Where Week 7 left the network#

• Week 7’s finding: every network in this district is a capacity with a curve, and the curve gets ugly at the end.
• What we sized for: median demand, plus the margin the budget slider would tolerate.
• What we did not size for: two loads arriving in the same fortnight.
• Charles Perrow, Normal Accidents (1984): systems that are interactively complex and tightly coupled fail in combinations nobody designed for. Our district is both. One power shortfall stops the pumps; stopped pumps stop fire response.
• So a hazard week is a network week. Nothing new is introduced today. The same trunks are asked a harder question.

A hazard is an interaction, not an event#

• Burton, Kates and White, The Environment as Hazard (1978): a hazard is the interaction of a natural event with a human use system.
• A fire on empty forest is not a hazard. A fire on the forest that carries your only westward road and your power transmission line is a hazard.
• Which half of that interaction is in your control? Not the fire season.
• Therefore the design question is never “will it burn”. It is “what did we route through it, and what does the loss cost against what the corridor earns”.
• Consequence for language: in this studio, “the ridge burned” is not a finding. “We ran the only westward crossing through it in Week 5” is a finding. Burton, Kates and White’s own shorthand for the same move: the event is the weather, the loss is the design decision — a paraphrase, not their 1945 line about floods, since this week’s load case is fire.

Three families of hazard, three design responses#

You already have McHarg’s method#

• Ian McHarg, Design with Nature (1969): stack the constraint layers first, then site the development in what is left.
• The info views are that stack. Open them in this order over Halstead Reach: water, then wind, then ground pollution, then noise, then fire hazard.
• McHarg’s discovery, and yours: the layers disagree. The high ground on Verrall Ridge is dry and windy — good for the water problem, bad for the fire problem, and the wind carries the Pellow cells’ air pollution straight down it, toward the corridor most of you routed beneath it in Week 5.
• The intersection of all five acceptable zones is smaller than any student expects, and it is not where most of you zoned residential in Week 3.
• This is not a reason to move Week 3’s decision. It is a reason to know what Week 3 cost.

Pollution is a condition, not an incident#

• The simulation carries four pollution channels and they behave nothing alike: air pollution drifts with wind direction, noise falls off with distance, water pollution travels downstream only, ground pollution stays.
• Air pollution is a geometry problem: what is upwind of the residential blocks, at the prevailing direction, for most of the year.
• Noise is a zoning-adjacency problem, and it is the one your land value overlay reacts to fastest.
• Water pollution is an ordering problem on a single line: everything upstream is your inflow.
• Ground pollution is a memory problem, and it is the subject of slide 10.
• Anne Whiston Spirn, The Granite Garden (1984): urban air, water and soil are a designed system whether or not anyone designed them.

Ground pollution, and the decade of memory#

• Ground pollution accumulates under industry and generation, seeps toward groundwater, and decays slowly after the source is gone. Closing the source is the start of the wait, not the end of the problem.
• Practical consequence: the industrial cell you demolish in year 12 is not developable land in year 13. Your capstone plan cannot quietly assume it is.
• The land value overlay reads stained ground as low value for the whole decay period. Fiscal performance carries the cost long after the livability cost is fixed.
• Design response, in order of leverage: site it once, correctly; separate it with distance rather than with a park; and put the ground you intend to reuse somewhere else.

Climate as a seasonal load case#

• The row that fails first is Power, because its worst supply and its worst demand are the same fortnight.
• Luke Howard, The Climate of London (1818–1820), documented the urban heat island two centuries ago: the city makes its own climate. Tim Oke (1973) gave it a size relationship to the city’s own scale.

Which resilience did you buy?#

• C. S. Holling (1973) separates two things the word carries. Engineering resilience: how fast the system returns to normal. Ecological resilience: how much disturbance it absorbs before it settles into a different state altogether.
• A district that recovers its service levels in a week but loses a thousand residents permanently has excellent engineering resilience and failed.
• Aaron Wildavsky, Searching for Safety (1988), sets the matching pair of strategies. Anticipation: predict the specific hazard and prevent it. Resilience: build general capacity to absorb what you did not predict.
• Anticipation is cheaper when you are right. You are right about seasonal hazards and wrong about episodic ones.
• So: anticipate the winter, and hold redundancy for the fire.

The safe development paradox#

• White (1945) observed the pattern for floods; Raymond Burby (2006) named it the safe development paradox. Protection reduces frequent, small losses. In doing so it makes the protected ground attractive, so more is built on it, so the loss when protection is exceeded is larger than the losses protection prevented. The mechanism is general — this week’s load case runs it on fire, not flood.
• In this simulation the mechanism is visible and fast. Defend the ridge flank with a perimeter and response capacity, and the land value overlay there rises within a few years. Rising land value pulls higher-density redevelopment right up to the treeline. Higher density on defended ground is exactly the exposure you thought you had reduced.
• So “defend the ridge flank” is not the cautious option. It is a bet that your response capacity outlives the density your own protection invites.
• The honest version of the defend option therefore includes a density cap on the protected ground — which gives back part of the fiscal argument that justified defending it.

What the hazard layers cannot see#

• Every layer in today’s lecture is a physical layer. Water goes downhill, wind goes downwind, soil remembers.
• Susan Cutter and colleagues (2003) built a social vulnerability index because the same hazard produces very different losses across households at the same exposure: tenure, income, mobility, who can leave and who cannot.
• Ben Wisner and colleagues, At Risk (2nd edn, 2004), make the same point structurally: vulnerability is produced by social processes, and the physical event only reveals it.
• There is no info view for that. Not one of the five overlays we opened today distinguishes between the households on the ridge flank.
• That is next week’s lecture, and it is not a digression. Knowing which layers your model does not have is the same skill as knowing what the capacity curve does.