By Ed Hayes, Forest Health Specialist
 Article first appeared in the April 99 issue of Tree Care Industry

On Sunday March 29, 1998, an F-3 class tornado ripped through St. Peter, Minnesota (population 9,420) devastating the community and destroying an extensive and mostly mature urban forest. This storm hit St. Peter in the leaf-off season. Only 2.5% of all the tornadoes from 1950 to 1994 were in the class F-3.

It was estimated by the city of St. Peter that 8,406 trees blew down when the storm struck the community. Toppled trees were removed in the first five days after the storm by an army of contractors, Department of Natural Resources crews and volunteers. On Friday, April 3, a tree inspection began to evaluate the safety and defect status of all the remaining standing trees throughout the hardest hit areas of the city. Ten trained DNR tree inspectors working in pairs, inspected an estimated 3500 trees in the following three days. A total of 1,083 trees were storm damaged, 623 were marked for removal, and 460 were saved and marked for pruning.

Criteria use for tree removal were based on the Minnesota's Risk Assessment Guidelines for Hazard Trees.  (See References, Albers and Hayes.)  Five criteria for defect assessment were given to the trained inspectors. The inspectors were instructed to record the reason for all tree removal and estimate the tree size in dbh, (diameter at breast height). Trees with more than one defect category were recorded as multiple defects.  All standing trees were inspected on both public and private property. On private property attempts were made to contact the homeowner for their input.

The five failure categories used for tree removal decisions were:

1. Wind-throw; Leaning trees with evidence of recent root lifting or soil movement.

2. Crown Damage; If greater than 1/3 of the crown was damaged consider removal, if greater than ½ of the crown was damaged, remove tree. (Tree removal is preferred over severe pruning or topping for cultural, aesthetics, long term maintenance  and future safety. (See References, Shigo).

3. Main Stem Failure; Failed weak unions resulting in compromise of 1/2 or more of the stem circumference.

4. Main Stem Failure; New cracks associated with an existing defect that what would have otherwise been considered less than threshold level, (moderate risk potential), defects.  These are decay columns, and or and canker and decay defects with new cracks, that affect 1/3 to ½  the stem circumference. (See References, Hayes.)

5. Main Stem Failure; New cracks that compromise up to or greater than ½ the stem circumference.

Multiple Defects; If more than one of the above defects contributed to the tree failure, the multiple defect category was used.

Table 1.    Number Of Trees Removed by Defect Category
Defect Category # Trees % of Total
1. Crown Damage 324 52
2. Wind-throw. 123 19.8
3. Failed Unions. 117 18.8
4. New Cracks w/ existing defects. 13 2.1
5. New Cracks. 9 1.4
Multiple Defects. 37 5.9

Of the 1,083 storm damaged, still standing trees surveyed in St. Peter following the March 29, 1998 tornado, 623 or (57.5%) were recommended for removal and 460 or (42.5%) were salvaged by recommending pruning and or removing lodged and hanging branches.  See Tables 2 and 3.

Table 2. Number Of Trees Removed, By Percent and Size.
Species # trees % Ave. dbh
All - Maple Sp.(Silver & Sugar) 208 33.4 19.5
Sugar Maple 68* - 21.5
Silver Maple 66* - 18.8
Elm Sp. 63 10.1 19.5
Green Ash 51 8.1 15.6
Spruce Sp. 42 6.7 10.5
Black Walnut 35 5.6 18.8
Boxelder 34 5.4 14.6
Hackberry 33 5.3 17.8
Locust Sp. 33 5.3 13.7
Norway Maple 21 3.3 10.8
Red & White Cedar 16 2.5 12.2
Basswood 14 2.2 15.8
Balsam Fir 11 1.7 8.0
Red Pine 11 1.7 11.2
White Pine 7 1.1 16.5
Black Willow 7 1.1 23.1
Other, 10 species 37 5.9 -
Total 623 - -

*=The number of Sugar and Silver identified to species from the total of all 208 maple records.

Table 3. Percent Of Tree Species Removed By Failure Category.
Percent of Species Removed
Crown Damage Windthrow Main Stern Failures Multiple defects
Unions Defects New Cracks
Sugar Maple 54 6 21 3 4 12
Silver Maple 59 7 20 0 2 9
Elm Sp. 70 10 11 0 0 10
Green Ash 53 27 20 0 0 0
Spruce Sp. 38 60 0 0 0 2
Black Walnut 60 18 13 3 0 6
Box elder 68 9 9 6 0 9
Hackberry 18 21 43 4 7 7
Locust Sp. 33 17 42 0 0 8
Norway Maple 33 0 19 19 5 24
Red & White Cedar 40 60 0 0 0 0
Basswood 36 50 7 0 0 7
Balsam Fir 64 36 0 0 0 0
Red Pine 19 81 18 0 0 0
White Pine 100 0 0 0 0 0
Black Willow 57 28 14 0 0 0

Basically, trees that resisted wind-throw, had their crowns ripped to shreds, or failed as weak co-dominant branches split the main stems in half.    Let's look at the top four reasons for removal and their the patterns of failure.

Patterns Of Failure

I. Crown Damage
The most common mode of failure was crown damage, 52% of all trees removed. Significant
crown damage occurred in the maples, elms,  green ash , black walnut, locusts, black willow, and some conifers. A clear pattern emerged for branch breakage. Generally trees greater than 16 inches that resisted wind-throw began to loose branches. Among the hardwoods, boxelder, sustained the highest amount of crown damage, (68%). No surprise here, the species is well known for significant branch breakage. For spruce species this cut off was 14 inches. Spruce at 14 inches and greater that resisted wind-throw, lost their branches.  Among the conifers, white pine sustained the highest amount of crown damage, (100%) but there were only seven white pine records. However, it is apparent that white pines loose branches rather than wind-throw.

Table 4. Percent of Crown Damage And Tree Size, When Significant Branch Breakage Began To Occur.
Species % Crown Damage Size,-dbh
Sugar Maple 54 16"
Silver Maple 59 16"
Elm sp. 70 16"
Green Ash 53 16"
Black Walnut 60 16"
Box elder 68 10"
Basswood 50 16"
Balsam Fir 64 >10"
White Pine 100 8"
Black Willow 57 16"

II. Wind-throw
For these storm damaged trees, both hardwoods and conifer, wind-throw was most
common in trees up to 14 inches in diameter. Trees 14 inches and less failed by  wind-throw rather than shedding their branches. Wind-throw in conifers was much higher. Remember this was a leaf-off storm. In conifers, (with crown sails =foliage), wind-throw was 60% for spruce and cedars, and 80% for red pine. Spruce up to 14 inches failed primarily by wind-throw. Balsam fir was more resisted to wind-throw, (36%), preferring stem failure instead, (64%).  In hardwoods, wind-throw was also most common in trees up to 14 inches diameter.

III. Main stem failure; failed branch unions
The most common pre-existing ?defect? on the remaining standing trees, that lead to main stem failure, was weak branch unions. Main stem failures originating from weak unions accounted for 18.8% of the removals. While main stem failures resulting from pre-existing, non-threshold level defects, with new cracks, resulted in only 2.1%. These would be existing decay columns and canker faces. (See References, Hayes.)

Generally 20% of the maple species sugar, silver, and Norway, sustained main stem failures originating at the sites of weak unions. Additionally weak union failures occurred in green ash, black walnut, box elder, black willow, and an unusually high amount in hackberry, (43%).

In silver maple most of the weak unions were shed or failed by the time the species reaches 20 inches in diameter. By contrast, weak unions in sugar maple continued to fail up through the 20 inch classes to the 30 inch diameter class. Weak unions in sugar maple were retained into larger diameter classes than for silver maple. These patterns support what we generally see for these two species.

IV. Multiple Defects
The only pattern that emerged confirms observations associated with Norway maple. Norway maple had the highest number trees removed for this category. Norway maple commonly have  cracks associated with pre-existing wounds and decay columns. Norway maple are not well acclimated to the climate of this region, and it shows.

Lessons in branch breakage and wind-throw
Few structures, let alone trees, survive the force of an F-3 class tornado. An F-3 class tornado has wind speeds of 158-206 mph and is defined in the Fujita scale as causing, ?roof and some walls torn off well constructed houses; trains overturned; most trees in forests uprooted.? In such a storm's main path, both massive wind-throw and the domino effect likely combine to level trees or stands of trees.  Even at wind speeds of 50 to 60 mph, the force of the wind can result in wind-throw, exceeding the threshold of failure for even structurally sound trees. This threshold is described by Mattheck as the tree's biological safety factor. (See References, Mattheck.)

In this sample of the storm damaged trees that were still at least upright, we see these thresholds at work in the patterns of failure.

The biological safety factor for structurally sound trees is about 4.5, or 4.5 times the trees? service load. The service load is the tree's above ground mass, (it's weight), times the acceleration of gravity, (it's weight pushing against the surface of the earth). This normal service load can be increased by a factor of 41/2 before the tree fails, for trees without defects and with full crowns. Trees with defects fail sooner.  Since the tree is a chain of links of equal strengths, the biological safety factor applies from the branches through the stem and through the root plate and into the ground.

The biological safety factor for branches is less than the biological safety factor for the main stem. This means that branches are made to be shed at the expense of the main stem. A nice ecological adaptation that says the strategy is, ?it is better to shed a branch than to lose one's life.? All depending of course, on how much it costs.

The cut off between wind-throw and branch breakage was generally 14 to 16 inches diameter for most of the hardwood species. Trees 16 inches and greater that resisted wind-throw, experienced more branch breakage. While trees 14 inches and less experienced more wind-throw than branch breakage. Larger trees would rather shed their branches than wind-throw, to a point.

A parting warning on white pine. Do not mix high value targets with large white pines. Give them plenty of room. White pine branches can be easily shed during storm events.

To resist wind-throw a tree needs to have a well developed circular root plate. Trees with only 60% of this requirement are at threshold for wind-throw. These trees have no safety reserves remaining. They can fail by a force equal to their service load.

If the load (force of the wind) exceeds the safety factor of the root plate, the tree fails by wind-throw, no matter what the size. For larger diameter trees to resist wind-throw, an adequate root plate radius need only to be as much as three times the stem radius. For small diameter trees, the root plate may need to be as much as fifteen times the stem radius. The reason for this is easy. As you increase mass (weight) you increase the friction between the root plate and the surrounding soil, decreasing the need for a larger root plate.
However, even in true wind-throw, it is the soil that fails first, not the tree roots. Which is why we often see wind-throw in broadleaf trees in the spring, following foliage development and in wet soils. Wet soils have less cohesion and the ?slippage? is dramatically higher. It is like slipping on wet pavement after a spring shower.

The most common pre-existing condition for predisposing trees in high use urban areas to wind-throw, is restricted root zones. Trees without adequately developed circular root plates, or with asymmetrical root plates, or with disturbed root systems, are predisposed to wind-throw. In a practical way it is difficult to apply Mattheck's wind-throw mode, and it can only be applied to trees with circular root plates.

However, it may be another risk assessment tool for use by arborists interested in evaluating trees for defect. At any rate it provides us with interesting insight into failure of trees by wind-throw.

The 16 inch rule
Sixteen inches should not necessarily be thought of as a diameter limit for tree replacement in urban forests. However, it points out the need for not only species diversity but also for size diversity. Size is an important component of age class diversity.

Trees Saved
In addition to the 460 trees summarized in Table 5, it is estimated that another 1800 trees located  on both boulevards and private land survived.  Their average size is estimated to be 9 to 10 inches in diameter.

Table 5. Number Of Trees Saved By Corrective Pruning, Percent and Size.
Species #trees % Ave. dbh
Maple Sp. (Silver & Sugar 158 34.3 19.5
Green Ash 46 10 14.2
Honey Locust 43 9.3 11.6
Norway Maple 43 9.3 12.9
Black Walnut 42 9.1 18.1
Elm Sp. 23 5.0 17.0
Hackberry 22 4.7 15.7
Basswood 14 3.0 14.8
Other(>dozen species) 69 17.6 -
Total 460    


Albers, J. and E, Hayes , 1992, How to detect, assess and correct hazard trees in recreational areas, . St. Paul, Mn. DNR. 63pp.

Hayes E. M.. 1997.? Risk assessment Guidelines For Hazard trees, An easy to use field guide?,  Tree care Industry, April,1997.

Mattheck, C., and H. Breloer. 1994. The Body Language Of Trees: A Handbook For Failure Analysis. HMSO, London.

Shigo, A.L 1991. Modern Arboriculture. Durham, NH: Shigo and Trees, Associates: 424 p.