From Land Potential to Household Access: How Living Region Built Its Food Model
Published: 2026-11-17
By LiberIT

TL;DR: Rural planners can use this history to examine food-system scenarios by separating land potential, household access, labour, prices, transition time and nutrition, then tracing the evidence behind each result. It follows Living Region’s April-to-August 2026 software development in the context of my longstanding food-growing and land-use work.
My work on these questions began long before Living Region’s food-model code. My biography traces permaculture experiments in a Don Mills ravine to around 2009. After moving to Owen Sound, I planted perennial food species on a small urban lot in 2015. A Lyis Food Forestry page followed in 2021, and I registered Lyis Forestry for the 2022 season. Those years of cultivation and study are the personal context for this software history.
The April-to-August interval describes development of software in the Living Region project, which models how households, land, energy and infrastructure interact during regional change. Its food-system module took shape around a specific modeling problem: an early Grey County scenario returned foodCoverage of 0.697 and a food-surplus result of minus 31.97 billion calories. In this model, foodCoverage compares modeled net food energy with modeled demand. Those figures describe one scenario’s inputs and prompted an examination of how the result was built. Model outputs are scenario results; a county implementation history and measured outcomes are separate lines of evidence.
During that software development, I asked for a municipality-level breakdown and a plain-language explanation of the deficit. The work moved through a series of linked questions. What population and land should anchor the model? How much labour does production require? How does a global supply shock reach a household? How does a food forest change its output over time? Can a modeled system provide a balanced diet as well as enough energy? The chronology matters because each question exposed a different part of the model.
Build a local baseline from land and labour
The first step was to replace arbitrary county scaling with population and land data at the lower-tier municipality level. A seed pass used 2021 Census population and land area while keeping settlement locations as map anchors. The distinction was important: the census supplied a real scale for the region, while the model generated household and land patches from that scale. Those generated patches were useful for simulation, with their synthetic status made clear in the project documentation.
On April 28, the repository began turning that baseline into a labour question. Hand-tool labour assumptions were aligned with productive land; the project added a labour-to-land report, post-harvest work, and comparisons for perennial staples and permaculture. A planner reading those outputs could ask whether a modeled food supply was compatible with the people-hours available to grow, harvest and preserve it. The early work also distinguished land’s present production potential from the labour and time needed to reach a different system.
Labour data required its own qualification. The project added an import path for Census occupation tables 98-10-0594-01, 0591-01 and 0592-01 to estimate agricultural occupations at the county or district level. Those direct occupation rows remained empty in the recorded import, so the model’s preferred basis stayed at an industry proxy: 4,721 agriculture-industry workers and an estimated 3,918.43 full-time equivalents. A proxy is a substitute measure used when a direct local measure is unavailable. Full-time equivalent, or FTE, converts work hours into the number of full-time jobs. These figures describe the model’s proxy basis; a verified Grey County farm-worker count requires county-level occupation records.
Selected code history: hand-tool labour baselines, labour-to-productive-land report, present potential and transition constraints, and occupation-table import path.
Separate production capacity from household food stress
The early food-gap output combined questions that planners need to inspect separately. One is physical: how much food energy could modeled land produce? Another concerns access: can households obtain food at the prices and through the supply channels represented in the scenario? A third concerns production labour. Treating these as separate ledgers makes the assumptions behind a shortfall easier to see.
The project added a food-balance decomposition that reports population, demand and production by municipality. It also moved food accounting from calories to joules, the standard unit of energy used across the project’s models of fuel, heat and electricity. This unit change made energy comparisons more consistent. Separate protein and other nutrient checks arrived later.
A further distinction came from the global input-loss scenario. I asked for the severeSystemicInputLoss33 scenario to represent a one-third global production shock through price pressure, availability and import affordability. A global loss assumption needs its own pathway into local food access; the model now treats local harvest changes as a separate assumption. The foodPriceMultiplierEstimate is labeled as a modeled price-pressure proxy. That label tells a reader what the number represents and which outside evidence would be needed for a local price forecast.
The scenario results make the difference concrete. Under one linearConservative fuel-shock case, the model produced a 6% fuel increase, a 1.26% food-price change, and modeled food insecurity of 26,595 people—1,369 above the trend comparison. Those values describe that case’s assumptions and calibration. They help compare pathways inside the model; current local prices and food access require observed local data.
Calibration became a recurring part of the work. Land-access tiers, source-quality levels, intake contracts and readiness links were added so a planner can see how an input entered the model and how suitable it is for the question at hand. A practical lesson followed: publish the input’s source and status beside the output. A population count, an industry proxy and a scenario assumption can each be useful, while carrying different levels of evidence.
Selected code history: food-gap and price-pressure reports, separate food-insecurity pressure from production labour, food-growing land-access tiers, and local calibration source checklist.
Model how production changes over time
By August, the project was asking a different question: how does the food supply change while a perennial system is being established? An annual snapshot can show a mature yield while hiding the years of planting and care required to reach it. On August 8, the repository added an evidence-based food-forest transition model, with outputs for establishment and production as the system matures. This shifted the planning view from a single land-potential number to a path across years.
The next milestone connected that path to household labour and age. On August 9, commits added ageing-in-place labour and livestock assumptions, then audited mature-system and multifunction-land assumptions. On August 10, the model separated dependent children’s food demand from their parents’ land capacity. That distinction lets a scenario represent household needs and available production separately: children contribute to household food demand, while parental land capacity remains its own production input.
For a planner, these changes make the timeline a core input. A transition scenario should show which foods provide the early bridge, when perennial production becomes available, how labour shifts across those years, and which household members contribute work. Those outputs remain scenario-dependent: local planting choices, land access, household composition and work patterns still need regional evidence.
Selected code history: food-forest transition model, ageing-in-place labour and livestock model, and separate child food demand from parental land capacity.
Test food output against a whole diet
On August 16, the model expanded beyond total food energy. Energy tells a planner how much modeled food can contribute to demand; nutrition checks ask what that food contains. The repository added protein as a separate constraint and an optional livestock layer, while retaining the plants-only result as a baseline. Later changes reconciled animal outputs with edible foods, scaled livestock systems to household demand, and joined food-succession results to the browser planner.
Here, “whole diet” means checking a range of food needs across the modeled year, alongside the energy total. The livestock work distinguishes edible output from animal live weight and records feed, water, housing, labour and processing as separate requirements. That matters because an animal can add useful food while also drawing on a property’s finite feed and labour. The model compares those demands with modeled household needs and makes the chosen feed assumptions visible.
The code’s nutrition boundary is also explicit. Total protein can be compared with a modeled requirement, while amino-acid adequacy, micronutrients, food safety, animal health and local biomass inventories remain separate evidence questions. This gives planners a way to locate the next data need. It also keeps a promising model result from being mistaken for a complete dietary assessment of a real household.
Selected code history: nutrient-aware livestock capacity, reconcile livestock output with food forms, integrate food succession into whole-diet nutrition, and reconcile whole-diet production with the browser planner.
Put labour on a timeline and expose open inputs
The August 16 livestock milestone added task-based animal labour. On August 24, the project added a time-aware food-production labour ledger. Moving from one total to a task-based ledger, the model separates establishment work from recurring seasonal work and tracks categories such as annual crops, perennial food forests, livestock, harvesting, preservation and system maintenance. The project documentation describes checkpoints from Year 1 through Year 10 and the mature state.
That shift lets a planner ask when work peaks, which tasks require different skills, and how many people-hours a proposed system needs across the year. It also gives residents and community groups a way to compare a food plan with the hours households can actually contribute, keeping labour capacity explicit in the scenario.
The final August commit exposed unresolved support-labour inputs. Site acquisition, access roads, water systems, fencing and initial earthworks sit outside the food-production ledger unless local records provide them. The code records these as open work. For planning, an open input deserves a named data task; this keeps the effort visible for estimation.
Selected code history: task-based livestock labour, time-aware food-production labour, and expose unresolved support-labour inputs.
A review sequence rural planners can reuse
The project’s history suggests a practical review sequence for a regional food model:
- Set the boundary. Name the region, population, land area, year and scenario. Keep measured census inputs distinct from generated households and land patches.
- Describe access to land. Record which land can grow food, the source for each area estimate, and the evidence quality attached to it.
- Separate food questions. Track physical food production, household demand, food access and price pressure as related but distinct quantities.
- Draw the transition timeline. Show establishment, early production and mature production, with labour assigned to seasons and years.
- Check food form and nutrition. Compare energy and tracked nutrients with household requirements, while showing feed, storage, labour and external inputs.
- Label each shock. State whether a scenario changes local harvest, fuel, fertilizer, prices or import affordability. Keep each pathway and its assumptions visible.
- List open evidence needs. Name missing local records and the decisions they affect so the next modeling pass can improve the baseline.
The value of this sequence is clarity. A planner can explain which number comes from a census table, which comes from a proxy, which comes from a scenario, and which still needs local measurement. The Living Region model developed toward that kind of audit trail between April and August 2026: first defining a population and land baseline, then separating household pressure from production, then adding time, nutrition and labour detail.
This history documents an evolving model whose outputs show what the recorded assumptions imply. A locally grounded plan still needs current evidence about land access, crop production, work patterns, household needs, market prices and supporting infrastructure. That boundary is useful: it turns each model result into a question planners can answer with the next local dataset or community conversation.
The full project history and selected milestones are available in the Living Region repository.