Category: Post 4: Sampling Strategies

Blog Post-4

TRU Open Learning 1 Comment

For the Virtual Forest Tutorial i sampled the Snyder-Middleswarth Natural area using area-based systematic, random, and haphazard methods. The technique with the fastest sampling time was the area-based systematic approach with a sampling time of 12 hours, 5 minutes, while random and haphazard techniques had sampling times of 12 hours 34 minutes, and 12 hours 36 minutes, respectively. The least abundant species showed the most accuracy in results, White Pine had a 0% error for two different methods, random and haphazard sampling strategies.

In general, the area-based random sampling technique was the most accurate for each species. Area based systematic sampling showed the least accuracy in results. For the two most common species the systematic approach gave the best percent error results, while the haphazard sampling gave the worst results. Alternatively, the percent error for the two rarest species was best using the haphazard sampling technique. It seems that the systematic approach is more useful for large amounts of common species, while the haphazard sampling was more accurate for the rarer species.

 

Below is a list of percent error for the two most rare species, Red Maple and White Pine, as well as the most common, Eastern Hemlock and Sweet Birch, for each sampling strategy.

Systematic:

Eastern Hemlock-45.4%

Sweet Birch-20.6%

Red Maple-82.5%

White Pine-50.0%

Haphzard:

Eastern Hemlock-25.0%

Sweet Birch-6.38%

Red Maple-22.6%

White Pine-0%

Random:

Eastern Hemlock-9.1%

Sweet Birch-6.38%

Red Maple-1.85%

White Pine-0%

 

Blog Post 4: Sampling Strategies

TRU Open Learning 1 Comment

My results for the simulated forest sampling are below. While the haphazard technique was the fastest, it’s overall accuracy as abysmal. I found it interesting that for the most common and least common species, the random sample was the actually the most accurate, while the systematic approach was the most accurate for both the second most common and second least common methods. With the systematic approach, the accuracy dropped as the species became less common. As mentioned, this didn’t prove true with the random sample however, which was a lot more random in its results as it’s nature would suggest.

In conclusion a systematic sampling approach in this setting may work the best if you know your target species is fairly common while a random approach may be the best method if you are unsure. Based on this experiment I would definitely try to avoid haphazard sampling if it can be avoided.

Systematic Area Based  –

Easter Hemlock – Actual – 469.9  Estimated – 524   Error –  12%

Sweet Birch – Actual – 117.5 Estimated – 136  Error –   16 %

Stripped Maple – Actual – 17.5 Estimated – 12  Error –   31 %

White Pine – Actual – 8.4 Estimated – 12  Error –   43%

 

Random –

Easter Hemlock – Actual – 469.9 Estimated – 487.5  Error –   4 %

Sweet Birch – Actual – 117.5 Estimated – 137.5  Error –   17%

Stripped Maple – Actual – 17.5 Estimated – 0  Error –   100 %

White Pine – Actual – 8.4 Estimated – 8.3  Error –   1 %

 

Haphazard

Easter Hemlock – Actual – 469.9 Estimated – 579.2  Error – 23%

Sweet Birch – Actual – 117.5 Estimated – 275  Error –   134%

Stripped Maple – Actual – 17.5 Estimated – 29.2  Error –   66 %

White Pine – Actual – 8.4 Estimated – 20.8  Error –   147 %

 

 

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

The method I chose for the virtual forests tutorial was the area based method. The fastest sampling method was the systematic sampling method with a time of 12 hours 37 minutes, which was actually only within 20 minutes of the other methods. For the two most common species the systematic was the most accurate and was within a -7.2%  and 5.5% sampling error, compared to 63.4% and 100.9% error for the haphazard, and 11.2% and-29% error for the random method. For the rare species the haphazard sampling method was the most accurate of the three methods, although it still had a 52.4% error for the white pine and -8.5% for the Striped Maple. The random sampling method gave an almost 200% error for the White Pine and a 100% error for the Striped Maple, and the systematic sampling method gave a 174% and -100% error for the Striped Maple and White Pine respectively.

In general the accuracy declined with both the systematic and random methods as the species got more rare, and slightly increased in accuracy for the haphazard method for rare species. The Shannon-Weiner diversity was calculated to be the identical for the true diversity and the systematic method which leads me to believe that it was the most accurate method, along with having lower percentage error for most species present. For the random and haphazard methods the diversity was found to be 1.4 which is 0.1 lower than the true diversity. I don’t believe there was enough points to capture the diversity of species because in both the systematic and the random methods, one species was not observed at all. Even though the Shannon-Weiner diversity was found to be accurate for the systematic sampling method, I would consider increasing the number of samples in order to guard against inaccuracy.

Blog Post 4 – Sampling Strategies

TRU Open Learning Leave a comment

For the sampling strategy virtual forest tutorial, I selected area-based methods. The most efficient technique in terms of time spent sampling was systemic sampling technique, as it had the fastest estimated sampling time. The systematic strategy was the most accurate sampling strategy for the most common species and the haphazard strategy was most accurate for the least common species. However, the systematic strategy was only the most accurate for the most common species and the haphazard strategy was most accurate for the rest of the species. Accuracy declined as species became less abundant (from common to rare species). Overall, the haphazard strategy was most accurate. The random strategy was the least accurate. I believe that in this case, the haphazard strategy was most accurate because I chose the sampling points within each sampling zone (separated by transects) randomly, so the method that I deployed was similar to the stratified random sampling technique, which is the sampling technique that I will use in my field research.

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

In the virtual forest tutorial, I chose all area-based methods. The fastest sampling time was when using the systematic technique. It took an estimated time to sample of 12 hours and 4 minutes, compared to 12 hours 42 minutes for the randomized, and 12 hours 27 minutes for haphazard.

The most common species was the eastern hemlock. Below are my calculations for percentage error of this species.

Systematic: PE (495.8-469.9)/469.9*100=5.5%        Most accurate

Random: PE (680.8-469.9)/469.9*100=44.9%

Haphazard: PE (704.2-469.9)/469.9*100=49.9%

The least common species was the white pine. Below are my calculations for PE of this species.

Systematic: PE (8.3-8.4)/8.4*100=1.2%              Most accurate

Random: PE (8.3-8.4)/8.4*100=1.2%                  Most accurate

Haphazard: PE (4.2-8.4)/8.4*100=50%

It seems that calculating rare species is more accurate, but only when using random or systematic sampling. Haphazard sampling was not accurate in either species. For abundant species, systematic seems to be the only accurate sampling method.

The actual data compared to the estimated data left significant percentage errors in most cases for all species in the middle. Data was most accurate at the top and bottom, or most common and least common. I suggest more than 24 data samples would be needed to eliminate this.

 

 

Blog Post 4 – Sampling Strategies

TRU Open Learning Leave a comment

Blog Post 4 – 03-02-20

Out of the three sampling techniques, Random Sampling was the most efficient in terms of time with the sampling taking a time of 12 hours and 4 minutes. Haphazard Sampling was the second most efficient technique in terms of time with the sampling taking 12 hours and 32 minutes and Systematic Sampling was the least efficient in terms of time with the sampling taking 12 hours and 44 minutes to complete. For the Eastern Hemlock Systematic data, the percent error for the density was 9.959% while for the Random data the percent error was 18.813% and for the Haphazard data the percent error was 6.895%. For the Sweet Birch Systematic data, the percent error for the density was 18.468% while for the Random data set, the percent error was 4.255% and for the Haphazard data the percent error was 24.085%. For the Yellow Birch Systematic data, the percent error for the density was 27.273% while for the Random data the percent error was 19.651% and for the Haphazard data the percent error was 31.129%.  For the Chestnut Oak Systematic data, the percent error for density was 9.456% while for the Random data the percent error was 23.771% and for the Haphazard data the percent error was 19.086%. For the Red Maple Systematic data, the percent error for density was 8.915% while for the Random data the percent error was 36.922% and for the Haphazard data the percent error was 19.176%. For the Striped Maple Systematic data, the percent error for density was 28.571% while for the Random data the percent error was 76.0% and for the Haphazard data the percent error was 114.285%. For the White Pine Systematic data, the percent error for density was 10.0% while for the Random data the percent error was 10.0% and for the Haphazard data the percent error was 50.0%. Based off of these calculated values, the most accurate sampling strategy for common species (the most common being the Eastern Hemlock) was Haphazard sampling, followed by Systematic sampling. In comparing the two most common species, Haphazard sampling an Systematic sampling were very close in terms of accuracy, while Random sampling was not close. Based off of the aforementioned calculated values, the most accurate sampling strategy for rare species (the most rare being the White Pine) was very clearly Systematic sampling. The accuracy generally declined for rare species, especially in the Haphazard sampling data. This suggested that as species abundance lowered, accuracy in density measurements also lowered. Having 24 sample points perhaps was not sufficient to measure the number of species in the area. Perhaps 24 sample points was enough to provide an accurate measure of the common trees species, but it was not enough to provide an accurate measure for the number of rare tree species in the area; however, 24 sample points appeared to be enough to capture the relative abundance of each of the tree species in the area. 

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

Shannon Myles

February 1st, 2020

 

I used the distance-based sampling methods to measure tree species abundance in the tutorial. Table 1 below illustrates the data gathered.

Table 1 Comparison of three distance-based sampling methods used to calculate the abundance of tree species in the Snyder-Middleswarth Natural Area. 

Tree Species  Actual  Systematic  % error  Random  % error  Haphazard  % error 
Eastern Hemlock  469.9  277.3  40.99  441.8  5.98  485.0  3.21 
Sweet Birch  117.5  109.6  6.72  144.1  22.64  121.2  3.15 
Yellow Birch  108.9  70.9  34.89  105.6  3.03  83.9  22.96 
Chestnut Oak  87.5  38.7  55.77  115.2  31.66  93.3  6.63 
Red Maple  118.9  90.3  24.05  86.4  27.33  74.6  37.26 
Striped Maple  17.5  45.1  157.71  28.8  64.57  18.7  6.86 
White Pine  8.4  12.9  53.57  0.0  100  18.7  122.62 
Estimated time to sample  4h 18min  53.39  4h 40min   36.46  4h 38min   28.96 

 

Table 1 shows that the estimated times to perform the studies were somewhat similar, but that the fastest method to sample 24 plots was the systematic approach by approximately 20 minutes.

The comparison of measurements of abundance showed that the most precise technique for the most abundant species of tree (Eastern Hemlock) was the haphazard technique. With a % error of only 3.21, it surpasses the random sampling method in accuracy by less than 3%. Both the random and haphazard techniques were far more accurate than the systematic method, which led to a terrible % error of 40.99%. As for the second most abundant species of tree in the area (Red Maple), the best sampling method was systematic, with a % error of 24.05.

For the rarest species of tree (White Pine), the most accurate sampling method was found to be the systematic approach. With a relatively high 53.57% error, it is far more accurate than the two other methods, who have both surpassed 100% error. The abundance of the second least abundant tree (Striped Maple) was very accurately measured by the haphazard method. It scored a 6.86% error, comparatively to 64.57% and 157.71% for random and systematic.

Generally speaking, the accuracy of all three methods seemed to have diminished as species abundance became lower. Except for a few odd data points, the majority of % error was inversely proportional to the actual species abundance, no matter the sampling strategy.

Finally, the average % error of the haphazard method (28.96%) was lower than for random and systematic methods (36.46% and 53.39%). Therefore, I conclude that the distance-based haphazard method is the most accurate for sampling abundance in a forest.

Blog post 4: Sampling Strategies

TRU Open Learning Leave a comment

 

Haphazard Sampling of an Area

Percent error of 2 most common: Red Maple 1.5%, White oak 7% (percent error is [(E-T)/T]*100

Percent error of 2 least common: White Ash 100%, Yellow Birch 100%

Sampling time: 24 hours 13 minutes

Random Sampling of an area

Percent error of 2 most common: Red Maple 9%, White Oak 1.3%

Percent error of 2 least common: White Ash 100%, Yellow Birch 662.5%

Sampling time: 26 hours 1 minute

Systematic Sampling of an Area

Percent error of 2 most common: Red Maple 1%, White oak 5%

Percent error of 2 least common: White ash 150%, Yellow Birch 100%

Sampling time: 26 hours 40 minutes

 

 

Haphazard has the least amount of time which is surprising considering there is not necessarily a technique used to save time due to how irregular it can be. This may be a certain instance where it took less time to complete a sample. The least abundant species have the worst percent error, which makes sense as there are less to find in an area and can easily be missed in a study. The most abundant species have the lowest percent area, because there are enough of them to create a more accurate set of data. The lowest overall percent error would be in Systematic Sampling which could be explained by the fact that entire sections may not be missed as they would be in random or haphazard. Different gradients are more likely to be covered with a systematic sample.

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

Systematic sampling (area):

Sampling time= 12 hrs 7 min

Hemlock= 637.5….Percentage error= (35.67%)

Red Maple= 116.7…Percentage error= (1.85%)

White Pine= 8.3…Percentage error= (1.19%)

Striped Maple= 12.5…Percentage error= (28.57%)

I was surprised when I found that the most abundant species was represented so inaccurately while the least abundant was very accurate. I believe this may have occurred because the quadrats followed a very specific gradient going south to north, therefore we miss out on the other species that may be more present to the east or to the west of our selected quadrats.

Random sampling (area):

Sampling time= 12 hrs 42 mins

Hemlock= 420.8….Percentage error= (10.45%)

Red Maple= 100…Percentage error= (15.91%)

White Pine= 16.7…Percentage error= (98.81%)

Striped Maple= 20.8…Percentage error= (18.96%)

Overall this was even more inaccurate than the systematic sampling, especially with the rarest species- White Pine. By chance, the program sampled double the proportions of White Pine than are actually in the forest which is surprising. The more abundant species were more accurately represented.

Haphazard sampling (area):

Sampling time= 12 hrs 46 mins

Hemlock= 420.8….Percentage error= (10.45%)

Red Maple= 104.2…Percentage error= (12.4%)

White Pine= 0…Percentage error= (100%)

Striped Maple= 12.5…Percentage error= (28.60%)

This was the most inaccurate of the sampling techniques with not a single White Pine being sampled. This does not surprise me as their actual representation is quite low and haphazardly choosing quadrats without attention to the different gradients could easily lead to this result. The species in abundance were more accurately represented which is what I would expect.

Conclusion:

The systematic sampling technique had the fastest sampling time and was the most accurate. As long as the entire sampling area had similar environmental factors such as sunlight exposure, space, soil type/quality, etc then I feel this would be the best technique to use. If the environment was more diverse, more sampling points would be necessary in order to correctly represent the gradients. Either way, an increase in sample points would have been beneficial as I think would always be the case, but then that is more time consuming.

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

The fastest technique was Distance: Haphazard which completed in a total of 11 minutes. The slowest was Area: Haphazard and Area random at 33 and 32 minutes each.

The most accurate method seemed to be Distance: Haphazard. Species that were found to be in high abundance seemed to have very large percent errors (300-900%).

Comparing the relative frequency of our sample to the actual value for the 2 most common and 2 rarest species, gave the following data:

Area: Haphazard:

Black Tupelo: 20 vs 4.2 = 376% error

Red Maple: 20.0 vs 35.0 = 42.85% error

White Oak: 20.0 vs 13.5 = 48.15% error

Chestnut Oak: 20.0 vs 10.8 = 85.19% error

Average error = 138.05%

33 minutes

Distance: Haphazard:

11 minutes:

Red Maple: 50 vs 35 = 42.86% error

White Hazel: 50 vs 13.8 = 262% error

Total % error: 0.6/1.8 = 66.67% error

Average: 123.78% error

Area: Random:

32 minutes

Eastern Hemlock: 20 vs 4.6 = 335% error

Chestnut Oak: 19.2 vs 9.4 = 104% error

Witch Hazel: 20 vs 13.8 = 45% error

Red Maple: 20 vs 35 = 43% error

Average = 131.75% error

Distance: Random:

12 minutes

White Oak: 25 vs 13.5 = 85% error

Witch Hazel: 25 vs 13.8 = 81% error

Red/Black Oaks: 25 vs  9.2 = 172% error

Sowny Juneberry: 25 vs 2.3 = 987% error

Average = 331.25% error