Peterborough data projectHow big is Little Lake? 604 square kilometres, says the draft
The County of Peterborough republishes the Ontario Hydro Network's waterbody layer, clipped to the county, as open data: 12,073 lakes, ponds and river sections with their areas. We asked an AI-style draft to convert three lake areas, then checked it. The code, conversions and counts are ours, from the layer as downloaded on 23 October 2026.
Three lakes: area in the file and our corrected conversion (our run, 23 October 2026; the file's areas are in square metres, as the layer's coordinates are)| Lake | Area in the file (m²) | Draft says | Corrected (ours) |
|---|
| Little Lake, in the city | 604,381 | 604.4 km² | 0.60 km², 60.4 ha |
| Chemong Lake | 16,471,699 | 16,471.7 km² | 16.47 km², 1,647.2 ha |
| Rice Lake | 92,363,066 | 92,363.1 km² | 92.36 km², 9,236.3 ha |
The draft's mistake is a classic: it divides square metres by 1,000 to get square kilometres and by 100 to get hectares, treating an area like a length. The right factors are 1,000,000 and 10,000. Nothing crashes, and the numbers even look precise. A learner who did not already know roughly how big Little Lake is would have no reason to doubt 604.4.
The program and its two checks:# Peterborough County waterbodies: an AI-style unit conversion, and the checks that catch it (our teaching example)
import glob, json
rows = []
for page in sorted(glob.glob("wb-*.json")):
rows += [f["attributes"] for f in json.load(open(page, encoding="utf-8"))["features"]]
def find(name):
return [r for r in rows if r["OFFICIAL_NAME_LABEL"] == name and r["WATERBODY_TYPE"] == "Lake"]
# The draft: areas are in square metres, so "divide by 1000 for km2 and by 100 for hectares".
def draft(m2):
return {"km2": m2 / 1000, "ha": m2 / 100}
# The fix: 1 km2 = 1,000,000 m2 and 1 ha = 10,000 m2.
def fixed(m2):
return {"km2": m2 / 1_000_000, "ha": m2 / 10_000}
def consistent(a):
"""Human check 1: 1 km2 is exactly 100 ha, whatever the input."""
return abs(a["ha"] - 100 * a["km2"]) < 1e-6 * max(1, a["ha"])
for name in ("Little Lake", "Chemong Lake", "Rice Lake"):
lake = max(find(name), key=lambda r: r["SYSTEM_CALCULATED_AREA"])
m2 = lake["SYSTEM_CALCULATED_AREA"]
d, f = draft(m2), fixed(m2)
print(name, "| draft consistent:", consistent(d), "| fixed", round(f["km2"], 2), "km2", round(f["ha"], 1), "ha")
# Human check 2: the file has two area columns. Where do they disagree by more than 1 percent?
odd = [r for r in rows if abs(r["Shape__Area"] - r["SYSTEM_CALCULATED_AREA"]) > 0.01 * r["SYSTEM_CALCULATED_AREA"]]
print("area columns disagree on", len(odd), "of", len(rows), "rows")
Check 1: units agree with each other
One square kilometre is exactly one hundred hectares. The draft's two outputs break that rule for every lake, so the check fails without anyone knowing the true area. Internal consistency is the cheapest test there is.
Check 2: two columns, one quantity
The layer carries both Shape__Area and SYSTEM_CALCULATED_AREA. They agree on all but 4 of 12,073 rows. On Crowe Lake the first is 0.835 of the second, and on Burnt River 0.735.
What we think, and what we know
Our guess is that those 4 polygons were cut at the county boundary, which would change one area but not the province's original figure. The snippet does say the layer is clipped to the county. We label this a guess and choose the column per question.
Names are rarer than water
Only 483 of the 12,073 polygons carry an official name, and most are ponds (6,066) or lakes (5,644). An agent asked to list Peterborough County lakes by name will miss thousands of unnamed ones, and the Otonabee River alone is split into 42 named polygons. Counting before summarising is the habit that keeps an answer honest.
This is county and provincial data, not the City of Peterborough's: the City's own portal terms allow use of unmodified datasets only, and its map items ask commercial users to arrange a paid agreement, so we chose the County layer, which covers the city's waters too. Used under the Open Government Licence - Ontario, which grants use including for commercial purposes (licence, County of Peterborough layer). Contains information licensed under the Open Government Licence - Ontario. Neither the County nor the province endorses this page.
