The comparisons Hydrionis will not draw

Directly comparable quantities Not currently buildable

Both are a depth of water over an area, in millimetres, which makes them the only pair here that can be compared as quantities rather than as shapes. Rain falls on the catchment; some evaporates, some enters storage, and what leaves past the gauge is the runoff. The gap between the two lines is everything that did not become streamflow, which is why this pairing is a water account rather than two lines sharing an axis.

Meaningful down to year resolution. Expected timing: None worth measuring at annual resolution: the delay between rain and river is days to months, and a year contains it. What survives is the ratio between the two depths, which is the catchment’s yield.

Held: 17 series of annual precipitation, 0 of annual runoff.

What this pairing could never show, even where it builds: The gap is not evaporation. It also contains change in groundwater storage, abstraction taken out of the catchment, and water imported into it by transfer schemes — which is why a catchment can yield more runoff than the rain that fell on it without anything being measured wrongly. The runoff depth also rests on the catchment area as its publisher delineated it, which Hydrionis does not measure, and a topographic area used where a contributing area was meant rescales every value by the ratio between them.

Contextual — separate panels only Not currently buildable

Rain that is not evaporated or stored reaches the channel, so a river’s flow is the catchment’s reply to its rainfall, delayed and smoothed by everything the water passes through on the way.

Meaningful down to month resolution. Expected timing: Within the month for a responsive catchment; one to three months where baseflow dominates.

Held: 375 series of monthly precipitation, 4,323 of mean river flow.

What this pairing could never show, even where it builds: Millimetres and cubic metres per second are not commensurable and are on separate panels for that reason: the apparent size of a response depends entirely on which scaling each panel was given. The rainfall series is also an area average over a region that is not the gauge’s catchment, so a storm that missed the catchment is still in the rainfall line.

Contextual — separate panels only Buildable

Rain that neither runs off nor is taken up by plants percolates to the water table. An aquifer integrates that recharge over months to years, which is why a groundwater line is smooth where the rainfall above it is not.

Meaningful down to month resolution. Expected timing: Months, and in a deep or confined aquifer years.

Held: 375 series of monthly precipitation, 1,095 of groundwater level.

Hydrionis can build this. The first eligible pair found is country-gbr against borehole-ea-006807ec-a7a4-406e-99b6-d657b0fcf185.

What this pairing could never show, even where it builds: A groundwater level is measured against a datum, and levels at different boreholes are not on the same datum and must not be read as one series. Recharge is strongly seasonal and largely confined to winter in temperate climates, so summer rain can be substantial and still move the level not at all. Abstraction moves the level too, and is not in either line.

Contextual — separate panels only Not currently buildable

A reservoir fills from the runoff of its catchment, so its storage carries the rainfall record filtered through the catchment and through whatever the operator did next.

Meaningful down to month resolution. Expected timing: Weeks to months, and asymmetric: reservoirs empty faster than they refill.

Held: 375 series of monthly precipitation, 179 of reservoir stored volume.

What this pairing could never show, even where it builds: This is the pairing most confounded by decisions rather than weather. Storage is drawn down to supply demand, released for downstream requirements, and held back in anticipation of dry weather, and none of that is in the rainfall line. A reservoir falling through a wet month is usually being operated, not failing to fill.

Contextual — separate panels only Not currently buildable

In a snow-dominated catchment the winter snowpack is the reservoir and the spring melt is the release. Snow-water equivalent is the depth of water held in the pack, so it is a forecast of the volume that will arrive when it melts.

Meaningful down to month resolution. Expected timing: The melt season: weeks to a few months after the peak pack.

Held: 847 series of snow water equivalent, 4,323 of mean river flow.

What this pairing could never show, even where it builds: The relationship is seasonal rather than continuous, and correlating the two across a whole year is dominated by the shared calendar rather than by the melt. How much of a pack becomes streamflow depends on the melt rate, on how dry the ground beneath it is, and on whether the melt is interrupted — a large pack that melts slowly into dry ground produces less flow than a smaller one that melts fast.

Contextual — separate panels only Not currently buildable

The Palmer Hydrological Drought Index is the Palmer index built to track water supply rather than weather: it holds a drought open until the moisture deficit has actually been made up. That makes it the index that should move with river flow rather than ahead of it.

Meaningful down to month resolution. Expected timing: Roughly coincident; the index is built to lag the weather, not the river.

Held: 344 series of palmer hydrological drought index, 4,323 of mean river flow.

What this pairing could never show, even where it builds: The index is computed for a climate division, which is not the gauge’s catchment, and it excludes irrigation, reservoirs and industrial use by construction — so it describes the natural balance while the gauge measures the managed one. It is also an anomaly against 1931–1990, not against a recent normal.