Disaster Recovery & Protective Systems

Community Resilience Through Distributed Infrastructure
Cascadia Subduction Zone · Pacific Northwest · Mesh Networks · Early Warning · Solar Infrastructure
M9.0+ megathrust threat 326 years since last rupture 10–15% chance in 50 years 1,100 km fault length 39 DART buoys (US network) 50M+ residents in ShakeAlert zone ~500 yr average recurrence
The Pacific Northwest sits atop the Cascadia Subduction Zone — the most dangerous fault system in North America. A 1,100-kilometer megathrust running from Cape Mendocino to Vancouver Island, it last ruptured on January 26, 1700, generating a magnitude 9 earthquake and a transoceanic tsunami that struck Japan nine hours later. The geological record shows at least 19 great earthquakes (M8+) over the past 10,000 years. The question is not if, but when. Everything on this page exists because of that certainty.

Cascadia Subduction Zone

The M9.0+ Threat

Seismic

The Cascadia Subduction Zone extends 1,100 km from northern California to British Columbia, where the Juan de Fuca plate subducts beneath the North American plate at approximately 40 mm/year. USGS estimates place the probability of a full-margin rupture (M8.7–9.2) at 10–15% in the next 50 years. Including partial-segment ruptures (M8.0–8.6), the probability of a major event rises to approximately 37% in 50 years. The locked zone extends 60–150 km offshore, accumulating strain that will release in seconds.

USGS Fact Sheet 2025-3050 · PNSN Cascadia Sources · Recurrence: ~200–600 years (full margin)

January 26, 1700 — The Last Full Rupture

Seismic Historical

At approximately 9 PM local time, the entire Cascadia margin ruptured in a M8.7–9.2 earthquake. The evidence: Japanese historical records document an "orphan tsunami" — a tsunami with no local earthquake — striking the Sanriku coast on January 27–28, 1700 (Japanese calendar). Kenji Satake and colleagues used the tsunami arrival times and wave heights to back-calculate the source, pinpointing the date and magnitude. Ghost forests along the Washington and Oregon coasts — stands of dead western red cedar killed by sudden saltwater inundation — stopped growing after the 1699 growing season, confirming the timing through dendrochronology.

Satake et al. (1996) Nature 379:246 · Atwater et al. (2005) "The Orphan Tsunami of 1700" USGS

Liquefaction Zones in Puget Sound

Seismic

Loose, water-saturated alluvial soils throughout the Puget Sound lowlands are susceptible to liquefaction during sustained shaking. High-risk areas include the Duwamish industrial corridor in Seattle, Interbay, stream corridors, waterfront areas throughout the region, and fill zones in Everett and Tacoma. The Washington Geological Survey developed liquefaction susceptibility maps after the 2001 M6.8 Nisqually earthquake demonstrated the hazard: buildings tilted, underground pipes broke, and the Port of Olympia sustained significant damage — from a moderate event at 52 km depth. A shallow M9 would produce shaking 30–100 times stronger, sustained for 3–5 minutes.

WA DNR Geologic Hazard Maps · USGS Open-File Report 2007-1175 · PNSN Hazard Maps

Tsunami Inundation Zones

Seismic Warning

A full Cascadia rupture generates tsunami waves across the entire Pacific basin. The outer coast (Washington, Oregon) faces 15–30 minute arrival times with wave heights of 6–10 meters in many areas, potentially exceeding 20 meters in constricted inlets. The Strait of Juan de Fuca channels waves toward the San Juan Islands and inner waterways. Within Puget Sound, reflected and refracted waves arrive in 30–60 minutes with reduced but still dangerous amplitudes of 1–3 meters, complicated by the narrow, deep channels that amplify wave energy. Evacuation to high ground is the only reliable protection on the outer coast.

NOAA PMEL Tsunami Modeling · WA DNR Inundation Maps · PTWC coverage zone

ShakeAlert Early Warning System

Warning Seismic

ShakeAlert is the USGS-operated earthquake early warning system covering California, Oregon, and Washington — over 50 million residents. The system detects P-waves from seismometers, calculates magnitude and location, and issues alerts through the Wireless Emergency Alert (WEA) system and the MyShake smartphone app before slower, destructive S-waves and surface waves arrive. For Cascadia events, warning times vary significantly: coastal areas near the rupture zone may receive only seconds of warning, while inland cities like Seattle and Portland could get 30–90 seconds. Limitations: the system depends on seismometer density, cellular infrastructure, and internet connectivity — all of which may degrade in the event itself.

shakealert.org · USGS · Delivered via WEA + MyShake app · Operational since 2021 (PNW)

Mesh Networking for Disaster Communication

When a M9 Cascadia earthquake hits, conventional communication infrastructure fails within minutes. Cell towers lose power (battery backup: 4–8 hours), fiber backhaul severs at fault crossings, and network congestion renders surviving towers useless. Mesh networks — decentralized, self-healing, locally powered — are the engineering answer to catastrophic infrastructure failure.

LoRa / LoRaWAN

Mesh

Long Range (LoRa) radio operates in the unlicensed ISM band (915 MHz in North America, 868 MHz in Europe) using chirp spread spectrum modulation. Range: 2–15 km line-of-sight, 1–5 km in urban environments. Data rate: 0.3–50 kbps (low, but sufficient for text, GPS coordinates, and sensor telemetry). Power consumption: as low as 10–50 mA transmit, microamps in sleep mode. A single 18650 lithium cell can power a LoRa node for months in low-duty-cycle operation. Spreading factor (SF7–SF12) trades data rate for range: SF12 reaches farthest but at only 250 bps.

ISM 915 MHz · Semtech SX126x / LR11xx · License-free · Battery: months to years

Meshtastic — Open Source LoRa Mesh

Mesh Open Source

Meshtastic is open-source firmware that turns inexpensive LoRa radio modules (ESP32, nRF52, RP2040-based) into a decentralized mesh network. Features: encrypted text messaging, GPS position sharing, telemetry (temperature, humidity, battery voltage), and store-and-forward for offline nodes. Maximum 7 hops per message. Pairs with smartphones via Bluetooth for a user-friendly interface. No internet, no cellular, no infrastructure required. Hardware cost: $25–60 per node. A network of solar-powered Meshtastic nodes provides persistent, zero-infrastructure communication across an entire community.

meshtastic.org · GitHub · 7-hop max · AES-256 encryption · $25–60/node

WiFi Mesh (802.11s)

Mesh

IEEE 802.11s defines a mesh networking amendment to the WiFi standard. Higher bandwidth than LoRa (tens to hundreds of Mbps) but shorter range (50–200m) and significantly higher power draw (2–10W per node). Useful for local high-bandwidth needs: sharing images, video feeds from cameras, accessing cached web content. In a disaster mesh, WiFi nodes serve as local access points while LoRa handles long-range backbone communication. Power-hungry — requires solar or generator to sustain operation beyond battery backup.

2.4/5 GHz · 50–200m range · High bandwidth · Power: 2–10W per node

AREDN — Amateur Radio Emergency Data Network

Mesh Ham Radio

AREDN transforms commodity WiFi routers into a high-speed amateur radio mesh network operating on 900 MHz, 2.4 GHz, 3.4 GHz, and 5.8 GHz bands. Requires an amateur radio license (Technician class or higher). Achieves 60+ Mbps on 10 MHz channels over 80+ km links with directional antennas. Self-healing mesh topology (transitioning from OLSR to Babel routing as of firmware 3.25.x). Supports email, chat, VoIP, video conferencing, and document sharing. Key advantage over commercial alternatives: amateur radio operators can legally use higher power levels and dedicated spectrum allocations during emergencies.

arednmesh.org · Amateur license required · 60+ Mbps · 80+ km links

Protocol Comparison

Protocol Range Bandwidth Power Draw License Cost/Node Best For
LoRa / Meshtastic 2–15 km 0.3–50 kbps 10–50 mA TX None (ISM) $25–60 Text, GPS, sensors
WiFi Mesh (802.11s) 50–200 m 50–300 Mbps 2–10 W None $40–150 Local high-bandwidth
AREDN 1–80+ km 5–60+ Mbps 5–15 W Ham (Tech+) $50–200 Long-range backbone
Starlink Global (LEO) 50–200 Mbps 75–100 W Subscription $299+/mo Internet uplink (if power available)

Topology: Why Mesh Survives

Mesh

Star topology (cellular): all nodes connect to a central tower. Tower fails = all nodes lose connectivity. Single point of failure. Mesh topology: every node connects to multiple neighbors. Any node can route traffic for any other node. When a node fails, traffic automatically reroutes through surviving neighbors. This is why mesh survives earthquakes — there is no central point to destroy. The math: in a mesh of N nodes with average degree k, the network remains connected even after removing up to (k−1) nodes. Redundancy is structural, not added.

Graph theory: mesh connectivity = min vertex cut · Self-healing via routing protocol convergence

Solar-Powered Node: Power Budget

Solar Mesh

A reference design for a self-sustaining mesh node: Load: LoRa radio + microcontroller = ~1W average (5W peak transmit). Panel: 20W solar panel (PNW: ~3.5 peak sun hours/day in winter = 70 Wh/day generation). Battery: 100 Wh LiFePO4 (safe chemistry, 2000+ cycles). Autonomy: 100 Wh / 24W = ~4 days without sun. With even partial sun, indefinite operation. The 20W panel produces 3.5x the daily load even in December in Seattle (47.6°N). Enclosure: NEMA 4X weatherproof, pole-mounted, elevated above flood zone.

5W avg load · 20W panel · 100Wh LiFePO4 · 3+ days autonomy (no sun) · Indefinite (partial sun)

Early Warning Systems

Modern disaster preparedness depends on layered warning systems operating at different timescales: seconds (seismic P-waves), minutes (tsunami propagation), hours (weather), and days (space weather). Each layer uses different physics, different sensors, and different communication channels.

Seismic: ShakeAlert & Strong Motion Sensors

Warning Seismic

ShakeAlert detects P-waves (compressional, fast, low damage) traveling at ~6 km/s and issues warnings before S-waves (shear, slow, destructive) arrive at ~3.5 km/s. The system uses a network of ~1,700 seismometers across the West Coast. For a Cascadia event originating offshore, P-waves reach coastal sensors first; the system calculates magnitude within 5–10 seconds and pushes alerts via WEA and app. Inland cities gain valuable seconds: Seattle could receive 30–90 seconds of warning for a Cascadia event. Strong motion accelerometers (SMAs) supplement broadband seismometers for near-field recording where broadband instruments clip.

~1,700 sensors · P-wave @ 6 km/s vs S-wave @ 3.5 km/s · 5–10 sec to detection

Tsunami: DART Buoys & PTWC

Warning

NOAA operates 39 DART (Deep-ocean Assessment and Reporting of Tsunamis) buoy stations across the Pacific, Atlantic, and Caribbean. Each station consists of a seafloor bottom pressure recorder (BPR) at 1,000–6,000m depth and a surface buoy communicating via Iridium satellite. The BPR detects tsunami waves as small as 1 cm in 6,000m of water by measuring pressure changes at 15-second intervals. In standard mode: hourly reports. In event mode (triggered by seismic alert): 15-second data transmitted in real time. The Pacific Tsunami Warning Center (PTWC) in Honolulu, Hawaii covers the Pacific basin internationally, while the National Tsunami Warning Center (NTWC) in Palmer, Alaska covers US and Canadian coasts. Both integrate DART data with seismic data and tide gauge readings to issue tsunami warnings, watches, and advisories.

39 DART stations · 1 cm detection in 6 km depth · Iridium satellite uplink · PTWC Honolulu / NTWC Palmer

Weather: NWS, CWOP & CoCoRaHS

Warning Citizen Science

The National Weather Service operates NEXRAD Doppler radar (160 stations, 230 km range each) and launches radiosondes twice daily from 92 upper-air stations. Citizen networks amplify this: CWOP (Citizen Weather Observer Program) contributes ~11,000 personal weather stations feeding real-time data into NWS models via APRS. CoCoRaHS (Community Collaborative Rain, Hail & Snow Network) provides daily precipitation measurements from ~23,000 volunteers, critical for flood forecasting in data-sparse areas. These citizen networks demonstrate the power of distributed sensing — the same architecture that makes mesh networks resilient.

NEXRAD: 160 stations · CWOP: ~11,000 stations · CoCoRaHS: ~23,000 observers

Space Weather: DSCOVR, ACE & GOES

Warning

Solar storms (coronal mass ejections) can destroy power grid transformers, disable satellites, and disrupt GPS and radio communication. The defense: sentinel spacecraft at the Sun-Earth L1 Lagrange point, 1.5 million km sunward. DSCOVR (2015–present) carries the PlasMag instrument suite (fluxgate magnetometer, Faraday Cup ion detector, electron analyzer) measuring solar wind speed, density, and magnetic field orientation. Provides 15–60 minutes warning before CME impact. ACE (1997–present) continues as backup. GOES-16/17/18 carry magnetometers in geostationary orbit for real-time geomagnetic field monitoring. SWFO-L1 (Space Weather Follow-On) is the planned successor.

DSCOVR @ L1 · 15–60 min CME warning · GOES magnetometers · SWPC Boulder, CO

Multi-Hazard Integration

Warning Seismic

Real resilience requires integrating warning systems across hazard types. A Cascadia earthquake triggers a cascade: ground shaking (seconds) → landslides (seconds to minutes) → liquefaction (during shaking) → tsunami (minutes to hours) → dam failures (hours) → aftershocks (days to years). Each hazard has its own warning system, detection physics, and communication pathway. The integration challenge: a M9 earthquake may simultaneously disable the communication infrastructure needed to deliver tsunami warnings. This is why local, self-powered, mesh-networked warning systems matter — they operate independently of the infrastructure the earthquake destroys.

Cascading hazards: EQ → landslide → liquefaction → tsunami → dam failure → aftershock

Solar + Compute + Communications Stack

The concept: every rooftop solar installation is a potential node in a resilient distributed network. Panels generate power. Batteries store it. Compute hardware collects sensor data and routes messages. Radios connect nodes into a mesh. Each node is independently viable. Together, they form a system that survives what centralized infrastructure cannot.

Power Architecture

Solar

Solar panels → MPPT charge controller → LiFePO4 battery bank → DC bus (12V or 48V) → loads. LiFePO4 chemistry chosen for safety (no thermal runaway), cycle life (2,000–5,000 cycles at 80% DoD), and temperature tolerance (−20°C to 60°C). The DC bus directly powers compute and radio hardware — no inverter losses for DC loads. A 48V architecture matches telecom standards and minimizes wire losses. Sizing: 1 kWh battery bank + 200W panel sustains a full communication node (radio + compute + sensors) indefinitely at 47°N latitude, including PNW winter cloud cover.

LiFePO4 · MPPT · 48V DC bus · No inverter for DC loads · Telecom-standard voltage

Edge Compute

Solar Compute

Raspberry Pi 5 (5–10W) or Jetson Nano (5–15W) class single-board computers serve as edge nodes. Functions: sensor data collection (temperature, humidity, barometric pressure, seismic, air quality), data logging to local storage (SD/NVMe), mesh network routing, MQTT message broker, and cached content server. Local-first architecture: all data stored on-node first, synchronized to peers when connectivity allows. No cloud dependency. A Pi 5 with 8GB RAM runs Linux, sensor daemons, MQTT broker, and a local web server simultaneously within a 10W power envelope.

Raspberry Pi 5: 5–10W · Jetson Nano: 5–15W · Local storage · MQTT telemetry

Communication Stack

Mesh Solar

Layered radio architecture on each node: Layer 1 — LoRa (915 MHz): long range (2–15 km), low bandwidth (text, GPS, critical alerts), ultra-low power, always-on backbone. Layer 2 — WiFi (2.4/5 GHz): short range (50–200m), high bandwidth (images, cached content, local services), higher power, activated on demand. Layer 3 — Cellular/Starlink: internet uplink when available, used for data sync to outside world, highest power, opportunistic. The node gracefully degrades: if power drops, Layer 3 shuts down first, then Layer 2. Layer 1 (LoRa) runs last on milliwatts.

L1: LoRa (always-on) → L2: WiFi (on-demand) → L3: Internet (opportunistic)

Data Architecture: Local-First

Mesh

All sensor data is written to local storage before any network transmission. MQTT (Message Queuing Telemetry Transport) handles sensor telemetry: lightweight publish/subscribe protocol designed for constrained devices and unreliable networks. Data format: structured JSON with timestamps, sensor ID, location, and readings. Synchronization: when mesh connectivity exists, nodes exchange data using a conflict-free replicated data type (CRDT) strategy — last-write-wins for sensor readings, append-only for event logs. No single node is authoritative. Every node has a complete local view of its neighborhood.

MQTT pub/sub · JSON telemetry · CRDT sync · Append-only event logs · No cloud dependency

Resilience Mathematics

Solar Mesh

Each node is independently: self-sustaining (solar + battery = indefinite power), self-routing (mesh protocol discovers and maintains routes automatically), self-healing (node failure triggers automatic route recalculation; no human intervention). Network resilience scales with density: a grid of nodes spaced 1 km apart in a 10×10 km area (100 nodes) maintains connectivity even with 40% node failure, because average node degree ≥ 4. The system gets more resilient as nodes are added — the opposite of centralized infrastructure, which gets more fragile as it scales.

N nodes, k=4+ avg degree · Tolerates (k−1) adjacent failures · Antifragile at scale

Community Resilience Patterns

Technology without community organization is hardware on a shelf. These programs represent decades of research into how communities actually self-organize, prepare for, and recover from disasters.

CERT — Community Emergency Response Teams

Community

FEMA's Community Emergency Response Team program trains civilians in basic disaster response: light search and rescue, fire suppression, medical triage (START protocol), and team organization. The training recognizes a fundamental truth: in a catastrophic event, professional first responders are overwhelmed immediately. The 2001 Nisqually earthquake and 2014 Oso landslide in Washington demonstrated this — neighbors rescued neighbors in the critical first hours before organized response arrived. CERT creates structured capability at the neighborhood level, turning bystanders into first responders.

FEMA · 20-hour basic training · Fire suppression, SAR, triage, org · Active in all 50 states

Map Your Neighborhood (MYN)

Community

Developed by Dr. LuAn Johnson at the Washington State Emergency Management Division and endorsed by FEMA and the American Red Cross. MYN is a 2.5-hour program that organizes 15–20 households into a prepared block unit. Participants create a neighborhood map identifying: who has medical training, who has tools (chainsaw, generator, jack), who has special needs (mobility, oxygen, medication), and where gas/water shutoffs are located. Successfully implemented in 22+ states. The genius: it's small enough to complete in one evening, and the social bonds formed are themselves a resilience resource.

WA EMD · FEMA/Red Cross endorsed · 15–20 homes per unit · 22+ states · 2.5-hour program

Mutual Aid Networks

Community

After every major disaster, mutual aid networks emerge spontaneously: neighbors sharing food, tools, shelter, and labor without formal organization. Research shows these networks form faster and distribute resources more efficiently than centralized distribution in the critical first 72 hours. The pattern is consistent across cultures and disaster types: Kobe 1995, Christchurch 2011, Nepal 2015, Puerto Rico 2017. The research question is not whether mutual aid works — it demonstrably does — but how to design infrastructure (communication, logistics, storage) that amplifies natural mutual aid behavior rather than replacing it.

Self-organizing · Critical first 72 hours · Faster than centralized distribution

The Convergence Zone — Everett/Mukilteo Microclimate

Community Weather

The Puget Sound Convergence Zone is a meteorological phenomenon where marine air splits around the Olympic Mountains and reconverges over the northern Puget Sound lowlands, typically near Everett and Mukilteo. This creates intense, localized precipitation events — sometimes dropping 2–4 inches of rain or 6–12 inches of snow in a narrow band while areas 10 miles north or south receive nothing. For disaster planning, this means hyper-local weather hazards that regional forecasts miss. Distributed rain gauges (CoCoRaHS), personal weather stations (CWOP), and mesh-connected sensors provide the spatial resolution needed to detect convergence zone events in real time.

Olympic rain shadow convergence · Hyper-local precipitation · Distributed sensing required

Community-Owned Infrastructure

Community Solar

The cooperative model for shared resilience has deep roots in the Pacific Northwest: rural electric cooperatives (formed in the 1930s–40s under REA), community solar programs, and mutual water systems. The principle: infrastructure owned and operated by the people it serves is maintained for their benefit, not for profit extraction. Applied to disaster preparedness, this means community-owned mesh networks, shared solar+battery installations, neighborhood tool libraries, and collectively maintained emergency supply caches. The infrastructure IS the community — maintained by the same social bonds that activate during disaster response.

PNW cooperative tradition · Rural electric co-ops (1930s) · Community solar · Mutual water systems

Rosetta Connections

Disaster resilience is not an isolated discipline. It connects to every cluster in the research framework through shared principles: distributed architecture, redundancy, self-healing, and graceful degradation.

Ecology: Mycorrhizal Networks as Biological Mesh

ECO

Forest mycorrhizal networks ("wood wide web") are biological mesh networks. Fungal hyphae connect tree root systems, routing carbon, nitrogen, phosphorus, and chemical warning signals between trees. When a tree is attacked by insects, it sends chemical signals through the mycorrhizal network to neighboring trees, which upregulate their defenses before the insects arrive. The network topology mirrors engineered mesh: no central node, multiple redundant pathways, graceful degradation when connections are severed. Hub trees ("mother trees") serve a role analogous to high-degree mesh nodes — more connections, more routing capacity, more critical to network integrity.

Suzanne Simard (UBC) · Carbon/nutrient routing · Chemical signaling · Damage-tolerant topology

Electronics: Sensor Networks & Power Management

LED BPS

The electronics cluster connects through the fundamental engineering of sensor nodes: analog-to-digital conversion (seismometer output to digital samples), signal conditioning (filtering noise from P-wave detection), power management (MPPT solar charging, sleep/wake duty cycling), and RF design (impedance matching for LoRa antennas). A seismometer node IS an electronics project: operational amplifiers for signal conditioning, ADCs for digitization, microcontrollers for processing, and radio transceivers for communication. The 555 timer (T55) connects through the same timing and oscillator principles that drive LoRa's chirp spread spectrum modulation.

ADC · Signal conditioning · MPPT · RF design · Chirp spread spectrum

Infrastructure: Distributed Systems & Failover

SYS K8S

Every principle from distributed computing applies directly to disaster infrastructure. Redundancy: no single point of failure (mesh topology). Consensus: nodes agree on routing tables without a central authority (distributed hash tables, gossip protocols). Partition tolerance: the network continues functioning when split into disconnected segments (each segment operates independently). Eventual consistency: when partitions heal, data synchronizes (CRDT merge). The CAP theorem from distributed systems theory precisely describes the tradeoffs in disaster communication networks.

CAP theorem · Gossip protocols · DHT routing · Partition tolerance · Eventual consistency

CERN: WLCG as Disaster Recovery Model

CERN

CERN's Worldwide LHC Computing Grid is a working example of disaster-resilient distributed infrastructure at planetary scale: 170+ sites across 42 countries, federated authentication, automatic failover between tiers, and data replication across geography. The three-tier architecture (Tier 0 at CERN, 13 Tier 1 national centers, 160+ Tier 2 university sites) provides a blueprint for disaster communication networks: critical data replicated to multiple geographic locations, any tier can operate independently if disconnected, and the system degrades gracefully rather than failing catastrophically.

170+ sites · 42 countries · Federated auth · Geographic replication · Graceful degradation

Music: Polyrhythm as Multi-Frequency Sensing

MUS

A disaster warning system operates like a polyrhythmic composition: seismometers sample at 100–200 Hz, DART buoys report at 15-second intervals, weather stations update every 5 minutes, satellite passes occur every 90 minutes, and tidal gauges measure continuously. Each sensor operates at its own cadence, yet the combined signal reveals patterns no single frequency can capture. This is polyrhythm: independent rhythmic lines that create emergent structure through their interaction. The Fourier analysis that decomposes seismometer signals into frequency components is the same mathematics that analyzes musical timbre.

Multi-cadence sensing · Fourier analysis · Emergent pattern detection · Signal decomposition

Seed Papers — Initial Knowledge Base

Papers to seed the disaster resilience research corpus. Each represents a foundational contribution to seismology, mesh networking, community preparedness, or distributed infrastructure.

What Comes Next

This page documents the science and engineering of community disaster resilience in the Pacific Northwest. The Cascadia Subduction Zone defines the threat. Mesh networking, solar infrastructure, early warning systems, and community organization define the response. Every section connects to the broader research framework: ecology teaches us how biological networks survive damage, electronics teaches us how to build the sensors, infrastructure teaches us how to design distributed systems, and CERN shows us that planetary-scale federated computing already works. The PNW is both the laboratory and the subject — a region where the science of preparedness meets the urgency of geological certainty.

Next steps: embed seed papers in pgvector · build Meshtastic node reference design page · map Cascadia liquefaction zones with interactive visualization · connect CWOP/CoCoRaHS station data to weather-map.html · document solar power budget calculations for PNW latitude/climate