Water purification shock (NDR)

Source: IMAGE-GNM (Global Nutrient Model, river N), Darwin threshold transfer Year: 2015 baseline → 2035 / 2050 / 2080 Category: Scenario · Ecosystem service (ES-direct, Tier-2) Coverage: Global Format: Raster grid (COG) Used in risk analysis: Yes — supplies the Water purification service shock in the Resilience — nature risks; not used in present-day materiality scoring.

What it shows

This layer family projects how much of the water-purification service a location loses under climate-and-land-use scenarios, relative to 2015, read from the nitrogen concentration of its rivers.

It is an ES-direct (Tier-2) layer: it carries an ecosystem service shock, not an ecosystem component shock. Where a natural-asset layer (Habitats, Soils, Species, Atmosphere, Water) is projected onto services through the ENCORE dependency matrix, an ES-direct layer is read straight into the service it names. When this layer resolves for the requested scenario and horizon, Resilience takes the Water purification degradation from the layer itself and bypasses the ENCORE projection; if it does not resolve, Water purification falls back to being projected from its supplying components.

How it is built

IMAGE-GNM routes nitrogen through soils, groundwater and the river network and publishes the in-stream total-nitrogen concentration C (mg N/L) per cell, in 5-yearly bands. The loss is built in three steps.

1. A threshold function turns a concentration into a degradation.

D(C) = clip( (C − 1.5) / (11.3 − 1.5), 0, 1 )      C in mg N/L

• 1.5 mg N/L — the mesotrophic/eutrophic boundary for total nitrogen in streams (Dodds, Jones & Welch, 1998). Below it the river's self-purification copes, and extra nitrogen costs the service nothing: D = 0. This floor is provisional: the reference it stands in for is the pre-industrial nitrogen concentration of each river, which is being derived; until then a literature boundary holds the place.
• 11.3 mg N/L — 50 mg NO₃⁻/L, the limit of the EU Nitrates Directive (and of the Drinking Water Directive). Above it the water is unusable without treatment: the service is gone, D = 1. IMAGE-GNM's concentration is total nitrogen, of which nitrate is only part, so reaching this anchor on total N is a conservative reading of the limit.

Between the two anchors the degradation is linear: D is the distance travelled towards the unusable state. The two anchors and the linear ramp are a V1 methodology choice.

2. Each horizon is a 25-year window, every year kept at its weight. A single IMAGE-GNM band is one simulated hydrological year, not a trend, and differencing two single years made the old shock flip sign from one horizon to the next. Each horizon therefore reads the five bands centred on it (τ − 10 … τ + 10): D is applied to each year, then averaged. A dry year at a high concentration keeps its full one-fifth weight, because for a site it is a real year without usable water. A band where the river carries no water at all is left out — the source writes a zero concentration there, which is a dry river, not clean water, and the absence of water is priced by the water-stress side.

3. The loss is the change against 2015.

loss (%) = ( mean D over the horizon window − mean D over the 2015 window ) × 100

The 2015 window (2005–2025) already contains scenario bands, so the baseline is computed per scenario. The loss runs from −100 to +100.

Sign convention: positive = worse. A positive value means the purification service degrades; a negative value means river water quality improves. This is a pressure layer, like PM2.5 and water stress — the opposite of the service/abundance layers (LUH2, MSA, GLOBIO-ES), whose values fall as the service degrades. The stress-test engine reads the value as a degradation fraction, floors an improvement at zero and caps the degradation at 100 %.

Why concentration and not retention. In-stream retention is the "supply-pure" quantity and was tested first, but it is nearly scenario-invariant: under high-input scenarios retention scales with the extra load, so the retained fraction barely moves. The degradation signal lives in the concentration, which is also the quantity the regulatory anchors are written on.

Horizon provenance. The last IMAGE-GNM band is 2070, so a window centred on 2080 does not exist. The 2080 degradation is extrapolated linearly from two disjoint windows, centred on 2040 (2030–2050) and 2060 (2050–2070), then clipped back to the 0–1 range. Treat 2080 Water purification as a directional trajectory rather than a modelled projection.

Scenario forcing. The pessimistic member of this family is driven by SSP3-RCP6.0, the forcing IMAGE-GNM publishes for the regional-rivalry pathway. The natural-asset layers of the same stress test use SSP3-RCP7.0, and the application labels the pessimistic scenario SSP3-7.0 throughout. The narrative is the same regional-rivalry storyline; the radiative forcing behind this particular service model is the slightly lower one.

How to read it

Values are the projected loss of the water-purification service, in percent, versus 2015. A value of 30 means river nitrogen has moved 30 % of the way from the eutrophication boundary to the Nitrates Directive limit, over and above where it stood in 2015. Most cells score zero: a river that stays under 1.5 mg N/L loses nothing whatever its relative change in nitrogen. Under SSP3 at 2050, about 23 % of river cells lose something; among those, the median loss is about 5 % and one in ten exceeds about 23 %.

Read scenario and horizon together — ssp3_2080 is the pessimistic pathway at the most distant (and extrapolated) horizon.

Class thresholds

The layer is rendered as a continuous symlog gradient rather than discrete risk classes. The legend is drawn in the service framing — green (improvement) on the right, dark red (loss) on the left — so its axis reads −100 on the loss end and +100 on the improvement end, the opposite sign of the raw value. The ±100 endpoints are the bounds of the loss itself, not published thresholds.

The layer's zone rule marks a cell as a high water-purification loss zone when the loss exceeds 30 % — river nitrogen moving more than 30 % of the way to the Nitrates limit. The Water purification materiality flag on the Sites & territories screen reads a different layer, Risk of pesticides pollution, so this rule does not drive it.

Source

IMAGE-GNM — the Global Nutrient Model coupled to the IMAGE integrated-assessment model (PBL Netherlands Environmental Assessment Agency), river-nitrogen component, 5-yearly bands. Threshold anchors after Dodds, Jones & Welch (1998), Water Research 32: 1455, and the EU Nitrates Directive (91/676/EEC).

Comparison with the WWF Risk Filter Suite

This is a forward-looking, horizon-resolved scenario layer and has no equivalent in the WWF Risk Filter Suite, which publishes present-day indicators only. Treat it as a Darwin extension that projects an ecosystem-service trajectory WWF does not itself produce.

Legend

Symbolised field: Water purification loss (%)

Generated from darwin/layers/layer-ndr-shock-ssp{1,3}-{2035,2050,2080}.toml (develop).