Yvynation · walkthrough open in the app →
Protected area · fire · step by step

How to read a protected area, one layer at a time

Pick a park from Canada's register of protected and conserved areas, and Yvynation rings it with three bands of neighbourhood and reads fifteen crop inventories, twenty-five years of forest loss and fire, and a second, forest-made map inside each. This walkthrough opens Okanagan Mountain Park, on the east shore of Okanagan Lake, where the fire of August 2003 started.

Loading the example park's analysis…

—measured from the CPCAD boundary measured
—of the park burned in its worst fire year
—of tree cover lost in the park since 2001
—change in coniferous forest, crop inventory

measured computed from satellite imagery or from the geometry itself. registered as the register publishes it. estimated a modelled product, with a larger error than the land-use map. missing a layer that did not respond this time.

1

Pick the park from the national register

The Canada page reads the Canadian Protected and Conserved Areas Database (CPCAD), published by Environment and Climate Change Canada: parks, ecological reserves, conservancies and privately owned conservation lands, each with its boundary, category and year. A name search, a click on the map or a link opens one.

In the app: the search box on the Canada page, a click on a park, or /canada?area= with the name as CPCAD spells it. A park made of many polygons can be analysed part by part.

2

Three bands around the boundary

A park's numbers mean more next to its surroundings. Every analysis also runs in three bands outside the boundary: up to 500 m, 500 m to 2 km, and 2 to 5 km. None overlaps another or the park, so no hectare is counted twice.

In the app: the zone buttons above every chart, and the rings on the map.

3

Fifteen years of cover inside the park

Agriculture and Agri-Food Canada's Annual Crop Inventory classifies every 30 m pixel each year, forest and grassland as well as crops. Each column is one year inside the park, from 2011 to the latest.

In the app: the Cover tab, with the year selector driving the map.

4

The park against its neighbourhood

The share of each zone that is natural vegetation: forest, wetland, grassland and shrub. Inside a park it should stay high, but natural vegetation is not the same as forest: after a fire, conifer turns to shrub and grass and still counts.

In the app: the Cover tab with the % switch, changing zones.

5

What became what

The flow diagram says what each pixel was in the first year and what it is in the last. The pale band joining a class to itself is what stayed the same; the others are the change.

In the app: the Transitions tab, where you choose the two years.

6

Tree cover lost, year by year

Hansen/GLAD maps the 30 m pixels that lost their tree cover each year since 2001, from any cause: logging, pests, building, fire. For a protected area the app would split the loss at the year it was established; a park created before 2001 has the whole series after protection.

In the app: the Forest tab, which also shows stand age from the national forest inventory (NTEMS).

7

The fire, seen from orbit

The fire record comes from NASA's MODIS (MCD64A1), month by month since 2001, in 500 m pixels of about 21 ha. Coarse for a parcel, but a park-sized fire is exactly what it was built to see. Each column is the share of the zone that burned that year.

In the app: the Fire tab, with the years with fire and the last fire for each zone.

8

Two maps, two answers

The crop inventory is built to see farms; VLCE2, from the Canadian Forest Service, is built to see forests. Yvynation sets one against the other, grouped into the same broad classes, pixel by pixel. In a park, the forest map is the better judge of what is forest, and the crop inventory of what is not.

In the app: the Validation tab, with the agreement matrix for each zone.

9

Where every number came from

Each analysis keeps its source, scale and time of calculation, and the export carries them with the numbers.

In the app: the download button: a spreadsheet with one sheet per table, and the PDF report.

yvynation.umaterra.com.br/canada
Loading…measured
Loading the example park…

What Yvynation does not say

  • It does not judge park management. The satellite shows cover and fire; whether a burn was suppressed, prescribed or left to run is BC Parks' and BC Wildfire Service's record, not this one.
  • It does not say why the forest changed. Fire, beetles and logging all read as loss; the cause comes from the fire and forest records.
  • The bands are not a legal buffer. They are comparison rings, the same for every area.
  • It does not analyse Indigenous lands. Indigenous-governed and Indigenous-owned areas are left out of the data entirely, following the First Nations principles of ownership, control, access and possession (OCAP®).
  • It cannot see below 30 m, or 500 m for fire. Small burns and thinning under the canopy escape both.

Open in Yvynation → The same walkthrough on a Brazilian reserve

More examples: Okanagan Mountain Park, this page's example · Myra-Bellevue Park · Kalamalka Lake Park · Fintry Protected Area. Change the area in the address (?area=…, with the name as CPCAD spells it, British Columbia only on this page).

Other Yvynation walkthroughs, on Brazilian examples because the Canada page has no drawing or batch: an Indigenous land · a drawn area · many areas in one batch. Versão em português (com uma reserva do Brasil).

Notes

  1. The three UmaTerra tools differ in who draws the boundary they analyse: the state, the analyst or the landholder. A protected area's boundary is the state's: official and enumerable, so areas can be compared across the country on equal terms. Indigenous-governed lands are excluded from the Canada data by design, not by omission. Biondo et al., Registry, coordinate, cadastre (manuscript in preparation, 2026).
  2. The rings are exclusive: each starts where the last ends and leaves the area itself out. Even so, they are a comparison region, not a counterfactual: protected areas tend to be placed on land that was less threatened to begin with (Joppa & Pfaff, 2009), so the screen compares, but does not credit the difference to the boundary. Biondo et al., Registry, coordinate, cadastre (manuscript in preparation, 2026).
  3. Hansen's loss layer records canopy loss from any cause (Curtis et al., 2018), and a year rather than a date: a pixel is booked in the first year its loss shows in the annual Landsat composite. A fire in late August can therefore spread its loss over that year and the next. Hansen et al. (2013), Science 342, 850–853.
  4. Every area on this page is a pixel count times the true pixel area, measured on the same grid as the count. The crop inventory counts 900 m² Albers pixels; measured on a lat/long grid instead, Kelowna's pixel would read 580 m² and every area a third short, which is what this page caught and fixed in October 2026. Biondo et al., Registry, coordinate, cadastre (manuscript in preparation, 2026).

Built with Reflex