Supply Shed Travel Time Assumptions¶
When you call /batch_supply_shed without specifying a travel_times value, the API automatically selects a
commodity-appropriate default. This page documents the methodology behind those defaults, the peer-reviewed
sources that support them, and the caveats you should be aware of when interpreting results.
What is a Travel Time Isochrone?¶
A supply shed is generated by computing a travel-time isochrone centred on the submitted processing facility. The isochrone captures all land reachable within the specified travel time using the actual road network, accounting for road type, speed limits, and topology. This is more ecologically meaningful than a simple radius buffer because agricultural sourcing relationships follow roads, not straight lines.
The travel time represents the maximum one-way journey a farmer would typically make to deliver raw material to the first industrial aggregation point — the nearest large, fixed, detectable processing facility that performs first-stage transformation of the commodity (e.g. a palm oil mill, slaughterhouse, or cooperative wet mill). Non-industrial intermediaries such as village traders or itinerant collectors are excluded because they are not identifiable as fixed facilities.
Why Travel Time Rather Than Distance?¶
A fixed kilometre radius is a poor proxy for agricultural supply sheds because the distance a farmer can practically cover in a given time varies enormously depending on local conditions:
- Road quality — a 50 km journey on a paved highway may take 45 minutes, while the same distance on an unpaved logging track or a wet-season dirt road may take 3 hours or be impassable entirely.
- Topography — mountainous terrain (e.g. coffee-growing highlands in Ethiopia or Colombia) sharply reduces effective speed even on established roads.
- River crossings and seasonal flooding — common in the Amazon basin and West Africa, these can double effective journey times or cut off access to whole catchment areas.
- Vehicle type — farmers in remote areas may travel by motorbike, ox-cart, or on foot for part of the journey, none of which follow straight-line distances.
Using travel time as the unit of measurement means the isochrone automatically contracts in areas with poor infrastructure and expands in areas with good road networks, producing a supply shed that reflects the real-world effort and cost a farmer faces — which is ultimately what determines whether they will deliver to a given facility or seek an alternative. A simple radius buffer makes no such distinction and will systematically overestimate supply sheds in rugged terrain and underestimate them along major highways.
How Travel Time is Computed¶
The isochrone is generated by modelling travel cost across the landscape from the facility outward, accounting for the real-world factors that determine how long a journey takes. Several independent geospatial layers are combined to produce a travel time estimate for every point in the surrounding area. The boundary where the accumulated travel time equals the specified limit becomes the supply shed polygon.
Road Network¶
The model uses a global road network with per-segment classification and surface type. Each road is assigned a speed appropriate for heavy commodity transport — the kind of vehicle a farmer or aggregator would use to deliver raw material:
| Road type | Approx. speed |
|---|---|
| Motorway / major highway | ~80 km/h |
| Trunk / primary road | ~60–65 km/h |
| Secondary / tertiary road | ~45–50 km/h |
| Residential / service road | ~24–30 km/h |
| Unpaved track | ~20 km/h |
Where road surface information is available, unpaved roads (dirt, gravel, mud) receive an additional speed reduction to reflect the substantially slower travel conditions common on wet-season rural tracks in tropical agricultural regions. Off-road movement — where no road exists — is modelled as much slower still, so the isochrone shape naturally follows the road network rather than expanding uniformly.
Where posted speed limits are available for a segment, they are used directly in place of the road-class estimate, improving accuracy on roads where the class-based speed would otherwise over- or under-state actual conditions. Roads that explicitly prohibit heavy goods vehicles are treated as impassable for commodity transport purposes.
Terrain and Slope¶
Elevation data is used to compute terrain gradient across the whole area. Slope has a non-linear effect on travel speed: a gentle incline has a modest impact, but steep terrain disproportionately slows loaded vehicles and is factored in accordingly. This is most consequential for:
- Coffee highlands (Ethiopia, Colombia, Honduras)
- Rubber and cocoa areas in hilly West Africa
- Timber concessions in montane Southeast Asia
Water Bodies¶
Rivers, lakes, and persistent floodplains are treated as impassable barriers, correctly fragmenting the supply shed in regions where water bodies physically interrupt the road network. This matters most in the Amazon basin, West African coastal regions, and river delta areas.
Ferry crossings
Where roads bridge a river, the crossing is modelled as normal road travel. Scheduled ferry crossings are also modelled as traversable routes at typical cargo ferry speeds (~15 km/h), allowing the supply shed to extend across waterways where ferry services operate. Unscheduled or informal river crossings (dugout canoes, seasonal pontoons) are not captured.
Land Cover¶
In areas away from roads, the type of terrain affects how quickly a vehicle can move. Dense forest, wetland, and mangrove areas slow off-road travel significantly and are assigned accordingly. In most agricultural supply chain contexts this layer plays a secondary role — the road network dominates — but it governs the shape of the supply shed boundary in sparsely roaded regions.
International Borders¶
Where the supply shed crosses a national border, a crossing time penalty is applied to reflect checkpoint and customs delays. Cross-border supply sheds are uncommon in EUDR contexts since the regulation is applied per country of production, but the penalty ensures the isochrone does not treat borders as frictionless.
Known Limitations¶
The model provides robust, globally consistent results for road-based commodity supply chains. The following factors are not yet accounted for and may affect accuracy in specific contexts:
| Factor | Where it matters most |
|---|---|
| Seasonal road conditions — wet-season accessibility (flooding, mud closures) is not yet modelled dynamically | West Africa, Amazon basin during rainy season |
| Informal river crossings — seasonal pontoons and dugout canoes are not captured; only mapped ferry routes are modelled | Remote areas of Congo Basin, Amazon |
Default Travel Times by Commodity¶
| Commodity | Default | Approx. distance | First industrial aggregation point |
|---|---|---|---|
| Cattle | 180 min | ~100–150 km | Slaughterhouse (state-licensed / SIE) |
| Soy | 120 min | ~30–80 km | Grain elevator / trading post |
| Palm oil | 90 min | ~50 km | Palm oil mill |
| Timber | 180 min | ~60–100 km | Sawmill |
| Rubber | 120 min | ~60–80 km | Crumb rubber / ribbed smoked sheet (RSS) factory |
| Cocoa | 90 min | ~22–40 km | Cooperative fermentation and drying centre |
| Coffee | 60 min | ~25–35 km | Cooperative wet mill (washing station) |
Distance approximations assume typical rural road speeds of ~40 km/h for agricultural roads and ~30 km/h for logging roads. Actual isochrone extents will vary based on local road network density.
Overriding defaults
You can always pass an explicit travel_times array to override the commodity default for a specific
run — for example "travel_times": [60, 120] to compare two shed sizes simultaneously. See the
Batch Supply Shed parameter reference.
Per-Commodity Rationale and Sources¶
Cattle — 180 min¶
State-licensed (SIE) slaughterhouses in the Brazilian Amazon are the most common first industrial aggregation point for cattle in deforestation-risk regions. Imazon's analysis of 128 Amazon slaughterhouses found that SIE plants purchase from farms within an average radius of 153 km; Gibbs et al. (2015) independently found that the 75th-percentile supplier distance to JBS (the world's largest meatpacker) is 145 km. At rural road speeds of 40–50 km/h, 153 km corresponds to approximately 180–230 minutes.
180 minutes is chosen as a conservative central estimate for state-licensed sourcing zones.
Limitation
Federal export-licensed (SIF) slaughterhouses extend their purchasing radius to ~360 km (~350+ min).
If your facility is a large SIF-licensed export plant, the default will underestimate the supply shed.
Use an explicit travel_times override in that case.
Sources: Imazon & ICV (2017); Gibbs et al. (2015), Conservation Letters 9(1): 65–76.
Soy — 120 min¶
Brazilian soy is predominantly grown in Mato Grosso, where farms sell to grain elevators and cooperative buying stations at paved road junctions before the commodity enters long-haul export logistics. Supply chain structure analysis (da Silva et al. 2020) places these first-buyer stations at approximately 30–80 km from farm gates, consistent with the dense elevator spacing in the Cerrado's grid-like road network.
120 minutes covers the full 30–80 km first-buyer range at typical Cerrado road speeds.
Note
Grain elevators are storage and aggregation facilities, not processing facilities in the transformative sense. They are nonetheless the first fixed, detectable commercial node where multi-farm commingling begins, and are therefore the appropriate EUDR traceability cut-off point.
Sources: da Silva et al. (2020), World Development 168: 106268; USDA AMS (2023), Brazil Soybean Transportation Guide.
Palm Oil — 90 min¶
Fresh Fruit Bunches (FFB) must reach a mill within 24–48 hours of harvest to prevent free fatty acid (FFA) content from spiking, which causes mill rejection. This biological constraint creates a hard ceiling on sourcing distance. Plasma and organised smallholders in Sumatra and Borneo typically deliver FFB to mills within 50–80 km (Cramb & McCarthy 2016). Recent analysis of 87 RSPO-certified mills places the operative mill supply shed at ~20 km for direct smallholder delivery (Meijaard et al. 2025).
90 minutes (~50 km) targets organised and plasma smallholder supply chains.
Independent smallholders
Independent smallholders often sell to roadside collectors at 5–20 km, who then transport to the mill. Collectors are non-industrial intermediaries and are excluded from the isochrone model. The palm oil mill itself is always the intended target.
Sources: Cramb & McCarthy (2016), The Oil Palm Complex, NUS Press; Meijaard et al. (2025), Communications Earth & Environment 6.
Timber — 180 min¶
The first industrial processing point for timber is the sawmill, where raw logs are cut into lumber. Log landings are non-processing staging areas and are excluded. FAO documentation on tropical timber logistics places concession-to-sawmill truck hauls at 50–150 km; Amazon-specific data from Holmes et al. (1997) shows averages of 80–150 km on the regional road network. At logging road speeds of 30–40 km/h, 100 km ≈ 150–200 minutes.
180 minutes falls within the documented tropical sawmill catchment range.
Limitation
Southeast Asian operations with high-value hardwoods sometimes haul logs 200+ km to sawmill
clusters, which would exceed this assumption. Consider using a custom travel_times value
for known long-haul concessions.
Sources: FAO Unasylva Nos. 65–67 (Timber Transportation in the Tropics); Holmes et al. (1997), Forest Ecology and Management 108(1–2): 17–34.
Rubber — 120 min¶
The first industrial aggregation point for rubber is the crumb rubber factory or ribbed smoked sheet (RSS) factory — large, fixed facilities where latex from multiple farms is coagulated, cleaned, and dried before export. These are distinct from downstream vulcanization plants (which manufacture finished products) and from non-industrial village dealers (which consolidate smallholder supply at the village level but are not detectable as fixed facilities).
ETRMA (2022) documents that crumb rubber factories source through dealer networks within 150–200 km total, with direct factory catchment estimated at ~60–80 km.
120 minutes (~60–80 km) represents the conservative direct catchment of the first detectable industrial processing facility.
Sources: ETRMA (2022), The Natural Rubber Supply Chain; IPB Journal (2015), Jurnal Manajemen dan Agribisnis 12(1); WWF (2021), Smallholder Solutions in the Rubber Sector.
Cocoa — 90 min¶
Cocoa cooperatives operate fermentation and drying centres — the first industrial processing step — that serve smallholder networks via "pisteurs" (collectors). Mighty Earth (2020) found that 80–90% of smallholders are within 22 km of their nearest cooperative; Ruf & Bini (2012) place the maximum practical pisteur circuit at 40 km before a new satellite aggregation point is established.
90 minutes accounts for poor rural road conditions in West Africa and covers the full pisteur circuit length.
Sources: Mighty Earth (2020); Ruf & Bini (2012).
Coffee — 60 min¶
The coffee cooperative wet mill (washing station) is the first industrial processing point for washed Arabica coffee. It is equivalent in structure to the cocoa cooperative model: a fixed facility with pulping machinery, fermentation tanks, and drying beds that serves a cluster of smallholder farms. The key biological constraint is that harvested cherries must be pulped within 6–12 hours of picking to avoid fermentation defects (TechnoServe 2022).
Field research in Ethiopia documents cooperative washing stations at 5–20 km from farm clusters in major producing regions (Minten et al. 2017). The UCDA similarly recommends stations be sited within 5–20 km for washed Arabica.
60 minutes (~25–35 km on highland roads) is a conservative upper bound for the cooperative wet mill catchment — non-industrial intermediate dealers are not modelled.
Dry/natural process coffee
Dry-processed (natural) coffee has no pulping time constraint, so the isochrone is not biologically bounded in the same way. The 60 min default is calibrated for washed Arabica cooperative supply chains. If your context is predominantly natural-process coffee, consider whether a larger travel time is appropriate.
Sources: Minten et al. (2017), Food Policy 83: 370–383; TechnoServe (2022), Coffee Wet Mill Processing Guide; Uganda Coffee Development Authority, Arabica Coffee Handbook.
Full Reference List¶
- Gibbs, H.K. et al. (2015). "Did Ranchers and Slaughterhouses Respond to Zero-Deforestation Agreements in the Brazilian Amazon?" Conservation Letters, 9(1): 65–76. DOI: 10.1111/conl.12175
- Imazon & Instituto Centro de Vida (2017). As zonas de influência dos frigoríficos no arco do desmatamento.
- da Silva, V.P. et al. (2020). "Not all supply chains are created equal." World Development, 168: 106268.
- USDA AMS (2023). Soybean Transportation Guide: Brazil 2023.
- Cramb, R. & McCarthy, J. (eds.) (2016). The Oil Palm Complex. NUS Press, Singapore.
- Meijaard, E. et al. (2025). "Uneven participation of independent and contract smallholders in certified palm oil mill markets in Indonesia." Communications Earth & Environment, 6.
- FAO. Unasylva: Timber Transportation in the Tropics, Issues 65, 66, 67.
- Holmes, T.P. et al. (1997). "Costs and benefits of forest management for timber production in eastern Amazonia." Forest Ecology and Management, 108(1–2): 17–34.
- ETRMA (2022). The Natural Rubber Supply Chain.
- IPB Journal (2015). "Supply Chain of Natural Rubber in Indonesia." Jurnal Manajemen dan Agribisnis, 12(1).
- WWF (2021). Exploring Smallholder Solutions in the Rubber Sector: Feasibility Study.
- Minten, B. et al. (2017). "Coffee Value Chains on the Move: Evidence in Ethiopia." Food Policy, 83: 370–383.
- TechnoServe (2022). Coffee Wet Mill Processing Guide.
- Uganda Coffee Development Authority (UCDA). Arabica Coffee Handbook.
- Mighty Earth (2020). [Cocoa smallholder distance study.]
- Ruf, F. & Bini, S. (2012). [Cocoa pisteur distance study.]