A 7-day, weather-driven modeled load and generation forecast for De Anza College (Cupertino, CEC CZ4) — a second campus alongside UC Davis. Predicted from a BigML load archetype, not observed.
This page renders a forward-looking 7-day forecast of campus electrical load and modeled solar generation for De Anza College — a second predictive site alongside the UC Davis campus that anchors TowerWatch. The forecast is produced entirely from a real NWS hourly weather forecast pushed through a BigML deepnet load archetype; solar output is modeled from clear-sky geometry.
Every number on this page is read live from the forecast JSON at load time, so re-running the forecast and re-staging the data updates the page with no code change.
net_kw = load_kw − gen_kw. Negative net (highlighted in green)
means modeled solar generation exceeds campus load — the campus would, in the model, be a net exporter for that hour.
Hourly modeled load, modeled clear-sky generation, and net load across the 156-hour NWS forecast window. Times are local (America/Los_Angeles).
De Anza College, Cupertino, California — the geographic anchor for the NWS forecast point.
A real NWS hourly weather forecast for the De Anza site is turned into model features, scored by a BigML deepnet trained on the De Anza load archetype to predict gross campus demand. Solar generation is modeled separately from clear-sky irradiance and a fixed nameplate.
Load (campus_demand_kw) — the real output of the BigML deepnet, trained on De Anza itself (1:1 anchor, no enrollment scaling).
Generation — modeled, not measured:
gen_kw = modeled_GHI(W/m²) × nameplate_kw / 1000 (STC) × availability_derate, capped at nameplate.
Nameplate 1500 kW DC × availability derate 0.71.
The following caveats are pulled verbatim from the forecast artifact's notes[]. This is a modeled, unvalidated forecast — read every absolute kW value with these constraints in mind.