Reading a hazard tree: the defects that put a tree into the span
No CPUC rule lists the defects an inspector must look for. GO 95 Rule 31.1 requires accepted good practice for particulars the order leaves unspecified, so the defect list is a choice the utility's procedure records. The published lists run to seven families, four of them carrying a measurable threshold. The Zogg Fire gray pine carried four at once.
A gray pine stood about 60 feet north of a PG&E 12 kV span on Zogg Mine Road in Igo, uphill of the conductors. It was 105.5 feet tall and 24.5 inches through at 54 inches above grade. It leaned 23 degrees from vertical, toward the wire.
A cavity roughly 4 feet high and 3 feet wide opened on the uphill face of the trunk and ran 14 to 15 inches into it. There were no roots on that side of the tree at all. In PG&E's own aerial photograph from July 2019, the crown is full and green.
On September 27, 2020, at about 1442 hours, the tree went over downhill. Branches roughly 67 feet up the trunk hit the conductors and arced. The Zogg Fire burned 56,338 acres, destroyed 204 structures and killed four people.
Four defects, each with a name and a measurement, on one stem. Whether a tree is a hazard tree at all is a prior question about reach, target and weather, and it is settled before any of this matters. What follows is the other half of the file: what each defect is called, what it has to be measured to, and which way it puts the tree into the span.
GO 95 names no defect, and Rule 31.1 says where to look instead
An inspector looking for the list of defects in the Commission's rules comes back empty. GO 95 Rule 35 uses three words for condition, and makes the company's actual knowledge of them the trigger. The words are dead, rotten and diseased. Public Resources Code section 4293 adds its own vocabulary of condition in its trailing paragraph. Both stop at the adjective. The test, the dimension and the way to tell one condition from another in the field are left open.
Rule 31.1 is what fills the hole, and the Safety and Enforcement Division reprinted the operative paragraph as Attachment F to the Zogg report:
For all particulars not specified in General Order 95, a supply or communications company is in compliance with this rule if it designs, constructs and maintains a facility in accordance with accepted good practice for the intended use and known local conditions.
The defect list is a particular not specified in General Order 95. So the rule that binds a company on defects is a rule that points outside itself, at accepted good practice, and leaves the company to say which practice it accepted. That is the whole regulatory answer, and it has one operational consequence. The defect vocabulary a program uses has to be written down somewhere the company can produce. Rule 31.1 is also the rule SED cites when a utility fails to follow its own procedures.
Rule 31.1 is the same rule that carries the general maintenance duty:
Electrical supply and communication systems shall be designed, constructed, and maintained for their intended use, regard being given to the conditions under which they are to be operated, to enable the furnishing of safe, proper, and adequate service.
SED cited that rule three times in the Zogg report. One of the three was for the tree.
Four defects, printed in a violation
SED's conclusion in Incident Investigation Report E20201009-01, dated June 2, 2022, is the clearest published statement in California of what a line-strike defect set looks like:
GO 95, Rule 31.1 for failing to maintain its facilities safely and properly, by failing to abate the Subject Tree which was identified for removal during 2018 post-Carr Fire VM work. The Subject Tree had a 23 degree lean from vertical towards the Subject Conductors, a large open cavity, no uphill supporting roots, and partial internal decay. This weakened the trunk and caused the Subject Tree to fail, fall onto the Subject Conductors, and ignite the fire.
Each of those four is a separate observation, made in a different way, from a different place.
Defect | How it was established | Measurement in the record |
Lean toward the conductors | LiDAR from the incident and PG&E's 2019 LiDAR survey | 23 degrees from vertical, southward |
Open basal cavity | CAL FIRE and the contracted arborist, on the ground | 4 feet high, 3 feet wide, 14 to 15 inches into a 24.5 inch stem |
No uphill supporting roots | Root excavation, quadrant by quadrant, December 14, 2020 | Boulder a few feet uphill; roots east and west about 12 inches in diameter |
Partial internal decay | Pulled out fibers on a saw cut face | Observed on a 12.5 foot cross-section |
The weather that afternoon is worth reading alongside them. The Clear Creek Weather Station, about four miles south-southeast, recorded 90 degrees Fahrenheit, north-northeast wind at 11 miles per hour with gusts to 22, and 13 percent relative humidity at 1450 hours. That is a hot, dry, ordinary September afternoon in Shasta County. The loading was ordinary for the season. The tree was the variable.
The cavity was on the other side of the tree
The single most useful sentence in the whole Zogg file is a note SED made on its own field visit on October 7, 2020:
Since the large open cavity was on the north-facing side of the trunk, it was not visible when standing from the south where PG&E's facilites were located.
The arborist's report reached the same conclusion and drew the working rule out of it. The lean should have been obvious from any vantage point near the conductors. The cavity and the absent roots were not visible directly from under the conductors, and would have been obvious from uphill of the tree or from either side.
That is a specification for how to walk a corridor, written by an expert witness after the fire. One face of a trunk is the face that points at the wire, and it is the only face an inspector standing in the span can see. On a tree 60 feet off the centerline, on a slope, the defect that decides the outcome is on the uphill face by preference, because that is where the tension roots are missing and where a grade cut or a boulder interrupted them.
The arborist also applied the International Society of Arboriculture's tree risk assessment protocol to the tree and reported that it would have yielded a risk rating of High. Which assessment level a program specifies, and therefore whether anybody was ever going to walk around the trunk, is a decision made in the procedure long before the inspector arrives.
Seven families, and the four that carry a number
The most widely circulated defect list in American arboriculture is How to Recognize Hazardous Defects in Trees, NA-FR-01-96, published in 1996 by the Minnesota Department of Natural Resources and the USDA Forest Service Northeastern Area. It is a brochure written for home owners and land managers, it is advisory, and no California statute, general order or CAL FIRE guide adopts it. It is also where the numbers are.
We recognize seven main types of tree defects: dead wood, cracks, weak branch unions, decay, cankers, root problems, and poor tree architecture. A tree with defects is not hazardous, however, unless some portion of it is within striking distance of a target.
Four of the seven come with a threshold an inspector can measure. Three are left to judgment.
Defect | The published trigger |
Dead wood | A tree is dead; a broken branch or top is lodged in the tree; a branch is dead and of sufficient size to cause injury |
Cracks | A crack extends deeply into or completely through the stem; two or more cracks occur in the same general area; a crack is in contact with another defect |
Weak branch unions | The union is on the main stem; the union is cracked; the union is associated with a crack, cavity or other defect |
Decay | The thickness of sound wood is less than 1 inch for every 6 inches of diameter at any point on the stem |
Cankers | A canker or multiple cankers affect more than half of the tree's circumference |
Root problems | More than half of the roots under the crown have been cut or crushed; the tree leans with recent root exposure, soil movement or soil mounding |
Poor architecture | The tree leans excessively; a large branch is out of proportion with the rest of the crown |
Take the decay line and run it on the Zogg tree. One inch of sound wood for every six inches of diameter, on a 24.5 inch stem, is a shell just over four inches thick. What the Zogg record holds instead is a cavity depth, taken off a tree lying on the ground, after a fire, by three parties who arrived to investigate a fatality. The threshold is only useful in advance of the failure, measured on a standing stem.
The brochure also carries the sentence that disposes of the green canopy argument: "green foliage in the crown does not ensure that a tree is safe. Tree trunks and branches can be quite defective and still support a lush green crown." PG&E's position on the July 2019 aerial photographs was that the gray pine displayed a full, green canopy with no apparent signs of disease, decay or fire damage.
SED's answer was that the photos do not provide a detailed view of the trunk, trunk base, or individual limbs and branches. An aerial image resolves the crown and nothing else, which is why the difference between a patrol and a detailed inspection is the question an investigator asks first.
The vocabulary California actually writes down
California has had a defect vocabulary in use since 1987, and it is the one the tree failure program collects. The reporting instrument is the International Tree Failure Database User Manual, prepared by E. Thomas Smiley, Nelda Matheny and Jim Clark. It is dated March 2005 and revised May 2006, and it was written under a USDA Forest Service grant. The University of California posts it as the user guide for the California program.
It is a post-mortem form. That is exactly why it is useful: it defines every defect operationally, because somebody has to check a box for it after a tree is already on the ground. The defect list comes in three parts, keyed to where the tree broke.
Trunk failure offers failed portion dead, decay, canker, multiple trunks or codominant stems, dense crown, flush cuts, topped, one-sided, low live crown ratio, included bark, lean, cracks in wood, lightning, animal, fire, insect and mechanical injury, and girdling.
Branch failure offers failed portion dead, decay, dense crown, heavy lateral limbs or heavy ends, included bark, crook, failed portion is an epicormic branch, cracks in wood, mistletoe or epiphyte, mechanical, lightning, insect and animal injury, and canker.
Root failure begins with a required choice among four mechanisms: roots broken, roots cut or severed, root plate lifted out of ground, and roots restricted. Its defect list then offers fire scar, basal wound, low live crown ratio, corrected lean, uncorrected lean, animal injury, cracks in wood prior to failure, and surface roots or root collar wounded.
Two instructions in that manual are worth adopting verbatim into a pre-inspection procedure. The first is the association test:
You may select as many of these defects as needed as long as the defect contributed to or was associated with the failure. Do not select a defect merely because it was present on the tree.
The second is the question the whole database exists to answer, which is also the question a Rule 35 file gets asked after an ignition: "Were the defects associated with failure visible before the tree failed?" Yes, No, Unknown.
The manual supplies the measurement protocols to go with the vocabulary. Sound wood thickness around a decay column is measured in at least three places around the circumference, bark excluded, and partially decayed wood is counted as decay. A cavity opening is recorded as a percentage of trunk circumference at cavity height. Early decay is read by pick-and-pry, which runs to four steps:
Drive a sharp screwdriver, knife tip or awl into the wood at an acute angle, across the grain.
Bend it back to snap a small piece from the surface.
Compare the break with known sound wood from the same tree.
Read a brash break across the grain as lost fiber strength, and a splintery break as sound wood.
Those are field procedures a crew can run with what is already on the truck.
Where the defect sits decides what reaches the wire
For a trunk failure the database requires two numbers: the height of failure above grade, and the diameter at the break inside bark. Those two say what came down. A stem that breaks at 30 feet puts the top of the tree in the span and leaves the bottom standing. A root failure puts the whole tree across the corridor.
The Zogg tree is the arithmetic in one line. A stem 105.5 feet long, rooted about 60 feet from the conductors, failing at the base, made contact about 67 feet up the trunk. The defect was at the bottom, so the whole length was available, and the conductors were well inside it.
The failure record splits by species along exactly these lines, which is why the profiles are worth reading before the walk rather than after it.
Coast live oak fails low, at the root and the base, with decay present in most trunk failures.
Valley oak drops a live lateral limb out of a sound crown, which is a branch failure into a single span.
Gray pine puts a codominant fork over the corridor, which is a trunk failure at height, and the Zogg tree failed at the base anyway. A profile narrows the search. It does not close it.
Redwood comes over on the root plate, and Douglas-fir fails standing dead at the base while the crown is still green.
The report form also carries two outcome categories that name this page's subject directly. Fire is defined to include a fire started by the tree contacting or arcing to a power line. Power outage is defined as a tree failure that damaged a power line and caused an outage on that circuit or beyond. Utility right-of-way is one of the settings, defined as sites within a maintained utility right-of-way, especially electrical power line rights-of-way.
Some of these defects are the corridor's own work
Several items on the defect lists are produced by line clearance, not merely found by it. A program that reads defects alongside its own cut history is auditing the whole file.
What the corridor does | The defect it creates | The published threshold |
Crown raising for clearance | Low live crown ratio | A majority of live branches in the upper third of total tree height |
Directional pruning and reduction | Cuts that do not compartmentalize | More than 25 percent of live tissue removed, or any cut larger than 4 inches in diameter |
Topping and heading cuts | Decayed stub, then epicormic branches on it | Record the diameter of the largest stub |
Trenching for a new pole or anchor | Roots cut or severed | Straight cut ends indicate a trenching machine, rock cutter or saw |
Grading an access road | Grade change within the dripline | Cuts and fills of 2 inches or more |
Any fixed barrier near the trunk | Roots restricted | A wall 1 foot from the trunk face restricts roughly 40 percent of the root zone |
The 1996 brochure is blunt about the heading cut:
Both of these practices are harmful and should never be used. The inevitable result of such pruning wounds is decay in the remaining stub, which then serves as a very poor support to any branches that subsequently form.
A stub cut under a conductor becomes a decay column, then a cluster of epicormic shoots on top of the decay column, and in eight or ten years the tree offers the crew a branch union that is weaker than the one that was cut. What trim type was actually performed, and what the record shows about it, is the evidence that closes or opens this loop.
Multiple defects on the main stem change the finding
The brochure sets one rule about combination, and it is categorical:
Multiple defects that are touching or are close to one another should be carefully examined. If more than one defect occurs on the tree's main stem, you should assume that the tree is extremely hazardous.
The Zogg tree carried four on the main stem and its root collar. Lean, cavity, missing roots, internal decay. The cavity and the missing roots were on the same side, and the arborist found that the oversized east and west roots had grown into the side of the cavity, meaning the two developed together over many years. Those four observations describe one mechanism, read four ways.
Species rating pushes the same direction. Revision 3 of PG&E's hazard tree rating and scoring procedure, TD-7102P-07, published August 14, 2019, puts gray pine at a Very High failure potential and names July through October as the months it is most likely to fail. The tree failed on September 27, inside the window its own utility's procedure had flagged, on a circuit its own utility patrolled.
A mark is not a finding
The Zogg tree was identified. That is what makes the file worth reading twice.
On August 23, 2018, under Work Order CA_GI 1101_26, a Mountain G Enterprises quality control inspector recorded two gray pines, Tree IDs 6557 and 6558, at coordinates SED later matched to the subject tree, as Priority 2 trees for removal. PG&E's account of what that record contains is the sentence a pre-inspection program should read out loud:
The Collector data does not, however, provide any information specifying why the QC inspector identified the two Gray Pines for removal.
A priority code and a coordinate pair. No defect named, no dimension, no note of where the inspector was standing. The next person on the circuit had nothing to re-read, nothing to rank against anything else, and nothing to defend. What a pre-inspection unit has to contain to be workable without a phone call is the standard that entry failed.
The work then stopped. In late September and October 2018 PG&E halted post-Carr Fire vegetation management work in the Zogg Mine Road area after interactions with an armed resident who believed trees were being marked unnecessarily, and decided to rely on routine patrols for the remainder. The trees were not removed.
Routine patrols in October 2018, April 2019 and April 2020 passed over them. What a refusal or an access obstruction actually requires of the company, and what can be done when a tree cannot wait, are the two doors that were available and not used.
SED's finding was that PG&E found the tree and left it standing.
What a defect record has to carry
The gap between the Zogg entry and a usable one is small, and every item in it is something an inspector already has in hand at the tree.
The species, and the failure profile the program holds for it.
Diameter, and the height the diameter was taken at. The arborist recorded 24.5 inches at 54 inches above grade, not "about two feet."
The defect, named in the vocabulary the procedure adopted, one name per observation.
The dimension that makes it reportable: cavity depth and width, sound shell thickness in three places, degrees of lean, percentage of circumference, number of roots cut.
Where the observer was standing. Uphill, downhill, in the span, or in a truck.
The date, the inspector, and the interval the judgment was scoped to.
Rule 31.1 asks for accepted good practice by that name and nothing more. SED then reads the file to see whether the company followed the practice it said it accepted. A record that names the defect, measures it, and says which way the tree would go answers that question. A priority code on a map does not.
Further reading
What makes a tree a hazard tree: Reach, target and weather envelope, which are settled before any defect on this page matters.
Field hazard tree assessment: The assessment level that decides how close somebody gets to the trunk, and who specifies it.
Patrol and detailed inspection: What a simple visual inspection is allowed to miss, and where a basal defect stops being anybody's excuse.
Frequently asked questions
Does any California rule list the defects an inspector has to look for?
No. GO 95 Rule 35 names dead, rotten and diseased. Public Resources Code section 4293 adds its own categories in its trailing paragraph, one of which turns on lean rather than on condition. Neither is a defect list with tests attached. GO 95 Rule 31.1 closes the gap by requiring that, for all particulars not specified in General Order 95, the company maintain the facility in accordance with accepted good practice for the intended use and known local conditions. Which published defect list a program adopts is therefore a decision, and the procedure is where it gets recorded.
Which defects made the Zogg Fire tree a hazard, and what did the CPUC find?
Four, and the Safety and Enforcement Division printed all four in the violation. Incident Investigation Report E20201009-01, dated June 2, 2022, records that the gray pine had a 23 degree lean from vertical toward the conductors, a large open cavity, no uphill supporting roots, and partial internal decay. SED found PG&E in violation of GO 95 Rule 31.1 for failing to abate the tree, which had been identified for removal during 2018 post-Carr Fire vegetation management work and never removed.
How much sound wood does a hollow trunk need?
The only published number a California inspector is likely to be handed comes from a USDA Forest Service and Minnesota Department of Natural Resources brochure, NA-FR-01-96, which says to take action if the thickness of sound wood is less than 1 inch for every 6 inches of diameter at any point on the stem. On the 24.5 inch gray pine at Zogg that threshold is a shell just over four inches thick. The brochure is advisory and no California regulation adopts it, so a program that uses the number has to name the source in its own specification.
Can an aerial photograph support a finding that a tree is sound?
Not on its own, and the Zogg record is the demonstration. PG&E held July 2019 aerial photos of the subject tree, and stated that in them the tree displayed a full, green canopy with no apparent signs of disease, decay or fire damage. SED's response was that the photos do not provide a detailed view of the trunk, trunk base, or individual limbs and branches. The cavity was 4 feet high and 3 feet wide at the base of that green tree.
Is marking a tree for removal enough to close the finding?
No. At Zogg the tree was identified. Mountain G Enterprises recorded two gray pines, Tree IDs 6557 and 6558, as Priority 2 trees for removal on August 23, 2018. PG&E later stated that the Collector data does not provide any information specifying why the inspector identified them. Work in the area stopped in October 2018 after interactions with an armed resident, the trees were never removed, and SED cited Rule 31.1 for the failure to abate.
