Mass appraisal with GIS: how land is valued in the multipurpose cadastre

When a municipality's cadastre is updated, the figure that matters most to the citizen —and to revenue— is the appraised value. But valuing the parcels of a 50,000-parcel municipality one by one would be impossible in time and cost. That's why the cadastre doesn't use individual appraisal but mass appraisal: a statistical-spatial method that assigns values to thousands of parcels from models, not parcel-by-parcel visits. This note explains how it works, the role GIS plays, and how you ensure the result is fair and defensible.

What mass appraisal is (and how it differs from market appraisal)

A market appraisal values an individual parcel with a site visit, specific comparables and a detailed expert assessment. It's precise but expensive and slow. A mass appraisal estimates the value of many parcels simultaneously by applying statistical models to variables known for the entire parcel universe: location, area, use, construction, zone.

| | Market appraisal | Mass appraisal | |---|---|---| | Unit | One parcel | Thousands of parcels | | Method | Individual expert assessment | Statistical-spatial model | | Cost per parcel | High | Very low | | Use | Collateral, transactions | Cadastre, property tax | | Key input | The parcel's comparables | Market research by zones |

Mass appraisal doesn't seek the exact price of each parcel, but an equitable, consistent estimate for the whole municipality. Its success criterion isn't individual perfection but relative equity: that similar parcels get similar values and that the ladder of values reflects the market.

The centerpiece: homogeneous zones

Under IGAC Resolution 388 of 2020, the multipurpose cadastre builds land value on two zoning layers:

1. Physical Homogeneous Zones (ZHF). They group the territory by shared physical characteristics: land use, access roads, public utilities, topography, environmental conditions, planning regulations. They're polygons where land behaves similarly.
2. Geoeconomic Homogeneous Zones (ZHG). On top of the ZHF, a unit value per square meter of land is assigned, derived from market research (listings, transactions, point appraisals). Each ZHG groups parcels with similar land value.

This is where GIS is indispensable: homogeneous zones are spatial entities. They are delineated, edited, topologically validated and intersected with parcel geometry to assign each parcel the unit value of the zone it falls in. Without GIS, the process would be unmanageable at municipal scale.

How land value is built, step by step

1. Market research. Listings, transactions and georeferenced point appraisals —the "research points"— are gathered. The quality of the mass appraisal depends directly on how many points there are and how well distributed they are.
2. Delineation of Physical Homogeneous Zones. The territory is segmented in GIS by overlaying use, road, utility, regulation and environmental layers.
3. Value assignment and ZHG delineation. Each physical zone receives a unit land value from the research points it contains, adjusting for differences between point and zone.
4. Per-parcel land value. ZHG unit value × the parcel's land area. Adjustments for particular conditions enter here (frontage, depth, shape, corner, easements).
5. Construction value. Valued separately by typology, materials, condition and built area, using replacement-cost tables.
6. Total value. Land + construction, with final adjustments.

Where statistics and AI fit in

Traditionally, the zone unit value is computed with adjusted averages. But when there are enough research points, you can use regression models —including geographically weighted regression (GWR), which recognizes that each variable's effect changes across space— to estimate value as a function of explanatory variables: distance to the center, to main roads, to amenities, socioeconomic stratum, utility coverage.

Machine learning (random forest, gradient boosting) is making strong inroads into mass appraisal internationally, because it captures non-linear relationships between variables and price. But there's an important caveat: a more complex model is useless if it isn't explainable and defensible. In cadastre, an appraisal must be justifiable to the citizen who disputes it and to the judge if the dispute escalates. A "black box" model that's right on average but can't be explained is a problem, not a solution.

How quality is controlled

Mass appraisal isn't validated parcel by parcel; it's validated statistically with standard indicators from international practice (IAAO) that the Colombian cadastre has gradually adopted:

• Assessment/sales ratio studies. The appraisal is compared against real market prices in a sample. The goal is a median near 1.
• Coefficient of Dispersion (COD). Measures uniformity: how spread out the ratios are around the median. A low COD means horizontal equity (similar parcels, similar appraisals).
• Price-Related Differential (PRD). Detects vertical inequity: whether expensive parcels are systematically under-assessed relative to cheap ones, or vice versa.

These indicators turn a political question —"is the appraisal fair?"— into a verifiable measurement. A well-built mass appraisal isn't the one that yields the highest number for revenue, but the one that passes the equity tests.

Common mistakes

1. Too few research points, or poorly distributed. The model extrapolates over gaps and produces badly valued zones. It's the root cause of most disputes.
2. Homogeneous zones that are too coarse. If a ZHG mixes different market realities, it creates inequity within the zone.
3. Dirty parcel geometry. Land value multiplies by parcel area; if the geometry is wrong, the appraisal inherits the error.
4. Not documenting the method. An appraisal that can't be reconstructed can't be defended. Traceability is part of quality.

GeoSAT's role

Mass appraisal combines three capabilities at the core of what we do: structuring parcel data under LADM-COL, spatial analysis in GIS, and analytical models. Throughout our track record in cadastral projects we've supported land-valuation processes where the real challenge isn't the calculation, but the quality of the inputs: enough research points, well-delineated homogeneous zones, and reliable parcel geometry. Our cadastral platform Terraes is built to manage that data chain end to end.

If your municipality or agency is structuring a mass appraisal and wants to make sure it passes the equity tests, the best starting point is a diagnosis of the market inputs and the parcel geometry. Get in touch and we'll review it together.