by design, and that’s what makes LoRaWAN great for vineyards. Sensors that operate using LoRa (Long Range) can send small data packets over several kilometers, while maintaining high energy efficiency.
Many of these sensors and probes are usually powered by an integrated battery, which may or may not be recharged using solar panels. This eliminates the need for excessive wiring while keeping the communications working. Additionally, many agricultural signals are small, slow, periodic, and spatially distributed. For example, a soil moisture measurement sensor doesn’t need a millisecond response time and water status doesn’t need to be live-streamed.
Agriculture is a gradual, dusty, uneven, and weather-beaten context in which sensors have to survive the elements and battery constraints. LoRaWAN is perfect for this; it’s not a super-sophisticated technology, and it doesn’t need to be.
Systems that rely on sparse communication have to develop a theory of importance in terms of which metrics should be measured, how often, at what depth, and what tolerances. What’s the threshold for alerts, and at what point does an alert demand action? These are all important questions.
Cultural Shift
Protocolized environments have a tendency to induce subtle shifts in culture and responsibility. Before the implementation of autonomous monitoring, growers had to have a good reason to irrigate a vineyard, but after the implementation of data-driven dashboards, it usually becomes difficult to find a reason not to irrigate. This is because default values and thresholds have weight.
Following implementation, growers initially remain clearly in charge, with the dashboard in an advisory role. The recommendation based on the data remains just that, a recommendation from a system that’s only supposed to support decision-making.
However, a dashboard checked every morning becomes part of the work rhythm, and thresholds that usually work become hard to ignore. A recommendation that saves time becomes trusted, used, and proven to work. This leads to the burden of explanation becoming reversed.
Dario Ferrai, co-founder of OpenClawVPS, describes this as a broader problem in sensor-driven systems:
In sensor-driven systems, the tension often appears when data signals authority even though operational reality is challenging that data. Conditions can change outside the parameters that sensors are able to sample, and that is when human judgment becomes essential rather than optional.
Once a system produces a legible recommendation, ignoring such a recommendation in favor of knowledge and experience becomes an accountable act. A grower who follows the dashboard is complying with the evidentiary protocol, but one who overrules isn’t, despite the fact that overruling might lead to better outcomes and a higher-quality product.
In these cases, experience becomes an override layer, and despite potentially better outcomes, not complying with recommendations creates conditions for later accountability. This is particularly true in areas where water is scarce, expensive, regulated, or tied to sustainability quotas. In such systems, the question of why a recommendation was followed or ignored becomes a managerial, financial, environmental, or even a legal matter.
But even then, a signal that the moisture level has dropped under a certain threshold doesn’t mean that the field is thirsty. It simply indicates that irrigation may be warranted, provided the surrounding field conditions support that decision.
It’s a capital mistake to initiate irrigation based on sensor data alone. A grower still has to compare sensor data against localized weather conditions, sensor placement, and depth, before making a decision. Doing otherwise and triggering an irrigation based on sensor data alone could, and often does, lead to sub-optimal outcomes.
Virone puts the same warning more practically:
Sensor data should start the conversation, but it should not end it. When a grower’s observation conflicts with a controller threshold, the first question should not be ‘who is right?’ but ‘what is missing?’ In the field, the issue may not be the sensor reading itself. It may be a leak, poor coverage, a setting error, an outdated zone, or a calibration assumption that no longer matches real conditions.
Edge Cases in the Rows
Ordinary cases where everything goes according to the established routine are never good for testing protocol.
There’s no great philosophical dilemma when the soil is dry, the vines look stressed, the forecast is clear, and the dashboard recommends irrigation. The system and the grower agree, and the water moves. In an ideal world, this works every time. However, we’re not living in an ideal world, and interesting things happen when the signs and data diverge.
In the real world, sensors may disagree, a probe might be poorly placed, one row might be shaded differently, the sensor’s battery might be out of juice, and a gateway might have missed a packet. Perhaps the weather forecast suggests waiting, while the soil threshold suggests watering immediately.
A good example of this would be sensor A saying that the soil is dry, while sensor B says that the moisture is adequate. So, what’s the protocol here? Average them out? Trust a deeper probe? Trigger a human inspection? Or simply wait for the next reading? Herein lies the hidden culture of precision agriculture. It’s not that LoRaWAN increases precision. It’s actually about deciding what kind of imprecision is acceptable.
Harrison Jordan, founder and managing lawyer of Substance Law, frames such conflict as a matter of limits rather than failure:
These are not moments of sensor failure as much as moments that reveal a grower’s understanding of sensor limitations. The best growers use precision irrigation data as a second opinion, not a final judgment. The mistake is forgetting that an algorithm is built around averages, while each vineyard is unique.
Older agricultural judgment was full of imprecision throughout history, but much of it was interpreted completely by humans, using unscientific methods. As previously discussed, these are not meaningless judgments based on imaginary things. These are compressed expressions of live pattern recognition based on years, even decades of experience.
The Dashboard Is Not the Field
Labeling humans as unreliable parts of the system is an age-old temptation in automation, because humans forget, misread, delay, over- and under-water, follow habits, and make decisions based on incomplete information. And while all of that is true, networks are unreliable in different ways. They only measure what they’re designed to at the locations in which they’re placed.
Jordan makes the same point in more human terms:
When a soil moisture sensor tells a grower to wait, but the plants show signs of stress, the experienced grower will often trust the plant. Sensors provide data for one measured element at one location, while a skilled professional is reading the whole plant, the weather, the soil, and the history of that specific block.
A good example of a common fault is when a dry vine and a dead sensor both appear as a normal reading. That’s why the best version of irrigation by protocol isn’t vineyards without farmers, but vineyards with farmers who use protocols to see more, waste less, and intervene more intelligently.
A well-designed irrigation protocol doesn’t just automate watering. It’s based on the difference between measurement and judgment, and makes uncertainties more visible. In a well-designed LoRaWAN irrigation system, this would imply allowing growers to annotate anomalies, reject suggestions, mark sensor distrust, and teach the system what they’ve known from experience.
Final Thoughts
While irrigation protocols relying on LoRaWAN might enable vineyards and fields to talk, it’s important to remember that what they’re saying often isn’t the whole truth. The point of protocolized irrigation isn’t to take in data points and present them as truth.
The point is to make some forms of agricultural knowledge more durable, visible, and actionable. But this visibility should never be mistaken for completeness, and when a field reports its thirst with greater precision, we must question the thresholds and consider the surrounding conditions as well.
 | A guest post by
|