How deep the roots are, through a season#

Written 2026-10-13. Two questions: how a well-watered crop's root depth runs with time, and what on a farm shows how deep the roots actually are. The soil-moisture advisory needs the answer every morning: the root depth sets how much water the bed holds for the crop, and so when to water and how much.

The curves#

Three ways root depth is modelled over a season, for a crop that is not short of water:

AquaCrop (Eq. 3.6a): Z = Zo + (Zx − Zo)·((t − t0/2) / (tx − t0/2))^(1/n). The roots start deepening at half the time to emergence, t0/2. With n = 1 they deepen at a steady rate; above 1 they deepen fastest just after planting. Their average rate is capped at 5 cm a day. The manual keeps roots and canopy apart: the time to the full canopy is set independently of the time to the deepest roots, so "the interdependence between root and shoot is not tight". A crop's height or cover is therefore no measure of its roots. Across the 65 crops with both in FAO-56's Tables 12 and 22, maximum height and maximum root depth correlate only loosely (r = 0.52).

Pedersen and Thorup-Kristensen fitted the straight line to the roots of vegetables at the Danish Institute of Agricultural Sciences, leek, carrot, cabbage and beetroot among them. Their crops' rates are not legible in the archived copy, so none are quoted here.

Tomatoes, side by side#

• FAO-56's 1998 edition: 15 cm at sowing to 70 cm on day 70, of a 145-day season.
• FAO-56's 2025 revision, which fieldmesh now uses: fresh-market tomatoes' roots to the shallow end of 0.6–1.2 m by 985 degree-days (Table 6.11's 325 and 660, base 7 °C, upper 28 °C), of 2,065 to the season's end. On McAllister's summers that is 59 to 74 days (median 66), and the season 113 to 132 days (median 122), not 145.
• AquaCrop's parameters, as AquaCrop-OSPy ships them (TomatoGDD, transplanted): 0.3 m to 1.0 m by 891 degree-days, of 1933 to maturity. The base is 7 °C and the upper limit 28 °C.

The two end their rooting at nearly the same point in the season: 48% of FAO-56's days and 46% of AquaCrop's degree-days. The difference is the clock. On McAllister Farm's thirty recorded summers, planted 1 May, 891 degree-days came after:

• 56 days at the earliest (2018);
• 69 days at the latest (1997);
• 62 days in the median summer.

These counts use the manual's methods 2 and 3, which give the same days here. FAO-56's fixed 70 days would be:

• about a week late in a typical summer;
• two weeks late in a warm one.

Running late is the safe direction for watering: a root zone counted shallower than it is gets watered a little sooner.

"Watered properly" matters to the curve#

The curves describe roots growing unhindered. AquaCrop slows root growth in three cases:

• when water stress starts to cut the crop's transpiration;
• when the soil below the roots is too dry;
• in a soil horizon the roots cannot easily penetrate.

That last case is set as a horizon's penetrability: at 0% the roots stop there. A watering that only ever refills the root zone leaves the soil just below it as it was. In a dry year that soil can be too dry for roots to enter, which no curve on its own shows.

What shows how deep they are#

• The stair-step. Where roots draw water, a probe's reading falls by day and stays level at night. As the crop grows, the steps show up at deeper probes. This is how growers read root depth off a multi-depth probe (Barrett, Specialty Crop Grower, 2018). A probe the roots have not reached does not step.
• Soil well below field capacity. Drainage slows to almost nothing near field capacity. So soil that keeps drying well below it, on days nothing wets it, is soil roots are drawing on. Near the surface, the sun dries it too.
• A spade, a soil probe, a penetrometer. These find a hardpan, the plough's pan, the compacted fill under a new garden, or the bottom of a raised bed. They are direct, and worth doing once per bed.
• Not the canopy. As above, the canopy's size is no measure of its roots.

What fieldmesh does with it (built 2026-10-13)#

• Bed.rooting_limit_mm: how deep the soil lets roots go. The crop's curve stops there, and a probe set into the pan is not counted.
• Root signs from the probes (root_signs), over the last two weeks of dry days. A probe whose soil stays level has no roots; one that falls and goes 0.02 under field capacity has roots. The stair-step, a fall by day against one as fast by night, was an option until 2026-10-17: on the simulated summers it found a pan no better and saw false stops on more mornings, and it was retired. Where a probe the curve passed at least 15 cm ago has no roots, under one that has, the roots likely stop between the two, and a note asks for the rooting limit.
• A range of roots, not one depth. When to water is timed by the shallow end, how much by the deep end: each probe's own deficit down to the deepest roots. A stop the probes find moves only the shallow end. So does a young planting beside an older one (Bed.plantings), which is the same problem: plants at different stages side by side; a probe named for its planting reads that planting's soil.
• Degree-days as the crop's clock (thermal.py), as the 2025 revision of FAO-56 gives them: its base and upper temperatures for each crop (Table 6.10), and for the 10 of fieldmesh's 20 sown crops it has field-observed values for, each stage's degree-days (Table 6.11). The other ten keep the 1998 days, taken as right for the farm's normal year from its past years' weather, so a warm season runs them ahead and a cool one behind, as the revision's own Table 6.12 turns those days into degree-days.

The backtests are in the advisory's README: the clock, plantings at different stages, and roots that stop. In brief, on thirty simulated summers, on the 2025 revision's curve: over a pan at 30 cm the curve alone left tomatoes short every summer for eleven days, and the root signs cut that by more than 99%, on a simulated soil that moved no water below field capacity. On a soil that moves it as a real one does (since 2026-10-15), water rising out of the soil under the pan kept the crop watered on the curve alone. Since 2026-10-16 the simulated roots draw most from the top, as real ones do, and the crop uses its water as FAO-56's does: over a dense pan, one that passes water slowly, the curve alone went short in 10 to 18 summers and the signs in 9 to 12, the rooting limit set by hand in 6 to 9. But a probe at 30 cm, drawn on lightly by roots densest above it, now reads near level with no pan at all, and the signs see a stop on about half of a season's mornings either way. A young planting of lettuce among tomatoes, on the weather's advice for the tomatoes alone, was short in 8 summers on the earliest soil, none on the soil of 2026-10-15, and 22 now that its patch dries at the surface; told of it, never. Timing the curve by degree-days moved a season's water by up to an eighth with the season's warmth, and left no crop short either way.

Next#

• The other crops' degree-days, for the ten the revision gives only approximate ranges for (its Table 6.12).
• Real beds: the curve's shape, straight or the nth root, the level line against a real probe's noise, and a pan found by its probes and then dug.
• A level line that expects the roots' own density: a probe deep in the roots, where they are sparse, should fall slowly; judged against that, rather than against nothing, it would not read as a stop.

Sources#

• R. G. Allen, L. S. Pereira, D. Raes and M. Smith, Crop evapotranspiration, FAO Irrigation and Drainage Paper 56, 1998: Table 22, and Annex 8.
• FAO, AquaCrop reference manual, chapter 3: calculation procedures, version 7.3: section 3.6, the effective rooting depth; section 3.3, growing degree-days.
• AquaCrop-OSPy, crop_params.py, Apache 2.0.
• A. Pedersen and K. Thorup-Kristensen, Modelling root distribution and rooting depth to analyse crop N exploration from different soil depths, Danish Institute of Agricultural Sciences, archived at Organic Eprints.
• C. Barrett, Soil moisture sensors improve irrigation, Specialty Crop Grower, 2018-07-31.
• P. Paredes et al., Base and upper temperature thresholds to support the calculation of growing degree days aiming at their use with the FAO56rev crop coefficients curve: a review, Agricultural Water Management 319 (2025) 109755, open access, CC BY 4.0 (2025a).
• L. S. Pereira, R. G. Allen, P. Paredes, R. López-Urrea, D. Raes, M. Smith, A. Kilic and M. Salman, Crop evapotranspiration, second edition, revised 2025, FAO Irrigation and Drainage Paper No. 56 Rev.1, CC BY 4.0: section 6.8.1, Tables 6.1, 6.2, 6.10, 6.11, 6.12 and 8.1–8.2, Eq. 8.5.
• P. Paredes et al., Estimating the lengths of crop growth stages to define the crop coefficient curves using growing degree days (GDD): application of the revised FAO56 guidelines, Agricultural Water Management 319 (2025) 109758 (2025b); the revision's Table 6.11 is its work.
• McAllister Farm's weather: NASA POWER daily, 1991 to 2020, CC BY 4.0.