A calibrated, uncertainty-aware, and spatially explicit emission inventory of New Zealand's agricultural diesel tractor fleet.

Non-road mobile machinery (NRMM) is a comparatively under-regulated source of nitrogen oxides and fine particulate matter, with rising relative importance as on-road sources are progressively controlled, yet national-scale activity-validated inventories that quantify uncertainty and spatial heterogeneity remain rare. This study presents a calibrated, uncertainty-aware and spatially explicit emission inventory for New Zealand's agricultural diesel tractor fleet, an internationally relevant case because the country combines ultra-low-sulphur diesel and tightening on-road standards with the complete absence of NRMM-specific emission regulation. The inventory integrates row-level Motor Vehicle Register microdata for 43,367 diesel tractors, the official New Zealand Transport Agency licensing-status table that defines an active licensed fleet of 31,253 units, and 29,278 dealer sales records spanning 2017 to 2025. An IPCC Tier 2 fuel-anchored framework was applied with EMEP/EEA emission factors, tier hours empirically calibrated against a CARB monitoring study, and population-level fuel calibrated against a national 204 million litre tractor-specific consumption target. Monte Carlo uncertainty propagation with 10,000 iterations yielded annual emissions of 2258 t NOx (95% uncertainty interval 1609 to 3177), 115 t PM2.5 (95% uncertainty interval 73 to 190), and 534 kilotonnes CO2 (95% uncertainty interval 419 to 670). Tractor-level Lorenz analysis revealed a highly concentrated emission distribution with Gini coefficients of 0.42 for NOx and 0.51 for PM2.5; the highest-emitting 30% of tractors generated 58% of NOx and 69% of PM2.5. Spatial analysis across 67 territorial authorities identified strong clustering of intensity (Moran's I = 0.293, p = 0.001) and four significant Getis-Ord Gi* hotspots: three formed a Canterbury-Southland belt, with an additional, geographically separate hotspot in Auckland. Mann-Kendall and Theil-Sen tests rejected strict lock-in (p = 0.001) but showed modernisation of only +2.7 percentage points per year, implying that voluntary turnover will not deliver Tier 4 Final and Stage V dominance until the 2040s. The findings provide a potentially transferable evidence base to inform NRMM regulation.