Sonoma County cellphone warnings scrambled by carriers, internal report says

Sonoma County cannot geographically target wireless emergency warnings as closely as it would like and the reach of those alerts appears to vary significantly by cellphone carrier, according to a new analysis of a widespread test conducted last month.

In a report released Friday, county officials said nearly ?3,700 survey responses to the Sept. 12 mobile alert trial showed some messages were received far beyond the five specific areas they were aimed.

The test also exposed key differences in how the county’s major cellphone carriers distribute emergency alerts. Such “inconsistent policies” will lead to “significant issues” for public officials trying to warn community members about an emergency, the report said.

“I’m glad we tested it, because at the end of the day, no matter how much they talk about how it works, until you do it, you don’t know,” said Supervisor David Rabbitt. “It would be great to continue to fine-tune it.”

During the Sept. 12 test of the federal Wireless Emergency Alerts system, the county and city of Santa Rosa sought to confine messages to specific areas: Guerneville, Glen Ellen and Kenwood, Healdsburg, Penngrove and Roseland. But survey results showed, for example, the Glen Ellen and Kenwood-targeted messages spilled over into Rohnert Park, the report said.

“It did kind of underscore the inability of the current technologies to keep messages tightly focused on the target areas,” said Christopher Godley, the county’s interim emergency manager.

Sonoma County conducted its test, believed to be the first of its kind on the West Coast, following outcry in the aftermath of last year’s wildfires that local government did not do more to warn people in harm’s way. The local test came three weeks before the federal government on Wednesday conducted an inaugural nationwide trial run of the same system.

The geographical limitations revealed in Friday’s report support early findings on the day of the test by county officials who monitored where people were filling out surveys linked to the cellphone message.

The initial findings also showed the reach of messages varying by cell carrier.

Verizon requires a cell tower to be located inside the boundaries of an alert area in order to send an emergency message, the report said. AT&T did not answer the county’s questions but because the carrier “had such a large bleed over,” the report concludes “any cell tower’s coverage that is within the boundaries of the alert issues the emergency message.”

“These differences cause significant issues for alert and warning officials when issuing alerts,” the report said. “With current policies of mobile carriers, it is almost impossible to target any area with any confidence.”

At the time of the test, numerous Guerneville-area residents reported they received no message at all even though they were supposed to, according to Supervisor Lynda Hopkins, who represents the community. Residents have continued to report that to her in the weeks since, Hopkins said Friday.

The challenges reaching Guerneville residents suggest to Hopkins the county needs to use multiple warning mechanisms, including sirens and “boots on the ground and people knocking on doors,” she said.

“This is not a perfect system,” Hopkins said of the pushed cellphone alerts. “We cannot rely on just that to save people in the event of an emergency.”

More than 800 survey respondents said they did not receive a cellphone message at all. Godley cautioned that some of them may have been outside the targeted areas and were never supposed to receive a test warning, but he could not provide a specific breakdown.

Test messages were sent in both English and Spanish. But fewer than ?160 survey respondents, or around 4 percent of the total, answered in Spanish, the report shows.

After testing the cellphone alerts Sept. 12, the county promptly conducted a separate trial of the federal Emergency Alert System, which broadcasts warnings through television and radio channels. It showed the “governance and operational procedures” of that system have not been maintained well enough at the state level, the report says.

It recommends forming a working group of all the county’s radio and TV broadcasters and cable providers to bolster the Emergency Alert System locally. County officials actually used live EAS codes - not test codes - during the Sept. 12 test, the report says. Yet one radio station waited until the end of a song to play the warning and another station had the wrong programming for it, delaying the message’s distribution by ?45 minutes, according to the report.

The tests appear to have benefited the county’s own SoCo Alert system. More than 50,000 phone numbers have now opted into the system, a 38 percent increase from early September, according to the report. Godley attributed the surge to the tests and publicity around them.

County officials conducted a test of the SoCoAlert system, which is also able to make reverse-911 calls to landline telephones, Sept. 10.

Local, state and federal officials must work with telecommunications providers to improve the systems’ reliability and effectiveness, the report concludes. The shortcomings in the federal warning systems, it noted, “significantly conflict with the public’s expectations for service.”

Godley plans to present the report to the Board of Supervisors next month.

You can reach Staff Writer J.D. Morris at 707-521-5337 or jd.morris@pressdemocrat.com. On Twitter @thejdmorris.