Field trial, 24 farms, one season

Environmental science poster

A season of battery free soil moisture sensors on 24 smallholder farms, laid out as an A0 landscape broadsheet. It is a field trial rather than a laboratory study, so the board reports the things field work actually turns on: whether the device survived, where it was weakest, and what it saved.

Create a poster with OneCraftA0 Landscape, printed at 1189 by 841 millimetres

The whole board

The poster at full size, exactly as it prints. Every number, citation and caption on it was written for this example, so the layout is being judged on real content.

BU
Battery free soil moisture sensors for smallholder irrigation: field performance over one growing season
K. Osei ¹, L. Marchetti ¹,², F. Ndlovu ²
1 Department of Agricultural Engineering, Brightwater University · 2 Regional Water Management Institute
Abstract
Soil moisture sensors are sold to smallholders and then abandoned, usually when the battery dies and no replacement is available within a day of travel. We built a passive sensor that harvests its power from the reader and holds no battery at all, and we ran 96 of them across 24 farms in three soil types for one growing season. Readings tracked fortnightly gravimetric sampling to a mean absolute error of 3.4 volumetric percent, which is inside the band that separates the irrigation decisions these farmers actually make. Ninety one of the 96 units were still readable at harvest, and none of the five losses was an electrical failure. Each farm irrigated one plot on the sensor readings and one on its usual schedule, and water applied fell 18% (95% CI 11 to 25) on the sensor scheduled plots with yield differing by 1.2% in their favour, well inside the noise for a single season.
Introduction
Irrigation on the farms in this district is scheduled by feel and by habit. Water is not metered and the cost of over watering is paid later, in pumping fuel and in nutrient washed below the root zone, so the feedback that would correct the habit never arrives. Where sensors have been introduced, the failure mode is consistent and it is not accuracy: the coin cell runs out after a season or two, the nearest replacement is a bus journey away, and the unit is left in the ground. Extension officers in this district could point us to buried sensors on nine of the first ten farms we visited. A sensor that never needs a battery removes that failure. It costs something in return, because a passive device has to be read by a handheld unit carried within range rather than reporting on its own. On these farms that is a smaller cost than it sounds, since the plots are walked daily anyway.
METHODS
Device. Passive LC resonator, no battery, read at 13.56 MHz by a handheld unit at up to 90 mm depth.
Sites. 24 farms across three soil types, 4 sensors per farm at 15 and 40 cm depth in two plots.
Reference. Gravimetric sampling fortnightly at each sensor location, oven dried at 105 degrees for 24 hours.
Comparison. Each farm irrigated one plot on sensor readings and one on its usual schedule.
Outcome. Agreement with gravimetric sampling, unit survival, water applied per hectare, and yield at harvest.
PARTICIPANT FLOW
Farms approached
31
Declined
7
Farms enrolled
24
Sensors installed
96
Units lost
5
Units read at harvest
91
Sensor scheduled
Usual practice
Nov
Dec
Jan
Feb
Mar
0200400600800
Figure 1. Water applied per hectare by month, sensor scheduled plots against usual practice.
Results
Across 24 farms and one season the sensors tracked gravimetric sampling to a mean absolute error of 3.4 volumetric percent, which is inside the band that separates the irrigation decisions farmers actually make. Agreement was best in the loam sites, where the error was 2.6 percent, and worst in the two heavy clay sites, where it rose to 5.1 percent in the fortnight after heavy rain and recovered as the profile drained. Depth mattered less than soil type: the 15 and 40 centimetre sensors differed by 0.4 percent on average. Ninety one of 96 units were still readable at harvest, a survival rate of 94.8%. Of the five lost, three were struck during ploughing and two were dug up by hand and not replaced. No unit failed electrically, which is the outcome the design was aimed at, and no unit drifted enough over the season to need recalibration. Water applied fell 18% (95% CI 11 to 25) on the sensor scheduled plots, from a mean of 2,797 to 2,329 cubic metres per hectare across the season. The gap was widest in January, the month with the heaviest rain, and almost absent in March. Yield differed by 1.2% in favour of the sensor plots, well inside the noise for a single season, so the honest reading is no measured penalty rather than a gain. Nineteen of the 24 farmers said they would keep using the reader without being paid to.
Discussion
The saving came mostly from not irrigating in the week after rain, which is the decision farmers told us they found hardest to judge by feel. That is worth noting, because it means the value of the device is concentrated in a handful of decisions each season rather than spread evenly across it. A reader shared between four farms would capture most of the benefit at a quarter of the cost. The clay sites are where the device is weakest and also where the decision is hardest, which is the wrong way round. The error there is not a calibration problem: bound water in clay shifts the resonant response in a way a single frequency cannot separate from free water, so the next iteration needs a second frequency rather than a better curve. One season, one district and 24 self selected farms is not a basis for a yield claim. The water figure is more robust than the yield figure because it was metered rather than estimated.
Future work
A three season trial across two districts is planned, with metered water, independently weighed yield at harvest, and enough farms to detect a yield difference rather than only rule out a large one. We will add a second resonant frequency to separate bound from free water in heavy clay, which is the one condition where the current device is not good enough to schedule on. We are also testing whether a reader mounted on a bicycle can collect a whole farm without the walk, and whether one reader shared between neighbouring farms is enough. The bill of materials and the reader firmware are being released under an open licence so other groups can build the device without licensing the design.
18%
less irrigation water applied
95% CI 11 to 25, n=24 farms
Conclusion
A sensor with no battery survived a season in the ground, tracked laboratory sampling closely enough to schedule irrigation, and cut water use by 18% with no measured yield penalty. The failures that remain are mechanical and avoidable, not electrical.
Mean absolute error 3.4 volumetric percent
91 of 96 units readable at harvest, no electrical failures
Water use down 18%, 95% CI 11 to 25
Weakest on heavy clay after rain
REFERENCES
1.Dobriyal P, et al. A review of methods for soil moisture measurement. Journal of Hydrology 2012;458:110-117.
2.Kashyap B, Kumar R. Sensing methodologies in agriculture for soil moisture. IEEE Access 2021;9:14095-14121.
3.Vereecken H, et al. On the value of soil moisture measurements. Water Resources Research 2008;44:W00D06.

Block by block

What each block on the board is for, in the order a reader walks it.

Title band and authors
The study title across the band with the university mark, three authors with generated superscripts and a QR to the open bill of materials.
Abstract
The failure the device is designed around, the trial size, the agreement figure, survival and the water saving, all in one block.
Introduction
Why the habit persists, what happens to sensors that need batteries, and the trade a passive device makes in return.
Methods as a labelled protocol
Device, sites, reference standard, comparison and outcome. The paired plot design is stated in the comparison entry rather than left to be inferred.
Participant flow
Farms approached, declined, enrolled, sensors installed, units lost and units read at harvest, with the counts reconciling at every step.
Water figure and key number
Water applied per hectare by month for both schedules, beside a stat block carrying the 18% reduction with its interval.
Results and discussion
Agreement by soil type, survival with the failure modes, the water saving, and why the saving concentrates in the week after rain.
Future work and references
The three season trial, the second resonant frequency, the shared reader question, and three references in journal style. The three references are invented for this fictional study, because the reference block is required on this layout; on your board, replace them with your sources.
How to adapt this board
Sites go into the second entry of the methods protocol, with how many and how they were chosen, because that is the first thing a field study is judged on. Seasons or sampling rounds go into the flow, where deployed, lost and analysed replace the participant path without changing the block. Sensor survival is a result in its own right and belongs in the results text rather than a footnote, since a reader has to know how much of the series is real before they read the chart. The key number block takes the single measurement that answers the question, and everything else stays in the figure.

What makes this board work

Survival is reported as a result

Ninety one of 96 units readable at harvest, with the five losses broken into three struck by ploughing and two dug up. For a device meant to live in a field, that is the finding, not a footnote.

The weakness is named and explained

Heavy clay after rain is where the error rises to 5.1 percent, and the board says why a single resonant frequency cannot separate bound from free water. Naming the failure mode makes the future work section mean something.

The claims are sized to the design

The water saving is metered and stated with an interval. The yield difference is called what it is, inside the noise for a single season, rather than being reported as a gain.

Questions people ask

What goes on an environmental science poster?

The problem in its setting, the sites and how they were chosen, the measurement and its reference standard, the results with intervals, and the limits of a single season or a single region.

How do I show field results without a control group?

Pair the plots. Each farm here irrigated one plot on the sensor and one on its usual schedule, so the comparison sits within a farm rather than between farms.

Should equipment failure go on the poster?

If the study is about equipment in the field, yes. Survival, the failure modes and what caused them are results in their own right and reviewers will ask.

Can I use landscape for a field study?

Landscape suits a study with several short columns and one wide figure, which is common for field work. Check the conference specification first, since some stands only take portrait.

Can I put a photo of the site on the board?

Not yet. Four of the forty layouts carry image figures, but those slots cannot be filled in the current builder, so every figure on a poster is a chart. Describe the site in the second methods entry and link photographs from the QR code.

How do I show a whole season on one chart?

A line chart with one series per site and the sampling dates on the x axis. Keep it to five or six series; more than that and the lines cannot be told apart at a metre. Mark equipment loss as a gap, not a zero.

What goes in the flow when there are no participants?

Sites or samples. This board runs deployed, lost to failure, and analysed as the main path, with the failure reasons in the aside. The block does not care what the unit is as long as the counts reconcile.

Build your own in about a minute

The button below opens the generator with this use case already described. Change the wording to match your own, generate, then edit anything you like.

Make my environmental science poster

Other poster examples

Epidemiology poster

An epidemiology poster has three obligations: print the case definition, reconcile the records flow, and put the denominators where a reader can redo the division. This A0 landscape example does all three for a fictional five year Ross River virus registry study, with incidence by region in one chart and a rate ratio of 2.4 as the key number.

Survey research poster

A survey research poster should put its response rate and weighting where a reviewer looks first, not in a footnote. This A0 landscape example runs four numbers across the middle of the board, 1,204 responses, 61.2 percent allowing remote work, 24.8 percent under ten staff and 78.5 percent at 50 to 199, each with its interval underneath.

Computer science poster

A computer science poster for a systems benchmark needs three things on the board: how the test set was made, what hardware ran it, and the cost of the win beside the win. This A0 landscape example shows a learned index benchmark with its data flow reconciling from 48,000 raw queries to 8,960 evaluated, and the 38 percent insert penalty printed in results.

Want the steps in the builder? Read Add charts and diagrams, then choose the template, theme and size. For everything this generator can do, see the poster maker.

Sources

Written and checked by the OneCraft team. Last checked .