It’s axiomatic that as pilots, we don’t trust weather forecasts. A pilot and his wife recently died in a crash partially caused by unexpectedly bad weather. But they weren’t led astray by a lousy forecast—it was the official weather observations that were seriously wrong. We have lots of data that show us what happened. It’s harder, however, to explain what didn’t happen.
The 72-year-old had been a private pilot for 10 years, logging a total of 1,600 hours. More than a thousand of them were in his 2005 SR22, registration N917SR. It was nicely equipped with several multi-function displays, two Garmin GNS 430 GPS units, and an Avidyne DFC90 autopilot. The plan was to fly the 310-horsepower single-engine plane from Craig-Moffat Airport (KCAG) in Colorado to Joslin Field-Magic Valley Regional Airport (KTWF), in Twin Falls, Idaho. The straight-line distance is 336 nautical miles. It was the first leg of a trip they’d made quarterly for several years, from their home in Craig up to Canada for medical appointments. His wife was about 80% blind from retinitis pigmentosa, a group of genetic disorders involving a breakdown and loss of cells in the retina. A Canadian doctor offered promising treatments. One of their daughters told a newspaper that for her father, “this was a way he could show up for her. She enjoyed travel, new places and a treatment that told her there’s hope.”
N917SR departed KCAG at 2:30 p.m., Feb. 15, 2019, for the visual flight rules (VFR) flight. Contacting Denver center air traffic control (ATC) for flight following, the pilot stated they were VFR en route direct to KTWF, “although it looks like I’m gonna have to go quite a ways south of direct because of convection.” There was a cold front along the direct line, with an active mix of clouds and snow advancing over the whole region. N917SR headed approximately west at 17,500 feet. As a visual rules-only pilot, he couldn’t go any higher, as flights higher than 18,000 feet are reserved for instrument flight rules (IFR). At 3:13 p.m., the pilot told the Salt Lake Center controller that he was still heading west to avoid some convection and would “hopefully turn back north” in 50 miles. At 3:51 p.m., ATC suggested that to avoid IFR weather, he could navigate direct to the Ely VOR and then north on course. The pilot reported they were going visual: “I hadn’t planned to go as far west as Ely, but if that’s what I have to do, I can.”
Ninety minutes into the flight, the prospect of safe VFR flight to KTWF seemed to dim. Mountains below, IFR altitudes above, layers of clouds, areas of snow, some blocks of restricted airspace. And the clouds were closing in. “We might have to go underneath the deck here in about 50 miles; right now, we’re in the clear, but it’s just too high a buildup off to the west to clear it,” the pilot told ATC. As he slowly descended to 10,500 feet to stay VFR, controllers started to lose N917SR on their radar scopes. They suggested Ely’s Yelland Field (KELY) airport as a divert option, 230 heading and 75 miles. Salt Lake Center reported precipitation showing on its radar to the north, saying, “the weather observation at KELY is showing a ceiling of 5,000 and overcast, broken at 6,500 as well.” It seemed like a very reasonable proposal. The pilot replied, “Roger that, thank you, we are heading to Ely, gonna put down there.”
What neither the controller nor the pilot knew was that the Ely Automated Surface Observing System (ASOS) weather station was broken. But the National Weather Service (NWS) knew it wasn’t working properly.
Local pilots had been complaining for weeks. The visibility value was sometimes missing, sometimes seemingly correct, and sometimes reporting 9 miles when the actual visibility was considerably less. The airport manager had reported these issues to the NWS. An Ely FBO manager told the National Transportation Safety Board (NTSB) of a landing jet pilot weeks earlier being very upset the ASOS was reporting 9 miles when the actual visibility was “1¼ at best.” That pilot reported the issue to Flight Service and Salt Lake Center ATC.
The KELY ASOS is owned and maintained by the NWS Weather Forecast Office in Elko, Nevada. An NTSB review of NWS records revealed that a “trouble ticket” for the ASOS at KELY was opened 10 days before the accident. The “priority 1” ticket advised of unreliable wind and visibility reporting. The next day, an entry in the maintenance shift log indicated that NWS technicians would have to visit the ASOS to determine visibility sensor accuracy. On Feb. 11, a maintenance shift log stated that technicians would be visiting Ely the following week to investigate the visibility reporting issue. A part was also ordered, which arrived in Elko on Feb. 14. The next day, a technician attempted to drive with the part to Ely, but he encountered bad winter weather and returned to the office. It was the day of the accident.
As the technician made his abortive attempt to drive to Ely, Cirrus N917SR was approaching the airport. At 3:53 p.m., the ASOS reported METAR 17014G22KT 10SM BKN050 OVC065 02/M04 A2967, the coded weather observation indicating that the visibility was 10 miles, which is what ATC relayed to the pilot. Over the next hour, the ASOS detected a wind shift, the start of light snow and lower cloud layers, but the reported visibility stayed at 9 or 10 miles. In reality, as one might suspect from increasingly snowy conditions, the weather was worse. A lot worse. Experienced local pilots estimate the visibility went from 4 or 5 miles to less than a ½ mile with frontal passage. Heavy snow was falling.
Descending out of radar contact, ATC terminated radar services and told the pilot to squawk VFR. We know what happened next from data saved by the onboard avionics. Starting at about 9,000 feet MSL, less than 3,000 feet above the Ely airport elevation of 6,259 feet, the pilot circled around several times, looking for the runway. He made a descending circle to the right, then turned left and right. If the weather had been as reported, he should have been able to see the field. But the weather was not as reported. A witness just east of KELY reported hearing an airplane flying low, noting the weather was very bad with clouds at tree-top level.
A second witness, a pilot since 1977, was monitoring the airport’s UNICOM frequency at home in his kitchen and heard the clicks of someone activating the pilot-controlled lighting. He estimates the visibility was about ¼ mile in snow. Then a voice transmission: “Are the runway lights on? I can see the runway.” About a minute later, he heard, “I’d like to land, but I cannot see the runway.” There were no further transmissions.
At 5:30 p.m., 3 miles from the airport, the SR22 came out of the clouds at 210 knots in a 6,400-feet-per-minute descent. The disorientated pilot had lost control. The aircraft immediately impacted snow-covered rocky terrain. The pilot, his wife and their dog died instantly in the crash.
After the accident, the NTSB determined the probable cause to be the “pilot’s decision to continue the visual flight rules flight into instrument meteorological conditions and icing conditions.” The pilot was not instrument rated, and the aircraft was not certified for flight in icing conditions. Visual flight into IFR conditions is one of the most common ways pilots lose control and crash. Frustratingly, this pilot was being prudent, tracking well off the direct course to avoid weather and listening to ATC suggestions, and he made the decision to divert and land at an airport reporting good VFR conditions. The NTSB correctly cites the inaccurate weather station as a contributing cause.
The reason the suspect ASOS wasn’t taken offline until it was checked and fixed isn’t given in the report. NWS policy states “because of the impact on the operation and safety!no data is preferred to bad data.” Missing is better than misleading. Why this didn’t happen here is troubling for all of us who use automated weather reporting.
The other thing that didn’t happen lay in the snow in the wreckage. The Cirrus Airframe Parachute System (CAPS), which can lower the entire plane, occupants and all, safely to the ground, was found unused. The rocket motor was still in the launch tube with its frangible link intact, the parachute packed. The NTSB report states, “no anomalies were noted with the CAPS system.”
It’s relatively easy not to trust a weather forecast. We’re all taught not to believe the “weather guessers.” It’s much harder not to believe a weather observation, which is supposed to accurately reflect the current conditions as measured by the automated system. When VFR cruising in a Champ or IFR landing in an Airbus, we must accept what our eyes tell us. We must accept and handle the sometimes-surprising reality in front of us. If lost in the clouds, climb, communicate, confess. If there are no runway lights at landing minimums, go around. In dire circumstances, we may have a parachute to deploy. The pilot, clearly expecting a different view of the world as he headed into Ely, did none of these things.
Read our previous After the Accident article, “No Red Line: The Tragic Crash Of A Mustang II in Georgia.”
