3.1.1.2 - Global water stores and transfers
Water is distributed very unevenly between oceans, ice, ground and atmosphere. Learn why the sizes of these stores differ, how changes of state move water between them, and why a rapid local transfer can coexist with slow global storage change.
Global water stores
Earth's largest water store is saline ocean water, while most freshwater is held in ice or below ground.
The hydrosphere includes Earth's water; in broad usage it encompasses frozen, underground and atmospheric water too. When comparing physical spheres, geographers often distinguish surface liquid water, the cryosphere of frozen water, water within the lithosphere of rocks and sediments, and the atmosphere of gases around Earth. State the classification being used: these sphere labels overlap, so their totals cannot simply be added without removing double counting.
Oceans form a connected surface store. Smaller liquid surface stores include lakes, rivers and wetlands. Groundwater occupies pores and fractures in rocks and sediments; soil water occupies the spaces in the unsaturated near-surface material. Neither is a vast universal underground lake. The cryosphere includes glaciers, ice sheets, snow, sea ice and frozen ground. Major land-ice stores occur in Antarctica and Greenland; seasonal snow and mountain glaciers occur elsewhere where temperatures and snowfall allow. Atmospheric water is mainly vapour, with much smaller quantities of liquid droplets and ice crystals in clouds.
The following selected reservoir estimates are adapted from the USGS global water-distribution table, which cites Shiklomanov (1993). They are longstanding global estimates, not a measurement of water stores in 2026. The table selects reservoirs and therefore does not sum to the whole Earth total.
| Selected reservoir | Volume, million cubic kilometres | Approximate share of all water |
|---|---|---|
| Oceans and seas | 1338.000 | 96.54% |
| Land ice and permanent snow | 24.064 | 1.74% |
| Groundwater, fresh and saline | 23.400 | 1.69% |
| Lakes, fresh and saline | 0.1764 | 0.013% |
| Atmosphere | 0.0129 | 0.001% |
Read the denominator before comparing percentages. Land ice holds about 1.74% of all water but about 68.7% of freshwater in this estimate. These statements are compatible because freshwater is a small subset of all water. Total groundwater is not all freshwater: its estimated fresh portion is 10.530 million cubic kilometres. A large store therefore does not necessarily mean easily accessible drinking water.
To compare magnitude, the ocean estimate is about 57 times total groundwater: 1338 ÷ 23.4 ≈ 57.2. Both figures use the same volume unit. Spatial distribution matters too: a large global groundwater total does not show how much usable water is present in a particular aquifer. Store size, accessibility and transfer rate answer different geographical questions.
Evaporation and its controls
Evaporation transfers liquid water into the atmosphere when molecules escape from a water surface as vapour.
Energy is needed for the change of state. Solar radiation often supplies it, but evaporation can occur below boiling point: boiling is not required. If other conditions remain the same, greater available energy and a warmer surface tend to increase evaporation. The process removes water from the liquid store and adds it to atmospheric water, without destroying water molecules.
The rate also depends on whether vapour can be transported away. Dry air creates a stronger difference in water-vapour concentration between the moist surface and surrounding air than already humid air. Wind replaces moist air near the surface with less saturated air and can maintain that difference. Exposed surface area and water availability matter. A hot dry land surface cannot continue evaporating much water once accessible moisture has been exhausted. USGS's evaporation explanation supports the energy and moisture controls.
Compare an exposed shallow pool and a shaded pool under otherwise similar conditions. The exposed pool receives more solar energy, so evaporation may remove water faster. Add strong wind over the exposed pool and removal of moist air can further increase the rate. This is a controlled comparison: in a real setting, differences in depth, humidity and water supply might complicate it. Do not infer evaporation from temperature alone.
Evaporated water can move across a basin boundary as atmospheric vapour. It need not fall as rain over the surface from which it originated. At global scale, evaporation from the oceans helps supply precipitation over land; atmospheric movement provides the spatial connection between source and destination.
Clouds and precipitation
Condensation and precipitation return atmospheric water to land or ocean, but cloud formation does not guarantee rain reaches the ground.
Condensation changes water vapour into liquid droplets. As moist air rises into lower pressure, it expands and cools. At the dew point, it becomes saturated for its temperature; condensation can occur on tiny particles called condensation nuclei. Clouds contain droplets or ice crystals, not visible water vapour. Condensation releases latent heat, which affects the cooling of rising air.
Droplets or ice particles must grow sufficiently to fall as precipitation. Droplets can collide and combine; in cold clouds, ice particles can grow and later fall as snow or melt into rain. Some falling water evaporates before reaching the surface. A cloud's presence alone therefore does not establish a precipitation input at the ground.
Three common causes of uplift explain contrasting rainfall patterns:
| Uplift process | Mechanism | Spatial implication |
|---|---|---|
| Convection | Strong surface heating warms moist air; buoyant air rises, expands and cools | Localised showers can form over heated areas; intensity and location vary |
| Relief or orographic uplift | Wind forces moist air up a hill or mountain slope | Windward uplift favours rainfall; descending leeward air warms and can produce a drier rain shadow |
| Frontal uplift | Warmer, less dense air is forced above colder air at a front | Rain can extend along a moving boundary between air masses |
The shared mechanism is uplift and cooling, but the trigger and geography differ. A hillside does not manufacture moisture: relief rainfall also requires a moisture supply and suitable wind. Nor does every leeward slope have no rain; other weather systems can bring precipitation there. The Met Office water-cycle explanation links cooling, cloud formation and precipitation.
For a coastal mountain, trace a complete sequence: onshore wind supplies moist air; the slope forces uplift; expansion cools the air; saturation allows condensation; particles grow and precipitation may fall. Descending air on the far side warms, reducing relative humidity if water-vapour content is unchanged. That explains an uneven distribution of inputs to neighbouring basins.
Cryospheric transfers
Frozen water stores grow when accumulation exceeds loss and shrink when losses exceed accumulation.
Freezing transfers liquid water into ice, while melting transfers ice into liquid water. Sublimation changes ice directly into vapour; deposition changes vapour directly into ice. These are transfers between stores or states, not additional sources of water. Snow accumulating and compacting over time can form glacier ice. A glacier also moves under gravity, so a store can transport water while it remains frozen.
Accumulation includes processes adding snow and ice to a glacier. Ablation includes losses such as melting, sublimation and calving, where ice breaks off into water. The balance between them determines net ice-storage change over the chosen period. A glacier with a large snowfall input can still shrink if ablation is larger. Seasonal snow can accumulate in winter and melt during warmer months; deeper glacier ice can persist much longer. USGS's account of ice and the water cycle explains these frozen stores and transfers.
Consider an original illustrative mountain snowpack. Winter snowfall increases its storage; warming in spring increases meltwater supplied to hillslopes and streams. Some meltwater infiltrates the ground and some reaches a channel quickly. The timing of a downstream input therefore depends on release from snow, not solely on the timing of the original precipitation.
Melting land ice transfers water towards the ocean and can add to ocean volume. Melting floating sea ice has little direct effect on sea level because it already displaces water; it still changes the proportions of liquid and frozen water and surface reflectivity. The distinction is about where ice was stored before melting. It does not imply that changing sea ice has no climate consequences.
Scale and transfer timescales
The same transfer can be part of a rapid local response and a much slower change in a larger store.
At hillslope scale, rainfall, infiltration and downhill movement redistribute water over short distances. At drainage-basin scale, many hillslopes and channels connect, so transfers from upstream can become inputs to downstream locations. At global scale, exchanges between land, atmosphere and oceans redistribute water across continents. A rise in one store can be offset by a fall elsewhere; a local water deficit does not imply a loss of water from Earth.
Transfer times vary with pathway and conditions. Evaporation and condensation can change small quantities of water over short periods. Storm runoff through a small basin may occur within hours or days; seasonal snow storage can delay release for months. Deep groundwater and large ice stores may retain water for much longer. These are broad comparisons, not fixed travel times for all basins: permeability, distance, climate and store depth matter.
Residence time is the average time water remains in a store. A simple estimate divides store volume by a steady output rate.
Use consistent volume units; dividing by a volume-per-year rate gives years.
For an original illustrative lake containing 12 million cubic metres with an average annual outflow of 3 million cubic metres, the estimate is 12 ÷ 3 = 4 years. This assumes the stated outflow is representative and the store is approximately stable. It is not a guarantee that each molecule stays four years; short-circuit flow and poorly mixed deeper water can produce very different individual travel times.
Read a transfer dataset by first identifying the store, unit, location and period. Compare like units; distinguish a total volume from a rate. Then describe supported differences and connect them to a pathway without pretending the data establish its cause. For example, an illustrative shallow soil store that drains in days differs from a deep aquifer that releases water over decades. That comparison explains how groundwater can sustain a river between rainfall events, but a residence-time estimate alone cannot determine today's river discharge.