This purpose of this study is to estimate what percentage of combustible yard fences is found within the Zone 0 of a primary structure in the WUI in Northern California, as compared to the total use of combustible fencing in the property.
Introduction
The use of yard fences throughout the United States of America appears to be a region-dependent cultural choice. While there are, of course, no absolute rules, and individual choices ultimately govern the use of yard fences, both the North-East and the South appear to favor short picket-like fences, while the Midwest appears to favor no fences at all. The coastal Western states, however, see significant use of privacy wood fences. This choice is certainly prevalent in Northern California. This study looks at the use of combustible fences within the WUI in the Very High Fire Hazard Severity Zone (FHSZ) in Berkeley, CA. Specifically, we studied the zone defined as “the Grizzly Peak Mitigation Area” (GPMA) by the city of Berkeley, which borders the eastern boundary of the city, alongside Tilden Park. How much of the combustible fencing around the whole house is used within the primary structure’s zone 0, and what impact would that have upon the use of forest products around the house?
Study methodology
The GPMA was analyzed in a previous study. A exhaustive analysis revealed an average lot size of 8,072 ft2 for built lots, and a median lot size of 6,553 ft2.
To minimize sample choice bias, the study chose to select a group of entire contiguous blocks where the average lot size would be close to the that of the GPMA. We expected the study result to show a very low percentage of combustible fencing within Zone 0—to ensure that we would obtain a worst-case result (from the point of view of the decrease of use of forest products, were we to stop using combustible fences in Zone 0), we skewed the choice of neighborhood towards a slightly lower average lot size, assuming that smaller lot sizes would increase the percentage of combustible fencing within the primary structure’s Zone 0. We identified a group of 5 entire contiguous blocks of houses, totaling 101 houses across different streets, with an average lot size of 7,159 ft2 (within 11.5% of the GPMA as a whole), and a median lot size of 5,967 ft2 (within 9% of the GPMA as a whole). The sample size validation is analyzed further below.
The study used both direct street-side visual inspection and satellite imaging measurements. The houses were first surveyed street-side by observers. The use of combustible fencing on each of the four sides of the house was estimated, as well as the percentage of combustible fencing within Zone 0. The numbers were then validated by satellite imaging inspection for the side and back fences where the street-side estimates were questionable. The back fences were directly observable on location in approximately 45% of the cases, and well defined on imaging in an additional 35% of the cases, for a total of approximately 80%. For worst-case analysis (i.e. maximizing the percentage of fencing within Zone 0), back or side fencing that was unavailable was logged as 0. All fencing within Zone 0 that was in question for in-person observers appeared well defined on imaging. When there was any uncertainty, again for worst-case analysis, 100% of the fencing in question was logged as being within Zone 0.
Only combustible fencing was logged in the study. Non-combustible fencing was frequently found but logged as 0.
Results
Throughout the sample, the average length of combustible fencing per lot was found to be 145 linear feet, of which the average length of combustible fencing within Zone 0 was 20 linear ft.
The percentage of combustible fencing within Zone 0 vs the total combustible fencing on the property was found to be 13.95%.
Statistical validity and margin of error
The standard deviation for our sample of 101 houses was 15.1%. Using a standard statistical equation for the margin of error, using a Z-score of 1.96 for a 95% confidence level, we get:
Margin of error = 1.96 x standard deviation/ SQRT(sample size) = 1.96 x 15.1%/ SQRT(101) = 2.9% (this represents 2.9% for the percentage of combustible fencing in Zone 0 as a ratio of fencing within Zone 0 to total combustible fencing).
Therefore, we have a 95% confidence level that the percentage of combustible fencing in Zone 0 as compared to the whole combustible fencing in the property in the GPMA is within 2.9% of 13.95%.
Do lot sizes count for much—sensitivity analysis?
In the whole experimental process, we were concerned about lot size. We felt that lot size could be important to this calculation, and that we needed to carefully match our sample to the lot size average and median—which we did. But does lot size really matter?
We had assumed that, the larger the lot size, the smaller the percentage of combustible fences in Zone 0 (because there is a lot more fence in large lots, but the house size does not change as much, and because there is less room to keep fences away from the house in a small lot).
A second analysis: slightly smaller lots
To verify our assumption, we did a second analysis, using another sample with a little smaller lot size—we used 103 properties, also contiguous, with an average lot size of 6,756 ft2 (6% smaller than before) and a median lot size of 5,662 ft2 (also 6% smaller). Some lots were included in both samples.
The results essentially did not change. We obtained an average of 148 feet of combustible fence per lot, and 21 feet of combustible fencing within zone 0, for a overall percentage of 14.39%, vs our previous 13.95% of combustible fencing within Zone 0.
A third analysis: much smaller lots
What would happen if we went with a radically smaller lot? This time we did not pick a different sample. Instead, we combined the two previous samples, and removed from the combined set every lot over 10,000 ft2, thereby obtaining interrupted (instead of contiguous) sets of houses, but with a much lower lot size average. There were 91 houses left, meaning the statistical relevance changed very little. The lot sizes, however, changed significantly, and went down to an average of 5,683 ft2 and a median of 5,500 ft2.
The results changed very little. We now have an overall combustible fencing length of 142 linear feet per property, and Zone 0 combustible fencing of 22 linear feet per property, for a proportion of Zone 0 combustible fencing over total combustible fencing of 15.58%.
Conclusion
Regardless of lot size, in the most dangerous part of the WUI (the Grizzly Peak Mitigation Area), in Berkeley, among 960 homes, the percentage of combustible fencing within Zone 0 remains around 15%, a very small percentage with respect to the use of forest products.
Study authors in alphabetical order: Nancy Gillette, Mary Pat Farrell, Staeppan Snyder, Michel Thouati. This study was undertaken by HelpBerkeley.org in collaboration with BerkeleyFirewise.org.