Rice Water Management by Growth Stage: Yield and Quality

Rice Water Management by Growth Stage: Yield and Quality

Rice moves through three stages: vegetative (germination through tillering), reproductive (panicle initiation through flowering), and ripening (milk stage through maturity). Water management decisions change yield most during the reproductive stage and change grain quality most during ripening. The reframe that matters: yield does not depend on how often a field dries. It depends on how deep the soil dries, and on which stage that drying lands. A field can dry down and re-flood many times without penalty. Two windows are not negotiable.

What are the key stages of rice cultivation?

Rice has three stages, and only one of them changes length. Shouichi Yoshida's Fundamentals of Rice Crop Science (IRRI, 1981) set out the arithmetic still used today: a 120-day variety grown in the tropics spends roughly 60 days vegetative, 30 days reproductive, and 30 days ripening. Differences in total duration come almost entirely from the vegetative phase.

That has a direct planning consequence: choosing a shorter-duration variety shortens tillering, not the reproductive window. Ricepedia, maintained by IRRI and CGIAR partners, places panicle initiation near day 52 in a 120-day variety, with flowering itself lasting about seven days. Ripening runs near 30 days in the tropics, stretching toward 65 days in cool climates.

The sensitive period arrives on the crop's schedule, not the irrigation schedule.

How deep can the soil dry before yield falls?

Deeper than most irrigation plans assume — up to a threshold that is sharp rather than gradual. Carrijo, Lundy and Linquist's 2017 meta-analysis in Field Crops Research, drawing on 56 studies and 528 paired comparisons, separated alternate wetting and drying (AWD) into two categories. Under mild AWD — soil water potential held at or above −20 kPa, or the perched water level never falling more than 15 cm below the soil surface — yields were not significantly reduced in most circumstances, while water use fell 23.4%. Under severe AWD, where soils dried past −20 kPa, yields fell 22.6%. Duration matters too, though less sharply: the same analysis found AWD confined to a single phase cost 0.5% of yield, while AWD run across the whole season cost 8.1%.

Same practice. Same name. One threshold between a saving and a loss.

Bo and colleagues confirmed this at global scale in Nature Food in November 2024. Across 1,187 paired field observations, variation in yield response was explained predominantly by the lowest soil water potential reached during drying, not by the number of dry spells.

The Sahel gives the cleanest field demonstration. Djaman and co-authors, at the Africa Rice Center's Fanaye station in the Senegal River Valley, published two 2015 seasons in Water in 2018. Continuous flooding applied 1,140 mm in the hot-dry season for 7.50 t/ha. AWD at −30 kPa applied 870 mm — 27.3% less — for 8.3 t/ha. AWD at −60 kPa saved more water and gave up 5.1% of yield. The threshold, not the practice, decided the outcome.

IRRI's field water tube makes the threshold visible without a sensor: a perforated pipe 30 cm long and 10–15 cm across, sunk so 15 cm stands above the soil, re-irrigated to 5 cm depth when the water inside drops 15 cm below the surface.

Which growth stage is most expensive to get wrong?

The reproductive and grain-filling stages, not tillering. A 2018 meta-analysis by Zhang and colleagues in IJERPH, covering 55 rice studies and more than 70 cultivars, found drought at tillering cut yield 20.0% while drought during grain filling cut it 31.5%.

Flowering is the sharpest point inside that window. Yang and co-authors quantified it in Scientific Reports in 2019. Holding soil water potential at −30 ± 5 kPa during flowering — a moderate stress, not a severe one — cut yield 23.2% in one cultivar and 24.0% in another. Spikelets per panicle fell 18–21% and filled grains fell 19%, while effective panicles and 1,000-grain weight were unchanged.

The finding that should change practice is what happened next. Twenty days after re-flooding, net photosynthesis was still 35% below the control and stomatal conductance 41% below. Re-watering did not restore the plant. This is why IRRI's safe-AWD protocol suspends the cycle entirely from one week before flowering to one week after, holding 5 cm of standing water through the window.

What the savings look like in currency

Take a one-hectare Boro rice field in Bangladesh. A 2015 assessment by Bangladesh Agricultural University for the Climate and Clean Air Coalition, drawing on IFDC demonstrations, measured pumping at 101.3 hours per hectare under conventional flooding and 66.7 hours under AWD. A 2019 CIMMYT study of the 2013–14 Boro season put average pump hire at BDT 80 per hour. The 34.6 hours saved are therefore worth about BDT 2,770 per hectare (calculated from the two sources).

BRRI's own Boro-season trial at Kushtia, published in 2009, measured the full loop: 15 irrigations down to 10, irrigation cost from Tk 14,059 to Tk 12,338 per hectare, and yield from 5.25 to 5.63 t/ha. Averaged across its three trial sites, the partial-budget benefit was Tk 4,224 per hectare.

Brad Watkins of the University of Arkansas published a 2022 budget exercise comparing multiple-inlet rice irrigation at 610 mm (24 acre-inches per acre) with cascade flooding at 813 mm (32 acre-inches), yields running 3–5% higher under multiple-inlet. At his published diesel pumping cost of USD 5.35–8.02 per acre-inch, the 203 mm saved are worth USD 106–158 per hectare.

One caution attaches to all these numbers. Enriquez and colleagues, reviewing 20 years of Philippine AWD in Frontiers in Sustainable Food Systems in 2021, found pump-irrigated farms saved up to 40–46% on fuel, worth USD 52–102 per hectare, while gravity-irrigated farms showed no significant income effect. Pandey and co-authors, writing in Water in 2020, name the reason: in Bangladesh, irrigation is priced largely per unit of area rather than per volume of water applied. Where the grower does not pay per unit of water, the saving is real in the aquifer and invisible in the ledger.

Where rice water management stops changing yield and starts changing quality

Yield is finished at physiological maturity. Milling quality is not.

Yang's 2019 flowering-drought study measured both. Alongside the 23–24% yield loss, the chalky kernel rate rose 53–67% and the degree of chalkiness rose 74–76% — while head rice rate and milled rice rate were statistically unchanged. Milling quality survived the flowering stress intact, because it is decided later, during ripening and at harvest.

The mechanism is fissuring. Kunze's 2008 review in the CIGR Ejournal returns to Kondo and Okamura's 1930 experiment, in which grains cycling from 12.6% moisture down to 9.5% and back to 12.4% fissured at a rate of 72% — and the fissures formed while the grain was gaining moisture, not while drying. Overnight dew on an over-dry crop does the same thing.

The cost is measurable. Huysman's data, also reported by Kunze, show that harvesting 50 days after flowering produced 25.5% sun-cracked kernels and a head rice yield of 36.8%, while harvesting five days earlier held cracking below 10% and head rice above 50%. The Arkansas Rice Production Handbook (MP192) places peak head rice yield near 20% harvest moisture for long-grain Cypress, and flags 14% as the point below which kernels fissure on rapid re-wetting. University of Arkansas publication FSA2164 converts that to money: each percentage point of head rice yield is worth USD 0.07 per 100 lb of rough rice, so a ten-point decline costs USD 0.32 per bushel.

Drain timing is the lever. Bruce Linquist of UC ANR, writing in the UC Rice Blog in 2021 and citing work by Cass Mutters, reports that California fields can drain safely 21–24 days after heading. Drain earlier and the soil may dry faster than the grain can finish; drain later and the crop dries past the milling optimum.

Frequently asked questions

How long can a rice field stay dry without losing yield? There is no fixed number of days. Depth matters more than duration. Carrijo and colleagues (2017) found no significant yield penalty in most circumstances while soil water potential stayed at or above −20 kPa, or the water level stayed within 15 cm of the surface — though AWD run across the whole season still cost 8.1%.

Does alternate wetting and drying reduce grain quality? Not milling quality, if the flowering window is protected. Yang and colleagues (2019) found flowering drought raised the chalky kernel rate 53–67% and the degree of chalkiness 74–76%, while leaving head rice yield unchanged. Milling losses come later, from fissuring at drain and harvest. Keep 5 cm of water through flowering.

When should a rice field be drained before harvest? UC ANR guidance, based on work by Cass Mutters, gives 21–24 days after heading as a safe drain point in California. The target is a harvest moisture near 20% for long grain and 22–24% for medium grain, following Siebenmorgen's Arkansas work, because head rice yield falls on both sides of that peak.

Valora Earth builds a stage calendar for a field from its planting date, variety and local weather, then sends flowering-window and drain-date alerts on the messaging channel the grower already uses.