Water loss in plants: updating (checking) the models with load cells
As the UK edges ever closer to a drought in July 2026, farmers are increasingly concerned about water supplies for their crops. However, watering them is just half the story. Plants require water for transpiration, much of which is lost through evaporation. Load cells have been used for decades to try and work out just how much water evaporates from plants, and to work out water management strategies for both fields and greenhouse growing.
Transpiration and water loss
Water evaporates from plants during transpiration, a process where small pores on their leaves open to pull up nutrients and water from the roots, and allows for the exchange of carbon dioxide in the air.
The interesting part of transpiration is that plants can regulate it dynamically, so they can protect themselves against dehydration and cellular level damage. According to a paper in “Plant, Cell and Environment”:
“The dynamic ability of plants to regulate their transpiration rates plays a vital role in facilitating the exchange of carbon dioxide and water, thereby improving water use efficiency and optimising growth. Various factors, both biological, such as plant size, and environmental such as solar radiation, temperature, humidity, soil water supply, carbon dioxide levels levels, and wind speed can impact the transpiration rates of different plants.” (1)
The data gathered about this dynamic process informs machine learning models, which in turn can predict transpiration rates in greenhouses and other controlled environments. This vital piece of information can then be used to develop water management strategies that provide the ideal balance between plant growth and water conservation.
Load cells lysimeters
‘An array of load cell lysimeters was utilised to continuous plant weight measurement and the derivation of both transpiration-induced water loss and daily plant mass accumulation.”
The experiments were carried out using mainly tomato and cereal crops grown under glass from June 2018 to March 2024.
As the paper points out:
‘Load-cells lysimeters are widely regarded as the gold standard for measuring whole-plant transpiration, as they enable direct quantification of evapotranspiration (ET) or transpiration flux.” (1)
To achieve the best results from their load cell lysimeters, the team also used:
- Load cell transducers which were temperature-compensated,
- Short cables to an individual A/D controller to reduce electrical interference
- Thermal insulation covers for each load cell to prevent overheating due to strong sunlight
- Compressed foam cushions to reduce vibration-induced noise.
And the results showed that…
The biggest factor affecting transpiration rates was that “Plant biomass plays the most significant role in predicting daily transpiration … The correlation between plant biomass and transpiration is well known, as larger plants have more leaves (and thus more stomata) leading to more water loss.”
What the data has provided is a long-term dataset that is much more comprehensive and reliable for the creation of future machine learning models, thanks to load cells.
“ML models trained on data derived from lysimeters can serve as reliable benchmarks (‘ground truth’) for validating or calibrating alternative indirect estimation methods “ (1)
Looking for load cells for horticultural research or commercial growing?
Contact us to discuss your requirement. We can help create bespoke load cell systems to monitor, test and record data from a wide variety of locations including glass houses, barns, silos, and in wet dirty, dry and dusty environments too.
Contact us to discuss your requirements.
About the author
Chris Beasley is Director of Richmond Industries Ltd, designers and manufacturers of advanced load cells and force transducers for industry, governments and universities all over the world.
REFERENCES:
(1) Friedman, S., Averbuch, N., Nevo, T., & Moshelion, M. (2026). Integrating load-cell lysimetry and machine learning for prediction of daily plant transpiration. Plant, Cell & Environment, 49(1), 410–429. https://doi.org/10.1111/pce.70222
