How do you plan water for a restoration project on rainfed land?

Why a budget rather than a rainfall figure

Annual rainfall is the number everyone quotes and it hides the thing that decides the project.

Land can receive perfectly adequate rain and still fail a planting, because the rain arrives in a few weeks and the plants have to survive the months after it. What matters is not the total. It is how much water is available in the longest dry period, at the depth the roots have reached by then.

A water budget puts numbers on that. It is simple arithmetic and it changes the design.

What goes into it

What arrives. Rainfall across the area of the block, and crucially how it is distributed through the year. A month by month profile, not an annual total.

What currently leaves. On degraded land, a large share runs off, because there is little cover to slow it and the surface is often crusted. That runoff takes topsoil with it, which is part of how the land became degraded.

What evaporates. Bare soil loses water directly to the air. Shaded soil loses much less, which is one of the quieter benefits of establishing cover.

What the planting needs. Highest in the establishment years, falling as roots reach deeper moisture. Different for each tier: shallow rooted understory crops draw from a different layer than deep rooted canopy.

What is left. The gap between what is available in the dry months and what the planting needs. That gap is the design problem.

Close the gap with retained rainfall

The instinct is to close it with a borewell. On restoration land that is usually the wrong answer.

This is land where the water table has typically been falling for years. Adding another draw treats the symptom and worsens the cause. It also makes the project dependent on an aquifer nobody controls.

The alternative is to keep more of the rain that already falls. Water that runs off a block is water the block received and did not use.

Contour bunds and trenches run along the slope rather than down it, so water spreads and sits instead of gathering speed. Sitting water soaks in.

Check dams across drainage lines slow the flow through the block, holding water long enough to infiltrate and dropping silt that would otherwise be lost.

Planting pits and basins around individual plants capture water at the point it is needed, which matters most in the first two summers.

Cover itself is a structure. Leaf litter and root channels increase how fast water enters the soil rather than running across it.

Where the structures go

Placement is what separates a scheme that works from one that looks like it should.

Structures follow the drainage, which means they follow the watershed rather than the plot boundary. Water does not respect a lease line. A bund placed by convenience holds less than the same bund placed on the contour.

That also explains an effect worth stating plainly: restoring one block lifts wells on neighbouring land. The benefit reaches people who were not party to the lease, which is one of the things this work produces that cannot be sold and is given away.

The establishment years are the honest part

A restoration block consumes more water than the bare land it replaced, during the years when the plants are small and the roots are shallow.

The budget should say so, and say how that water will be provided and for how long. A plan that claims no establishment water requirement on arid land is describing something other than the site.

What changes over the term is the balance. As cover closes, runoff falls, infiltration rises and evaporation from bare soil drops. Whether the block ends up net positive on water depends on species, spacing and structure placement, which is exactly why it is modelled at the design stage rather than asserted in a brochure.

What to measure

Three things, on a schedule fixed before planting.

Well levels, in the block and nearby, read at the same point in the season each year. This is the slow indicator that shows whether the recharge is working.

Soil moisture at depth, which tells you whether the roots are reaching water or the plants are living on the last rain.

Runoff, or the simpler proxy of whether the structures are holding and silting as intended after each monsoon.

None of these produces a result in the first year. They produce a curve over several, which is the unit restoration is actually measured in. The schedule and the method are published before the readings start, as set out on the measurement page.

See also degraded land and multi tier planting in the glossary.

Related questions

Why not just irrigate from a borewell?

Because on the kind of land restoration usually takes on, the water table is already falling, and a project that adds another draw makes the underlying problem worse. It also creates a dependency: if the borewell fails or the aquifer drops further, the planting fails with it. Retained rainfall is the more durable design and neighbours are not paying for it.

How long does establishment irrigation last?

Until roots reach a reliable moisture layer, which depends on species, soil depth and rainfall rather than on a calendar. Deep rooting species on deep soil in a good rainfall band may need support only through the first dry season. Shallow soil or a long dry period can extend it. The plan should state the assumption and what happens if it proves wrong.

What do recharge structures actually do?

They slow water down. Contour bunds, trenches and check dams give rain time to soak in where it falls rather than running off and taking topsoil with it. The water goes into the soil profile and into the shallow aquifer, which is what raises the water table over years and lifts nearby wells.

Does restoration use more water than it saves?

In the establishment years a block consumes more than the bare land did. Over the term, cover and roots increase how much rain soaks in, reduce runoff and reduce evaporation from bare soil. Which way it nets out depends on species, spacing and how well the recharge structures are placed, so it is worth modelling rather than assuming either answer.