Methodology

What the atlas shows

Three views, tracked month by month for every region on Earth, from January 2015 onward.

• Water Deficit — where is water short? How dry a region is right now, scored against its own history: the rainfall balance, the water in the soil, the health of the vegetation, the groundwater beneath, plus how often drought recurs here.
• Addressable Shortage — how much of that shortage could be closed? The rain missing from the last six months, set against the water actually passing overhead in clouds that could plausibly respond to rain enhancement, at a conversion rate you choose.
• Water Resource — how much water is passing overhead? The atmospheric water arriving above a region, the part that leaves again without ever raining, and the rain that did fall.

Deficit and Resource are measured independently. The first is a scarcity score on the ground, the second a physical quantity in the air. The Addressable view puts the two side by side.

A region is a country, a river basin, or a ~45 km hexagon — switchable at any time, each computed separately (Regions).

How to read a value

Two different grammars are in play, and mixing them up is the easiest mistake to make.

• Deficit layers say “unusual for here.” Every value is a rank against that same region’s own 30-year record (never against another region). A reading of 90 means drier than 90 % of the months in this region’s history.
• Resource layers say “this much water.” Physical water quantities, in millimetres of depth — if the water spread evenly over the region, how deep it would lie. Depth is comparable across regions (regardless of their size); a panel toggle converts it to a whole-region total (in billions of m³ — 1 bn m³ is typically the water to sustain ~2 million inhabitants for a year, though consumption varies a lot by region).

Vocabulary

• Anomaly, standard deviation (sd) — distance from normal, measured in the region’s own year-to-year swing. About one year in six is drier than −1 sd; −2 sd is genuinely rare.
• Percentile — rank against a region’s own history on a 0–100 scale. On the dryness layers, 0 is the wettest the region has been and 100 the driest.
• Atmospheric water — water vapour, cloud droplets and ice moving above the region. Most of it never falls here; it crosses the boundary and moves on.
• Enhancement-eligible atmospheric water — the share of the atmospheric water passing overhead that would plausibly respond to a rain enhancement program. It describes what the technique can act on, not what it would yield — that is the Δ conversion rate you set.
• Trailing six months — droughts accumulate. The shortfall and the rainfall balance read the six months ending in the month you selected, not that month alone.
• mm and bn m³ — 1 mm of depth over 1 km² is 1,000 m³ of water. The unit toggle switches between the two (1 bn m³ corresponds to a square of 1x1x1km dimensions).

What it does not claim

• Not a forecast. Every value describes a month that has already happened.
• The Addressable view is a scenario, not a yield. It answers “if a program converted Δ % of the enhancement-eligible atmospheric water passing overhead, how much of the gap would that have closed” — with Δ set by you, not measured.
• No data is interpolated. Where a source has not measured, the region is marked no-data rather than filled in.

The three bands

Both shortfall views sort every measured region into one of three bands, and paint each band on its own scale. The bands are mutually exclusive — a region is in exactly one, and no region is ever double-coded.

Two readings this separates: a region a millimetre from drought is a different finding from one having its wettest year on record, and a region merely below normal still has a gap that enhancement could close.

The two views split the work. Water Deficit answers which band, in colour: deep red for drought, light grey for below normal. Addressable answers how much of it is reachable, and measures every region against the same bar — the gap to normal — so one green scale covers the map and two regions showing 40 % mean the same thing. Hovering a region in drought adds the nearer reading: the share of the gap to the drought line, which is the more urgent one.

Water Deficit

Land only. Each component is a dryness percentile or a standardised anomaly against a fixed 1991–2020 baseline (clipped to each product’s real record — see Shared components), so a wet country and a dry one can both read “extreme”: the question is how unusual this is for here.

Drought severity headline

• The worst of the four dryness percentiles, taken at region level.
• Reading. 70 → at least one dimension is drier than 70 % of this region’s history; the binding one sets the number.
• Unit. Percentile 0–100, higher = drier.

Addressable Shortage

The one view where Deficit and Resource combine. It divides a shortage measured on the ground by a supply measured in the air, both in millimetres, over the same six months.

Share of shortfalls addressable headline

• Reading at Δ = 0.5 %. 0.20 → a rain enhancement program running through those six months could have closed 20 % of the gap. At Δ = 0 nothing lights up.
• One bar for every region: the gap to normal. The denominator is always the rain missing to make the last six months an ordinary six months, whether the region is in drought or merely below normal. That is what lets one green scale cover the whole map and two regions showing 40 % mean the same thing.
• The nearer bar is still there, on hover. A region in drought has a more urgent gap — the one to the drought line — and its hover card gives that share too, alongside the gap in millimetres, so the percentage sits on a quantity rather than floating free.
• Δ scales the result linearly — the slider changes the magnitude, not which regions rank highest.
• Nothing is carried from one place, or one month, to another. Wherever a share covers more than one region — a country, a continent, the world — or more than one month, each part is first held to its own gap, and only then are they added up. A region with more water overhead than it needed contributes what it needed and no more; the surplus is not spent on a region that fell short, because no one can move it there. The consequence is the one you would expect of an honest total: a country can never read higher than the regions it is made of. If each of them could have closed 30 % of its gap, the country reads 30 %, not 100 %. The same rule holds down the whole chain — month to period, region to country, country to continent to world.
• Unit. Ratio, shown as %.

Shortfall, and the gap “to normal”

The two numbers on the shortfall row answer different questions, and the difference between them is exactly one standard deviation of the region’s six-month rainfall:

• Shortfall to the drought line — the solid curve and the big value. The bar is the drought line: one standard deviation below the 1991–2020 mean, the same z ≤ −1 convention as the drought-frequency layer. It reads zero whenever the region is above that line, so a merely below-average season carries no deficit at all.
• To normal — the dashed companion curve. The bar is the 1991–2020 mean itself: the rain missing to make those six months an ordinary six months. It turns positive as soon as rainfall dips below average, and it stays visible through the below-normal-but-not-yet-drought band where the shortfall is still zero.

Read together, the gap between the curves is the safety margin: a wide gap means the region is running dry but has not crossed into drought; a narrow one means normal and drought are close together, which is the signature of a low-variability climate where a small rainfall loss is already an emergency.

The coverage headline divides by the gap to normal, everywhere (The three bands). So the atlas sizes the addressable problem against restoring an ordinary year — for regions in drought and for regions merely below normal alike. The drought-line share is the more urgent reading where there is a drought, and it is on the hover card.

Why six months, and what the ratio assumes

• The window is matched on both sides. The shortfall accumulates six months of missing rain, so the numerator is the total enhancement-eligible water of the same trailing six months. The panel row shows one month, so dividing that row by the shortfall gives roughly a sixth of the headline.
• Seasons are compared to like seasons. μ and σ are computed for windows ending in the same calendar month, so a monsoon climate’s dry season is measured against other dry seasons rather than a seasonless average.
• Incomplete windows show nothing rather than a misleadingly low number.
• The 0–2 % range is an assumption, not a measurement. Converting 0.5 % of the enhancement-eligible passing water is of the same order as the 5–15 % rainfall enhancement that randomised rain enhancement programs typically report (the region panel shows the implied ”% of natural rainfall” next to the coverage so the conversion rate is transparent).
• Two departures from a full “back to normal” accounting. The shortfall is precipitation-only — a hot drought’s inflated evaporative demand is not water that extra rain replaces mm-for-mm — and the bar is exiting drought, not restoration to the mean.
• Caveat on the eligible share. A monthly cloud-property fraction is applied to the month’s flux, assuming the flux is spread evenly across the month’s cloudy and clear hours.
• The eligible share is modelled, and the amount of cloud water is uncertain to a factor of two. Where the liquid sits in the vertical, and how cold it is there, comes from a weather model rather than an instrument — no satellite measures the phase of cloud water through the depth of a cloud, and reanalyses are known to turn supercooled liquid into ice more readily than the atmosphere does, which makes this share a low estimate in mixed-phase conditions and lowest over the Southern Ocean. Independently of that, the model and the MODIS satellite retrieval disagree about how much cloud liquid water there is at all: they agree within about 15 % across the tropics and differ by three to four times across the northern mid-latitudes in winter, with a global median ratio near 2. That envelope sits under every figure derived from the eligible share.

Seasonal timing — when in the year the largest share is addressable

Everything above answers where the shortage is and how much of it is addressable now. Seasonal timing answers the other question an operator planning a programme actually has: in which month of the year would a programme close the largest share of the gap, and whether that is a winter or a summer story. It is a per-region profile of the calendar rather than a layer — there is no month to paint it on — and it is the same share as the Addressable Shortage reading above it, taken per calendar month instead of per month of the record.

• Two readings, and Δ moves only one. The share of the gap that is addressable, and how often the region is short in that month — separated into the years it was in drought and the years it was below normal without being in drought. The conversion assumption changes what is addressable; it does not change how often the region is short.
• The numerator averages every year; the denominator only the short ones. The eligible water is a six-month total, the same window the shortfall accumulates over, averaged over the whole record because water passing overhead does not depend on whether the region needed it. The gap is the size of the shortfall when there is one, so it averages only the years that month actually came in short.
• Short means below normal, and the drought line is the second reading. The share divides by the gap to normal, the same gap the Addressable Shortage reading uses. Where a month has enough drought years behind it, the share of the nearer gap — to the drought line — is given as well; where it does not, that reading is absent rather than zero.
• Both underlying numbers are stated. “short 4 years in 10 · 62 mm passing overhead”. Neither can be recovered from the share, and the frequency is also the size of the sample the gap was averaged over.
• Countries and basins only. The high-resolution hexagons carry a trailing twelve months, which is exactly one sample per calendar month. One year is a realisation, not a climatology, and drawing it as one would be a fabrication — so the row is simply not offered there. A cell’s Region tab is an ordinary first-level region and does show it.
• Eleven or twelve years is thin, and it is a climatology rather than a long one. A calendar month with fewer than eight years behind it is left out entirely; a gap averaged over fewer than three short years is left blank while the rest of that month’s reading stands; and a region whose water readings the no-resource floor withheld gets no profile at all rather than a row of zeros.

Water Resource

Observed and reanalysis quantities only — no percentiles, no baselines, no ranking against history. These are physical amounts of water: how much arrived above the region, how much left again without raining, and how much reached the ground.

How the flux layers are measured

• Boundary integrals, not cell sums. The two flux layers are line integrals of the ERA5 total-water flux around the region’s outline, divided by its area. That makes the depth comparable across regions of any size and keeps water from being counted twice as it moves from one grid cell to the next.
• All three phases count. Vapour, cloud liquid and cloud ice. Vapour dominates — roughly two orders of magnitude larger than the cloud-liquid part, and the reservoir that feeds most rain by condensing in place over the region.
• Inflow and outflow are independent measurements. Outflow is not inflow minus rainfall: subtracting vapour-fed precipitation from a condensate-only inflow mixes two different reservoirs, so both layers are integrated directly from the same flux field.
• This is the water passing through, not water available for use. The Addressable view is what applies the eligibility gate and the conversion rate to it.

The worldwide figures

The Worldwide button in the top bar opens the whole planet for one six-month window: how many countries a drought touches, how much of the world’s land is short of rain, how much rain that actually is, and how much of it a rain enhancement program could add. It is computed once by the pipeline, one row per month, and read straight off — not a summary of what happens to be on screen. It covers every region on Earth whether or not the camera is pointed at it.

Every figure covers the six months ending at the month you pick — the same window the shortfall layers accumulate over. The months on offer are the trailing twelve the high-resolution cells carry.

It does not follow the region family you are viewing. Switching between countries, basins and hexagons changes the addressable ratio by a factor of three for the same month — the atmospheric-water layers are boundary integrals divided by area, so a smaller polygon reports a larger throughput for the same weather. The world figures are pinned to the hexagons, which are equal-area, with one exception: countries-touched is counted on the first-level regions, because a hexagon has no country.

• “Measured land” is the honest denominator, and it is within-family. Every share divides by the regions that actually have a shortfall reading that month — never by a fixed number for the Earth’s land area. The areas the atlas holds are each region’s buffered supply extent (land plus a 5 km coastal margin, see Masks), so they overlap slightly and a family’s areas add up to more than the real land surface — which is why every share is compared within the same family.
• A region reading zero is counted, not skipped. “No shortfall” is a measurement. Only a region with no reading at all — masked, unobserved, or outside the record — leaves the denominator.
• “Drought touches N countries” counts distinct countries with at least one first-level region in drought. No threshold, no share-of-the-country test: one region is enough to say the country is touched, and the figure is stated that way. It is the one line that comes from the first-level regions rather than the cells, because a hexagon’s parent is a region rather than a state.
• The addressable share is a ratio of sums, weighted by area. The eligible water of every region in drought over the rain missing from every region in drought — so a large region counts for more than a small one, and a single near-zero-shortfall region cannot distort the headline. Regions whose six-month supply window is incomplete are left out of both sides rather than counted as zero.
• Two addressable readings, against two different needs. One divides by the rain missing to reach normal across every region that is short, which is the denominator the Addressable map paints. The other divides by what the regions in drought need simply to clear the drought line, over those regions alone — the more urgent need, with the merely-dry regions set aside, so the two are not averaged into one number that answers neither question.
• Volumes lead, the percentage follows. A ratio is not comparable across region sets; a volume of missing rain is a quantity. So the figures lead with how much rain would bring every region that is short back to a normal six months, how much of that is what regions in drought need simply to clear the drought line, and what Δ could add against the first of those. Where the eligible water exceeds the gap, that is stated in words rather than as a share above 100 %: how far past the gap it goes is an artefact of a small denominator, not a finding.
• By continent, from the same rows. The scope picker re-reads the table at a continent instead of the planet. It is a grouping of the same aggregation, not a second measurement: every region sits in exactly one continent or in none, so the continent rows and the world row partition the same population and can be read against each other. The regions in none are the disputed areas geoBoundaries files under an internal number rather than a country — nineteen of them, in the world figures and in no continent, because assigning them one would be asserting a sovereignty nobody agrees on. River basins have no continent rows at all: a basin crosses borders by construction.

Shared components

Baselines

Nominal window 1991–2020, per-region and source-pinned ( the baseline comes from the same product the month is scored against). The window is clipped to each record (minimum 15 years / 60 months):

Zonal aggregation

Every layer is aggregated to the region as an area-weighted mean (exactextract coverage × cos-latitude weights; a region needs ≥ 10 % valid coverage to score) — except the two atmospheric-water flux layers, which are boundary line-integrals divided by area (not a sum of internal grid cells). Coarse sources stay at their native grid; a region smaller than the source cell inherits that cell’s value.

Percentiles

Dryness percentiles pool all baseline months into one distribution (midpoint ties), not one distribution per calendar month. De-seasonalisation happens upstream — the SPEI z-score, the GDO anomalies, GRACE’s own percentile, and soil moisture standardised per calendar month — so a score reads as “this month’s anomaly ranked against the pool of all monthly anomalies”.

Units: per area or total

Water quantities are stored as depths in mm — the water spread over each square metre, comparable across regions of any size. A panel control switches the Water Resource flux layers and the Addressable mm layers (Shortfalls, enhancement-eligible total) to a whole-region total: depth (mm) × area (km²) × 10⁻⁶ = billion m³, shown as “bn m³”. Totals are real, but a large region reads high simply because it is large.

Availability gating

Each source sits behind an availability gate. A month the source has not yet published is marked unavailable in the panel rather than drawn empty, and nothing is interpolated: a region with no measurement is marked no-data and contributes nothing. One exception — for a single month past a layer’s record it falls back to its most recent available month, always dated in the panel.

Masks

• Water Deficit — land only.
• Water Resource & Addressable — land plus a 5 km coastal buffer; open ocean masked.

The no-resource floor

Some places have essentially nothing overhead to work with, and the map marks them. A grid cell counts as no-resource only when both tests agree: almost no rain reaches it (mean annual precipitation, CHIRPS) and almost no cloud water passes over it (MODIS liquid water path weighted by cloud cover). Both thresholds are set from the two places the method names as the definition of “no resource” — the Sahara core and the Arabian Empty Quarter — at the level that catches essentially all of both.

• Why two tests and not one. Cloud water alone cannot tell a hyper-arid core from a farmed semi-arid highland: the deep Sahara’s readings are inflated by dust and land higher than the Sanʿaʾ highlands or Sudan’s Gezira. Any single cloud threshold either leaves half the Sahara unmarked or swallows real agriculture. Rainfall separates them cleanly — under about 110 mm a year for the two cores against 278 mm or more everywhere that must be kept.
• The cloud test can only ever remove a place from the floor, never add one. That is its purpose: to hand back the deserts that are dry on the ground but cloudy overhead — the Lima garúa, the Galápagos, Cape Verde, and the Gulf, where the largest operational rain enhancement programme in the world runs. With the cloud source switched off, the rainfall test runs alone.
• It is graded, not on or off. A region is not uniformly desert — Abu Dhabi’s south is erg and its coast is not — so what is stored is the share of the region that is no-resource.
• Values are only withheld at the extreme. Below a 90 % share the region keeps every reading; at or above it the resource layers and the shortfall are blanked.

Permanent ice

• Permanent ice has no ground readings. The Greenland and Antarctica ice sheets, and any hexagon at least half covered by glacier (Randolph Glacier Inventory 7.0), carry no drought severity, soil dryness or groundwater — those measure a land surface that is not there. The rainfall balance, rainfall and the atmospheric-water layers still apply, because they describe the air above rather than the ground below.

Regions

Region and country

A region’s panel carries a scope switch: the same rows, for the wider area it sits in, built by combining the first-level regions inside it. Nothing is stored per country — it is assembled as you open it.

• Water quantities are area-weighted means over the regions that were measured that month, so a country reports its measured half rather than a half-strength whole. The per-area / total switch then works exactly as it does for one region: the country’s mean depth over the country’s total area.
• The ranked indicators are left out, not averaged. Drought severity, soil dryness, vegetation stress, groundwater, rainfall balance and baseline water stress are all ranks of a region against its own history. Averaging a country’s ranks produces a number that ranks it against nothing at all, so those rows are simply absent from the country view.
• What replaces them is a count. The Regions short of rain chart shows, month by month, how many of the country’s regions were in drought and how many were below normal — the same two bands the map paints. Hovering it names the regions counted in that month, and each name opens that region.
• Static facts are stated as counts too — how many regions sit below the no-resource floor, how many are at high or extremely high baseline water stress — because a boolean mask and a fixed score have no meaningful country-level average either.
• A hexagon can climb two rungs. A cell sits inside an administrative region which sits inside a country, so its panel offers Cell | Region | Country — you can read the ~45 km cell against the region around it and against the whole country without leaving the panel or changing what the map is drawing. A first-level region offers Region | Country.
• No wider scope where there is none. River basins cross borders by construction, and a handful of disputed territories are filed under themselves rather than under a country — in both cases the switch is not offered.

Sharing a view

The address bar always mirrors what is on screen — the view, the indicator, the region family, the month or range, the conversion rate, the units, the theme, the camera, and any open region panel. Copy it to share exactly what you are looking at. Nothing is added to your browser history as you explore, so Back still leaves the page rather than stepping through every slider move.

Note on high resolution hexagons

• Equal-area cells make the map honest — the contrast you see is variation in the water, not in the size of the polygons a country happens to have.
• Recomputed, not downscaled — same sources, same aggregation, own 1991–2020 baselines. No country or basin value is interpolated onto a hexagon.
• A cell exists wherever it touches land, not only where its centre falls on it, so narrow coastal strips and peninsulas are covered. Cells that catch only a hairline of a coastline (under 5 % land) are dropped rather than drawn from a few offshore grid cells.
• A coastal cell is measured over its land. The hexagon you see is the whole hexagon; the values in it are averaged over the part of it that is land, so a cell that is mostly sea reports the weather on its coast rather than the ocean beside it. Its water volumes are over that land area too.
• ~45 km is near the floor the sources support (coarsest input 0.25°), so neighbours inside one source cell can share a value.
• Trailing 12 months only, single months, no period mode or playback — and therefore no seasonal profile (Seasonal timing).
• Glaciated cells carry no ground readings (Masks).

Months, lags and periods

• The globe opens on the most recent month available, and each layer carries the date its source actually published that month.
• Typical lags. 6-month rainfall balance, vegetation, groundwater and atmospheric water arrive within days to a week, so they are present in the newest month; soil moisture runs 2–3 months behind and rainfall ~3.5 months, so they are absent from it.
• Period replaces the single month with a range and shows the mean across it — except in Addressable Shortage, where the map divides the sum of the six-month eligible-water totals by the sum of the shortfalls across the months where a shortfall exists.
• High-Resolution is trailing 12 months only — its slider covers just those months (Regions).

A ratio of sums cannot be blown up by one near-zero-deficit month the way an average of monthly ratios would be, and both sides carry the same overlap weighting from the trailing six-month windows, so neither side is double-counted relative to the other.