Cost-Effectiveness Overview¶
A synthesis of what the model structure implies about cost-effectiveness of measures — before running scenarios. This page should be updated and verified as scenario results become available.
How to think about cost-effectiveness in this model¶
Cost-effectiveness is not a fixed property of a measure — it depends on:
- Retention (
TotRet,SurfRet): A cheap measure in a high-retention catchment delivers little N to the coast. The same measure in a low-retention catchment near the coast may be highly cost-effective. - Field characteristics:
prodcost(i),leaching(i),soil(i),livestock(i)all shift costs and effects. - P co-benefits: Measures with P effects (FO, BZ, LRh, etc.) effectively get a "discount" when both N and P targets must be met — the optimizer uses them partly for N and partly for P.
- Tripartite floors: Force sub-optimal measures into the solution, raising the average cost.
For this reason, the table below gives gross cost-effectiveness (before retention) and notes the retention sensitivity.
N measures — indicative cost-effectiveness¶
All values are approximate ranges from model parameters. "At coast" values assume illustrative retention levels.
| Measure | N effect (kg/ha/yr) | Cost (DKK/ha/yr) | Gross CE (DKK/kg N) | Retention type | Notes |
|---|---|---|---|---|---|
| N10 | 2–5 (field-specific) | 44.5 | ~9–22 | TR | Very cheap; best on high-Nhan fields |
| N20 | 4–10 | 178 | ~18–45 | TR | Less CE than N10 per kg N; 4× cost for 2× effect |
| CCS | 12–45 | 315–396 | ~7–33 | TR | Excellent CE on sandy high-livestock fields |
| IC | 14 | 325 | 23 | TR | Stable CE; modest effect |
| EW | 17 | 200 | 12 | TR | Good CE but limited potential (30% of winter area) |
| EC | 34–51 | prodcost ± adj | Variable | TR | Can be cheap or costly; depends on crop value vs EC value |
| CCW | 12–45 | 1,980–3,168 | ~44–264 | TR | Expensive; same N effect as CCS at 5–8× the cost |
| BZ10 | leaching−12 | prodcost | ~prodcost/(leach−12) | SR | CE highly variable; best where leaching high, prodcost low, SurfRet low |
| BZ20 | leaching−12 | prodcost | same as BZ10 (wider strip = more area) | SR | Same CE per ha as BZ10; different total cost |
| WL | 90 (flat) | 3,486 + prodcost | ~39–78 gross | NR | Key advantage: no retention discount. Best CE in high-retention catchments |
| LRl | 40 | prodcost + 1,016 | ~25–75 | NR | Also NR; competes with WL. Fixed 40 kg/ha regardless of field leaching |
| FO | leaching−8 | prodcost − annuity | ~0 to 30 | TR | Can have near-zero or negative net cost where annuity is large |
| LRh | leaching−6 to −18 (k-specific) | prodcost [+200 if livestock ≥0.8] | ~15–50 | TR | National: 108 catchments (updated 2026-04-05 from original 13) |
| SA | leaching−6 to −27 (k-specific) | prodcost + 250 [+200 if livestock ≥0.8] | ~30–150 | TR | National: 108 catchments (updated 2026-04-05); tripartite floor forces it beyond CE frontier |
Key insight: WL and LRl are NR measures — their cost-effectiveness does not deteriorate with distance from the coast (unlike all TR/SR measures). In catchments with TotRet > 50%, WL becomes progressively more competitive.
P measures — indicative cost-effectiveness¶
P cost-effectiveness is harder to generalize because it depends on erosion_field(i), macropore_field(i), and matrix_ha(i) which vary enormously across fields.
| Measure | Main P pathway | P effect | Cost (DKK/ha or unit) | Indicative CE (DKK/kg P) | Notes |
|---|---|---|---|---|---|
| PPC | Erosion | 0.9 × erosion_field | 80 + 0.1×prodcost | Highly variable | Cheapest erosion measure; best on high-erosion fields |
| NPB10 | All 3 | 0.05 × (E+M+X) | 50 | Highly variable | Best where all 3 pathways active |
| NPB20 | All 3 | 0.20 × (E+M+X) | 100 | ~2× better CE than NPB10 | Generally preferred over NPB10 |
| OT | Erosion | 0.5 × erosion_field | 200 | Variable | Less CE than PPC for erosion alone |
| IBZ | Field-specific | IBZ_eff(i) | 7,938 | High if IBZ_eff large | Precision targeting may justify cost |
| PWET | Erosion (flat) | 15 × PotV | prodcost + 9,733 | ~650–2,000 | Fixed rate; best where P loads are very high |
| Ochre trap | Ochre P | 140 kg P/trap | 121,537 | ~868 | High certainty; expensive upfront |
| Sand trap | Eroded sand | 7–26 kg P/trap | 9,328–12,460 | ~360–1,780 | Cheap; variable effect by geo zone |
| Re-meandering | Stream-specific | w_P_red | 7,000–25,000/km | Highly variable | High CE on small streams with strong effect |
| Raising | Stream-specific | w_P_red | 5,000–19,000/km | Highly variable | Competes with re-meandering; cannot combine |
| Trees | Stretch-specific | tree_eff | 290.8/yr | Usually very low | Best CE of all P measures if effect > 0 |
N vs P cost structure¶
A rough structural observation from the model: - N costs are dominated by opportunity costs (prodcost) on field measures — they scale with land productivity. - P costs are dominated by infrastructure costs (ochre, sand traps, re-meandering) and erosion-reduction measures — they scale with P loss intensity. - Fields with both high leaching and high erosion get double use from measures like BZ10/20, FO, LRh — these deliver both N and P benefits, effectively halving their per-target cost.
The NR advantage — a key model insight¶
Measures in the NR group (WL, LRl) do not suffer the retention penalty that makes TR/SR measures less effective in distant catchments. This creates a counter-intuitive result:
In catchments with 70–80% total N retention, WL at 90 kg/ha and 3,486+prodcost cost may be cheaper per ton N at coast than CCS at 45 kg/ha and 315 DKK/ha — even though CCS is far cheaper per kg N gross.
This is one of the model's most policy-relevant findings: the optimal measure mix is spatially differentiated, and retention geography matters as much as measure efficiency.
What scenarios will confirm¶
This page currently reflects analytical reasoning from model structure. Once scenario results are available:
- Run N_only_no_tripartite to get empirical cost-effectiveness rankings from the model's own choices
- Extract Total_cost_k.l(k) / Total_reduction_k.l(k) for each catchment — the model's revealed shadow price of N reduction
- Extract N_reduction_per_measure.l(j) to see which measures dominate nationwide