The warning system said moderate. Over half of Maricopa County's 608 heat-related deaths in 2024 occurred on days when the National Weather Service HeatRisk tool classified conditions at that level. Moderate. Below high. Below extreme.
That word should sit with anyone who trusts institutional risk categories to match what's actually happening on the ground. Because a study published last week in Geophysical Research Letters explains exactly why the classification failed. Researchers Di Cai and Gerrit Lohmann found that compound hot-dry extremes, heat and drought arriving simultaneously, have roughly doubled across Earth's land areas since preindustrial times. The physical mechanism reinforces itself: drought dries soil, reducing evaporative cooling, which amplifies surface temperatures further. Heat and drought together multiply.
Drought dries soil, eliminating evaporative cooling. Surface temperatures spike higher. Heat deepens drought. The feedback loop means simultaneous heat and drought produce nonlinear damage worse than the sum of each alone.
Maricopa County already knows this in its bones. The multiplication is written into the death records, the grid logs, the reservoir levels, and absent from every agency's emergency plan.
The room with the AC unit
Among people who died indoors from heat in Maricopa County in 2024, 88% had an air conditioning unit in their home. Seventy percent of those units weren't running.
The protective technology sat in the room, inert.
The county's surveillance doesn't distinguish why those units failed. Some broke from age or neglect. Some sat in apartments where tenants couldn't afford the electric bill. Some may have failed under grid strain. That ambiguity is itself part of the compound problem: a single-hazard data system can't parse a multi-hazard death. But trace the system-level forces pressing on that room and the compounding becomes visible.
Drought depletes the reservoirs feeding hydropower. Glen Canyon Dam operated at roughly 59% of nameplate capacity as of early 2023, its generating power diminished by the same low water driving Colorado River shortage declarations. By 2024, generation was on track for its second-lowest year since 1964. That lost hydropower is precisely the flexible, fast-ramping electricity the grid needs most during heat spikes. Meanwhile, heat drives demand the other direction: Arizona's three largest utilities set new peak demand records in three consecutive summers, with all three exceeding their own forecasts in August 2025.
Supply falling and demand spiking at the same time. The person inside that room had done what they were supposed to do. They had the unit. The system around it couldn't hold.
Two agencies, no shared map
Arizona runs its drought response and its heat response through entirely separate institutional architectures.
Drought governance flows through the Department of Water Resources and an Interagency Coordinating Group that has met twice yearly since a drought emergency was declared in 1999. Its mandate: water supply, Colorado River allocations, agricultural impacts, hydropower. Heat governance runs through the Department of Health Services, the Governor's Office of Resiliency, and a Chief Heat Officer appointed in 2024. Its mandate: cooling centers, mortality surveillance, wellness checks.
The drought system knows about energy. At its May 2025 meeting, a Western Area Power Administration vice president presented data on hydropower losses and replacement power costs reaching $1,000 per megawatt-hour during peak months. The heat system acknowledges drought conceptually. But heat activation protocols are triggered by NWS heat warnings alone. There is no drought severity co-trigger.
The clearest evidence these systems don't talk comes from the agencies themselves. Among ADHS's top recommendations in the state's first Extreme Heat Preparedness Plan:
"Unifying coordination of grassroots heat response activities with state agency efforts."
You don't call for unification when things are already unified.
What the instruments can't see
Maricopa County runs one of the most detailed heat death surveillance systems in the country. It tracks housing status, AC functionality, substance use, comorbidities, demographics, indoor versus outdoor location. It can tell you that 49% of the dead were experiencing homelessness. That African Americans and American Indians die at rates far exceeding their share of the county's population. That 57% of deaths involved substance use.
What it cannot track: whether drought conditions contributed to the death. Whether water access was a factor. Whether AC failure resulted from equipment breakdown or grid stress compounded by low hydropower.
The GRL study's core finding, that simultaneous heat and drought produce nonlinear damage, cannot be confirmed or denied by Maricopa's surveillance because the surveillance was designed to see one stressor at a time. Nobody built this gap on purpose. Every institution inherited a single-hazard framework and nobody was made responsible for the interaction between hazards.
The margin that keeps narrowing
Maricopa's preliminary 2025 death toll dropped to 427, a decline driven partly by expanded heat relief funded through the American Rescue Plan Act. That ARPA funding expires after the 2026 season. According to the state climatologist's report, Arizona received roughly 55–60% of normal precipitation in Water Year 2025. Post-2026 Colorado River operating rules remain unresolved after basin states missed their February deadline, with the Bureau of Reclamation preparing to impose terms that fall heaviest on Arizona's junior water rights.
On April 10, Maricopa County confirmed its first heat death of 2026. A historic March heatwave had already pushed Phoenix past 100 degrees. The Heat Relief Network doesn't activate until May 1.
Twenty-one days between a death and the system designed to prevent it. The warning said moderate. The AC was in the room. Someone died in the space between institutions that each thought someone else was watching.
- Arizona's snowpack entering summer: FERC's June 2025 update reported Arizona snowpack at 15% of median, severely limiting the hydroelectric generation available precisely when compound heat-drought conditions drive peak electricity demand.
- Grid reliability under compound stress: NERC's 2025 Long-Term Reliability Assessment found that compound annual growth rates for summer peak demand are the highest since tracking began in 1995, with thirteen of twenty-three North American assessment areas facing elevated or high resource adequacy risks over the next five years.
- Wildfire smoke as a third stressor: A February 2026 study in Science Advances found that chronic exposure to wildfire smoke PM2.5 contributes to an average of 24,100 deaths per year in the lower 48 states, with no evidence of a safe threshold and no EPA regulation of wildfire-sourced particulates.
- The post-2026 water cliff: The seven Colorado River basin states missed their February 2026 negotiation deadline, and the Bureau of Reclamation is now preparing to impose one of five alternative operating plans for January 2027, all of which place the heaviest cuts on Arizona's junior water rights.