3P Sampling

3P Cruising Large Areas, One Cruiser’s Experience

by Wayne Beck

Cruiser, Boise National Forest

Published July 1997


During the winter of 1993-94 I began working on a cruise design for a sale to be marked early that spring. The sale consisted of three stands that were to be thinned from below to a basal area of 70-80 square feet. All three of the stands had basal areas that ranged between 0 and 155. The three of them combined covered approximately 340 acres. The sale was to be sold as a tree measurement sale, which meant that the cruise would have to meet a 10% sample error with a T value of 2. As I worked through the numbers, it became clear that a plot cruise would be problematic. The sale would have plots on top of plots and I knew that the area determination on the stands would also be a large task. By default, I chose frequency cruising, because at that time we only had these two options available in our cruise expansion. We cruised the sale using this method and met our error, but we also cruised over 500 trees. As I worked through this process I began to look back through my forest mensuration book and wondered if there was a better way. I hit the idea of 3P around, but did not fully understand it and did not know how to generate a random number list.

As we marked and cruised this sale, my determination to find a better method for cruising sales to tree measurement standards grew. I continued to study 3P and found out about the Variable Probability Sampling course taught by Bell and Associates at Oregon State University. As the time approached to start designing/preparing a large cruise based sale, I began to talk to my boss about attending the Bell course in hopes of finding this better way. The sale that drove me to the Bell courses, is called Logging Gulch. It has 60 units covering approximately 2800 acres most of which have been cut-over in the past. The cutting that had occurred ranged from salvage of insect mortality to shelter-wood cuts consequently the variation in the stands is extremely high.

About the same time I attended the Bell course, the NATCRS program became available and it supported 3P cruising. I had the opportunity at the Bell course to test the 3P data recorder software by Holmboe Enterprises and felt it had great application. Therefore, armed with the knowledge that I had gained at the Bell course and the ability to process a 3P cruise (with NATCRS) we forged ahead. We purchased some data recorders and software and tested the system on two small sales (less than a million board feet). The test sales produced excellent results with sample errors of less than 10 percent. The tests showed that the CV (coefficient of variation) of 35 recommended in the literature fit our cruise very well. The test sales also demonstrated to us that if you stratify you had better have a good idea of the XKPI for each stratum. On our first try we over estimated the XKPI for one stratum and had to return to measure the reserve trees that we had selected during the cruise.

With the knowledge gained during the test sales, I began to design the Logging Gulch cruise. I decided to make four strata: Ponderosa Pine, Douglas-fir and other to correspond to the species breakdown in our typical timber sale contract. I also added a catch all strata to pick up anything that I did not expect to see. The sale was estimated to have approximately 12.5 MMBF and I estimated that the strata mix (for the cut volume) would be approximately 56% Douglas-fir, 38% Ponderosa Pine and 6% other species. I designed each stratum to meet a 10% sample error. This meant that approximately 50 trees should be cruised in each stratum:


I then doubled the number of samples to provide for reserve trees (these extra trees would be numbered and mapped in case we needed more samples). A 10% error for each stratum may sound like overkill. It is. However, with this overkill we would still only have to measure 150 trees. I calculated Next Column

Wayne Beckthat would mean about one measure tree per day. Besides any less made everyone nervous, including me (remembering that an over estimation of volume can produce too few samples). I decided that we would estimate volume in Decimal C (otherwise, we may exceed NATCRS’s XKPI field size of five characters). This produced the following K+Z values 7,020, 4,060 and 800 for the Douglas-fir, Ponderosa Pine, and other strata that I had chosen. This meant that for example, a Douglas-fir would be selected approximately every 70,200 bf (converted from decimal C) and every other one would be cruised. In other words a measure tree would be selected by the 3P selection rule approximately every 140,400 board feet.

The other thing that we had learned from our test sales was ideal crew mechanics. Three to four marker/estimators and one tally person works best. The markers move through the woods and select the trees to be cut using the provided marking guidelines. As each tree is marked, the species, estimated diameter, and adjustment factor are called to the tally person. These estimates are made while the tree is being painted and should take no longer. The estimates must be quick and consistent. The tally person then calls back the estimates to confirm them. The diameter is entered into the data recorder and is used to calculate the gross KPI for the tree (the software uses a regression formula for this task). An adjustment is made if there is defect in the tree or if it is abnormally tall or short. It is only called if the marker feels that there is a need for an adjustment of more than 10 percent. This adjustment percentage is used to adjust the gross KPI producing net KPI for the tree. The data recorder software then uses the species and estimated diameter to select automatically the appropriate strata. It then compares the net KPI to a random number based on that strata’s K+Z value. If the tree is selected the tally person calls out the cruise number for the tree and it is painted on the tree along with an additional paint band. The tree is then flagged, cruised, and mapped or GPS location recorded. I feel this process produces the best results, because it reduces the chance for personal errors. The species, diameter and adjustment factor (if needed) are confirmed by the tally person before entered and the software ensures that each tree is placed in the proper strata and compared to the appropriate random number list. The cruise information is also entered in the data recorder as the trees are cruised further, reducing the chance of personal error.

The marking and cruising of Logging Gulch has not been top priority due to a large salvage project on the district that was a result of the 1994 Boise River Fire. However, to date we have marked approximately 40 percent of the acres of this sale and the results have been great. The units have been, as we suspected, highly variable. On average, approximately 24 trees per acre have been marked equaling 420 KPI per acre (KPI is in decimal C BF). The diameters have ranged from 8 to 56 inches. The coefficient of variation has stayed around 35 percent for net KPI just as we found during our test sales. Of the nearly 28,000 trees estimated only, 37 trees have been cruised producing a combined sample error just over 22% with a T of 2. I still fully expect to meet a final error of 10% even though about 2.8 MMBF has been lost to fire, insect and riparian buffers. The best thing about this system is we do not have to go back to any steep brushy units, when the painting is done the cruising is done.

My message is: do not let large areas scare you away from trying 3P cruising. If you are going to mark the trees for cut, it should definitely be considered when designing your cruise. In today’s world of ecosystem management I feel it will become a favorite with many cruise designers and cruisers even on large areas. I hope that my sharing the experience that I have had with 3P cruising on large areas will be helpful to some of you.


Return to Home
Back to Contents