Category: Post 4: Sampling Strategies

Post 4: Sampling Strategies

TRU Open Learning Leave a comment

In the online sampling simulation, I chose to sample the Mohn Mill site using distance sampling. I found that systematic distance sampling had the fastest estimated sampling time with 4 hours and 7 minutes.

The percent error for species density of the two most common and two rarest species at the Mohn Mill site using distance systematic sampling is as follows:

red maple = |(392.4-403.7)/403.7|x100 = 2.799%

white oak = |(49.9-74.5)/74.5|x100 = 33.0%

yellow birch = |(0.0-0.8)/0.8|x100 = 100%

white ash = |(0.0-0.8)/0.8|x100 = 100%

The percent error for species density for the same species as above using distance random sampling is as follows:

red maple = |(380.8-403.7)/403.7|x100 = 5.673%

white oak = |(53.3-74.5)/74.5|x100 = 28.5%

yellow birch = |(7.6-0.8)/0.8|x100 = 850%

white ash = |(0.0-0.8)/0.8|x100 = 100%

The percent error for species density for the same species as above using distance haphazard sampling is as follows:

red maple = |(735.5-403.7)/403.7|x100 = 82.19%

white oak = |(157.6-74.5)/74.5|x100 = 112%

yellow birch = |(0.0-0.8)/0.8|x100 = 100%

white ash = |(0.0-0.8)/0.8|x100 = 100%

The more abundant the species, the more accurate the sampling. Conversely, the less abundant the species, the less accurate the sampling.

Systematic sampling appears to be the most accurate sampling method, followed by random sampling, and finally haphazard sampling.

BLOG POST 4

TRU Open Learning Leave a comment

The three sampling strategies used in the virtual forest tutorial are haphazard sampling, random sampling and systematic sampling. Haphazard sampling uses samples that are readily available, these samples are almost never random samples. Random sampling ensures that there is equal chance of being sampled. Systematic sampling avoids bias compared to haphazard sampling and is easier compared to random sampling.

Haphazard sampling had the fastest estimated sampling time. This is because there is less travel time between sample points.

The overall accuracy of the species that were more common was greater than the species that were less common.

The systemic sampling and random sampling strategy were the most accurate because the areas that were analyzed did not overlap.

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

During the virtual forest tutorial, three sampling strategies were used to compare the abundance of different tree species throughout Middleswarth Natural Area. These strategies were the haphazard strategy, the random sampling strategy, and the systematic sampling strategy. I chose to sample by area not distance for the tutorial. 

After completing the tutorial, it is estimated that the haphazard strategy was fastest for sampling (12 hours, 29 minutes) followed by the systematic sampling strategy (12 hours, 37 minutes), and the random sampling strategy taking the most time (12 hours, 42 minutes). 

For the systematic sampling strategy, the percent errors for the two most common species were 20.2% and 31.2%, while the percent error for both of the two least common species was 100%. For the random sampling strategy, the percent errors for the two most common trees were 4.6% and 8.9%, while the percent error for the two rarest species were 100% and 50%. Finally, the percent errors for the two most common species using the haphazard strategy were 7.8% and 40.5%. The percent errors for the two most rare species were 4.6% and 100%. A percent error value of 100% occurred when no trees of those species were found during the sampling. From this, it can be observed that the lowest percent errors were found by using the random sampling strategy on the most common tree types, while the largest percent errors occurred when systematically sampling the least common species. Even more generally, the systematic sampling strategy yielded an overall percent error of 62.9%, the random sampling strategy yielded 40.9%, and the haphazard strategy yielded 38.2%. From these results, if one strategy had to be used, the haphazard strategy would yield the most accurate data on average.

Generally, as the abundance of a certain species decreased, the percent error increased. This indicated a decrease in accuracy, as the degree of closeness between measured and true values decreased. Conversely, the more accurate data was found in the samples of the most abundant tree types. The most accurate strategy for sampling the most common tree species was the random sampling strategy, and the most accurate strategy for sampling the rarest tree species was the haphazard strategy. It should be noted that the average percent error for the rare species was still much larger than the average for common species. Thus, it is important to understand the expected density and abundance of the species of interest to choose the sampling strategy that will yield the most accurate data. From this tutorial, it is evident that best strategy may not be the same for sampling all tree types. Ensuring the proper strategy is used will allow for sampling data to best represent the population data. 

Post 4: Sampling Strategies

TRU Open Learning Leave a comment

The technique that is most time-efficient for area sampling is haphazard, with a total time of 12 hours and 29 minutes. Systemic sampling took 12 hours and 36 minutes, and random sampling took 12 hours and 45 minutes.

Eastern Hemlock – Most Common

  • Actual Density: 469.9
  • Systemic Sampling: 524.0; Percentage Error: 11.5%
  • Random: 666.7; Percentage Error: 41.9%
  • Haphazard: 650.0; Percentage Error: 38.3%

 Red Maple

  • Actual Density: 118.9
  • Systemic Sampling: 100.0; Percentage Error: 15.9%
  • Random: 79.2; Percentage Error: 33.4%
  • Haphazard: 112.5; Percentage Error: 5.4%

 White Pine – Most Rare

  • Actual Density: 8.4
  • Systemic Sampling: 12.0; Percentage Error: 42.9%
  • Random: 8.3; Percentage Error: 1.2%
  • Haphazard: 0.0; Percentage Error: 100%

Striped Maple

  • Actual Density: 17.5
  • Systemic Sampling: 12.0; Percentage Error: 31.3%
  • Random: 66.7; Percentage Error: 281.1%
  • Haphazard: 20.8; Percentage Error: 18.9%

Sweet Birch

  • Actual Density: 117.5
  • Systemic Sampling: 136.0; Percentage Error: 15.7%
  • Random: 170.8; Percentage Error: 45.4%
  • Haphazard: 183.3; Percentage Error: 56.0%

Yellow Birch

  • Actual Density: 108.9
  • Systemic Sampling: 128.0; Percentage Error: 17.5%
  • Random: 70.8; Percentage Error: 35.0%
  • Haphazard: 162.5; Percentage Error: 49.2%

Chestnut Oak

  • Actual Density: 87.5
  • Systemic Sampling: 104.0; Percentage Error: 18.9%
  • Random: 54.2; Percentage Error: 38.0%
  • Haphazard: 54.2; Percentage Error: 38.0%

When comparing the two most common species, the most accurate sampling strategy for Eastern Hemlock was systemic sampling with a percentage error of 11.5%. The most accurate sampling strategy for Red Maple was haphazard sampling, with a percentage error of 5.4%. If you average the percentage errors of the two species of systemic vs. haphazard sampling, the average error of systemic is 13.7% and haphazard is 21.9%. Therefore, the most accurate sampling strategy is systemic sampling.

When comparing the two most rare species, the most accurate sampling strategy for White Pine was random sampling with a percentage error of 1.2%. The most accurate strategy for the Striped Maple was haphazard sampling with a percentage error of 18.9%. Comparing the averages of the percentage errors for these two strategies, the most accurate sampling strategy is haphazard with an average percentage error of 59.5%. The average percentage error of random sampling is 141.2%. The accuracy of the sampling appears to decline with rarer species, as evidenced by the increase in the percentage error averages.

 I do not believe that 24 was a sufficient number of sample points. As the density of the tree species varies so greatly throughout the study area, using this number of points resulted in missing some species in the sample that were present. For example, although the density of White Pine trees was 8.4, with haphazard sampling this species was not sampled resulting in a 100% percentage error.

Blog Post 4 – Sampling Strategies

TRU Open Learning Leave a comment

For the virtual forest tutorial, I used the area-based model for my systematic, randomized and haphazard sampling of vegetation. The technique which required the least amount of time was the haphazard method. However, all sampling techniques had very similar sampling times around 12.5 hours. 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. Based on the percent error data for all six species the haphazard method presented the best results on average, followed by the systematic and the randomized approach.

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

Adrienne Burns

August 21, 2019

 

The first sampling method I conducted was the ‘Area; Systematic” approach. I choose one randomized number ‘y’ axis, and received my subsequent data from adding 10 to each of the ‘x’ axis and alternating between the randomized number and ‘y+10’. The density data from this sampling method has some very accurate of methods for certain species of trees, but very inaccurate data for others. When comparing it to the ‘Actual density’ data, for example, the actual density of Sweet Birch was 117.5, and sampling density found 116, and for White Pines species, the actual density was 8.4 and the data showed a density of 28.0. I found it interesting that for the species Striped Maple, this method did not count any of the trees. The density for Striped Maple was 17.5 and this method accounted for 0.0. This type of sampling didn’t correctly depict the distribution of tree species over the entire forest area. Also, the ‘Area; Systematic’ method took a long time to complete. It took 12 hours and 35 minutes to complete the sampling.

 

The second method was the ‘Distance; Random’ sampling technique. This method had given me 24 random ‘x, y’ axis to sample. It was the fasting sampling method which took 4hr 38minutes. This would be the preferred method of sampling if the ecologist had time constraints. Along with the first method this one also showed varying correctness for the distribution of the trees. For example the actual Hemlock density was 469.9 and the data showed a density of 445.1, but for the Red Maple the actual density was 118.9, but the data showed 145.2. Of all 3 of the sampling methods I used, the ‘Distance; Random’ technique was the most accurate especially with regards to frequency.

 

The last method, ‘Area; Haphazard’ took the longest timeframe to complete 13h and 1minute. It also had the largest variation in results. For instance, Eastern Hemlock actual density was 469.9. Both ‘Area: Systematic’ and ‘Distance; Random’ data were close in proximity to 440.0 and 445.1, yet the ‘Area; Haphazard’ showed 669. As it had the largest variation and the longest timeframe, I would need to seriously consider I was going to use this method.

 

None of the methods were very accurate. All had some tree species data that was accurate, and others that were far from the actual data.

 

Error Percentage Eastern Hemlock

 

‘Area; Systematic’

(440-469.9)/ 469.9*100 = 6.36% Error

 

‘Distance; Random’

(445.1-469.9)/469.9*100 = 5.28% Error

 

‘Area; Haphazard’

(664.0-469.9)/469.9*100 =41.31% Error

 

 

Error Percentage White Pine

 

‘Area; Systematic’

(28-8.4)/8.4*100 = 233.33% Error

 

‘Distance; Random’

(9.7-8.4)/8.4*100 = 15.48% Error

 

 

‘Area; Haphazard’

(4.0-8.4)/8.4*100 = 52.38% Error

 

 

 

Blogpost 4: Sampling strategies

TRU Open Learning Leave a comment

The sampling strategy in the virtual forest tutorial that had the fasted estimated sampling time was haphazard sampling. For one of the two rarest species, White Pine, haphazard sampling had the most accurate results with a 1.2% error. The other rarest species, Striped Maple, has no accurate sampling strategy with this tutorial (all sampling strategies had >100% error).

For the most common species, Eastern Hemlock, systematic sampling narrowly beat out random sampling with 20% error (compared to 20.6% random sampling). For the second most common species, the Red Maple, systematic sampling had a much wider margin on accuracy with a 32.7% error (as opposed to 53% for both other strategies).

 

Species abundance did not seem to have a massive effect on accuracy; the lowest percent error was for a rare species. This leads me to believe more replicate samples should be done in a study like this for a better representative sampling.

 

Overall, it seems that a systematic sampling strategy had the lowest percent error for more species than the other two sampling strategies. Haphazard sampling yielded the lowest percent error for a few species and random samples did not produce the lowest percent error for any of the species in the tutorial.

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

Sampling Using Virtual Forests (Mohn Mill – Sampling by Area)

Compare the results. Which technique is the most efficient in terms of time spent sampling? Compare the actual densities with the estimated (data) densities from your sampling. Calculate the percentage error for the different sampling techniques for both common and rare species. What was the most accurate sampling strategy for common species? What was the most accurate for rare species? Did the accuracy stay the same or decline for rare species? Was 24 a sufficient number of sample points to capture the number of species in this community? Was it enough sample points to accurately estimate the abundance of these species?

 

  • Which technique is the most efficient in terms of time spent sampling?

 

Time Spent Sampling:

Area Haphazard (t= 12:45, m=765)

Area Random (t=12:09, m=729)

Area Systematic (t=12:44, m=764)

In terms of time spent sampling area, random sampling was the most efficient by approximately 9.5% compared to both haphazard and systematic sampling which differed by only 1 minute.

 

  • Calculate the percentage error for the different sampling techniques for both common and rare species:

 

Red Maple Density

Actual (403.7); Area Haphazard Data (337.5) – Percentage Error= 16.4%

Actual (403.7); Area Random (425.0) – Percentage Error= 5.3%

Actual (403.7); Area Systematic Data (445.8) – Percentage Error= 10.4%

White Oak Density

Actual (74.5); Area Haphazard Data (41.7) – Percentage Error= 44%

Actual (74.5); Area Random Data (54.2) – Percentage Error= 27.24%

Actual (74.5); Area Systematic Data (125.0) – Percentage Error= 67.78%

Chestnut Oak Density

Actual (82.9); Area Haphazard Data (54.2) – Percentage Error= 34.62%

Actual (82.9); Area Random Data (91.7) – Percentage Error= 10.62%

Actual (82.9); Area Systematic Data (41.7) – Percentage Error= 49.69%

 

Witch Hazel Density

Actual (142.4); Area Haphazard Data (191.7) – Percentage Error= 34.62%

Actual (142.4); Area Random Data (166.7) – Percentage Error= 17.06%

Actual (142.4); Area Systematic Data (150.0) – Percentage Error= 5.33%

Red/Black Oak Density

Actual (46.7); Area Haphazard Data (29.2) – Percentage Error= 37.47%

Actual (46.7); Area Random Data (58.3) – Percentage Error= 24.84%

Actual (46.7); Area Systematic Data (0.0) – Percentage Error= 100%

Eastern Hemlock Density

Actual (45.6); Area Haphazard Data (8.3) – Percentage Error= 81.8%

Actual (45.6); Area Random Data (62.5) – Percentage Error= 37.06%

Actual (45.6); Area Systematic Data (41.7) – Percentage Error= 8.55%

Black Tupelo Density

Actual (35.5); Area Haphazard Data (33.3) – Percentage Error= 6.2%

Actual (35.5); Area Random Data (0.0) – Percentage Error= 100%

Actual (35.5); Area Systematic Data (33.3) – Percentage Error= 6.2%

White Pine Density

Actual (12.8); Area Haphazard Data (4.2) – Percentage Error= 67.2%

Actual (12.8); Area Random Data (12.5) – Percentage Error= 2.34%

Actual (12.8); Area Systematic Data (12.5) – Percentage Error= 2.34%

Downy Juneberry Density

Actual (9.9); Area Haphazard Data (8.3) – Percentage Error= 16.16%

Actual (9.9); Area Random Data (12.5) – Percentage Error= 26.26%

Actual (9.9); Area Systematic Data (16.7) – Percentage Error= 68.68%

Striped Maple Density

Actual (13.6); Area Haphazard Data (0.0) – Percentage Error= 100%

Actual (13.6); Area Random Data (4.2) – Percentage Error= 69.12%

Actual (13.6); Area Systematic Data (0.0) – Percentage Error= 100%

 

Hawthorn Density

Actual (4.5); Area Haphazard Data (0.0) – Percentage Error= 100%

Actual (4.5); Area Random Data (0.0) – Percentage Error= 100%

Actual (4.5); Area Systematic Data (4.2) – Percentage Error= 6.6%

Black Cherry Density

Actual (1.5); Area Haphazard Data (0.0) – Percentage Error= 100%

Actual (1.5); Area Random Data (0.0) – Percentage Error= 100%

Actual (1.5); Area Systematic Data (0.0) – Percentage Error=100%

Sweet Birch Density

Actual (1.2); Area Haphazard Data (0.0) – Percentage Error= 100%

Actual (1.2); Area Random Data (4.0) – Percentage Error= 233%

Actual (1.2); Area Systematic Data (0.0) – Percentage Error= 100%

American Basswood Density

Actual (1.5); Area Haphazard (0.0) – Percentage Error= 100%

Actual (1.5); Area Random Data (0.0) – Percentage Error= 100%

Actual (1.5); Area Systematic Data (0.0) – Percentage Error= 100%

Yellow Birch Density

Actual (0.8); Area Haphazard Data (0.0) – Percentage Error= 100%

Actual (0.8); Area Random Data (4.2) – Percentage Error= 425%

Actual (0.8); Area Systematic Data (16.7) – Percentage Error= 1987.5%

White Ash Density

Actual (0.8); Area Haphazard Data (0.0) – Percentage Error= (0.0-0.8)/0.8*100= 100%

Actual (0.8); Area Random Data (0.0) – Percentage Error= (0.0-0.8)/0.8*100= 100%

Actual (0.8); Area Systematic Data (0.0) – Percentage Error= (0.0-0.8)/0.8*100= 100%

 

  • What was the most accurate sampling strategy for common species?

The most accurate sampling strategy for common species overall tended to be random sampling.

 

  • What was the most accurate for rare species?

All the sampling methods were relatively poor for rare species, either missing them entirely or grossly overrepresenting them.  For example, both White Ash and American Basswood were missed entirely regardless of sampling method with consistent percent error of 100%.  Similarly, a mixture of missing entirely and grossly overestimating density was observed for both Yellow Birch and Sweet Birch, regardless of sampling method.

 

  • Did the accuracy stay the same or decline for rare species?

Accuracy for rare species declined drastically in almost all cases.

 

  • Was 24 a sufficient number of sample points to capture the number of species in this community?

 

Overall no, 24 was not a enough sample points to capture the number of species in this community.  Based on the analyses, using the sampling methods employed here with 24 sample points, I would have entirely missed White Ash, American Basswood and Black Cherry.

 

  • Was it enough sample points to accurately estimate the abundance of these species?

Overall no, some species were very accurately measured with some sampling methods with 24 sample points while others were very inaccurately measured.

 

 

 

 

Blog Post 4!

TRU Open Learning Leave a comment

This post describes the results from the virtual forest tutorial!

Result Summary- All Sampling Techniques (Here is a link to visually show my results).

The shortest estimated time to sample was 4 hours and 49 minutes for the distance, random or systematic.

Comparing percentage error between species for the most and least common species:

Red maple: (distance systematic, area haphazard, area random) 24.8%, 12.5%, 6.3%

White Oak: (distance systematic, area haphazard, area random) 50%, 4.9%, 51%

Yellow birch:(distance systematic, area haphazard, area random) 100%, 100%, 100%

White ash:(distance systematic, area haphazard, area random) 100%, 100%, 100%

Most accurate for the most abundant species: Area, random

Least abundant: American basswood was only picked up on the distance systematic. The accuracy decline with the rare species compared to the more dense species.

I propose that more than 24 sample points would have given a better overview of the area. There were rare species that were not sampled so with 40 points they might have been. 24 was enough sample points to measure the larger species but for the smaller ones more points should be used.

BLOG POST 4

TRU Open Learning Leave a comment

In the virtual forest exercise, of all the three sampling methods on which I performed the sampling analysis, the haphazard sampling technique was the shortest with an estimated  time of 1hr, followed by the area random or systemic which was 11 hours , 29 minutes and then a systemic sampling along a topographic gradient at 21 hours 23 minutes.

Of the two most common species, Red Maple and Witch Hazel, the hapharzard sampling had a percentage error of 0.9% for Red Maple and 0.3% for Witch Hazel. With random sampling, the percentage error was 1.4% for Red Maple and 1.5% for Witch hazel. With the systemic sampling along a topographic area, the Red Maple had a percentage error of 0.92% while the Witch Hazel had a percentage error of 1.53%.  for these two common species, it would appear that the Haphazard sampling method had the lowest percentage error for both the Red Maple and the Witch Hazel.  If you average the percentage error of the two sample techniques, the haphazard sampling method had the lowest percentage error for the common species of trees.

Of the two rarest species, White Ash and Yellow Birch the percentage error for the systematic sampling technique was -100% for both species. Random sampling had a percentage error of -100% for White Ash and 5.6% for Yellow Birch. For haphazard or subjective sampling, the percentage error for White Ash was 52.8% and -100% for Yellow Birch. The systemic sampling techniques failed to record any occurrences of both the White Ash and Yellow Birch trees. Random sampling had the smallest percentage error for White Ash making it the most effective sampling technique for this rare species.

Accuracy for all species was relatively consistent with the 3 sampling techniques except for the rarest species of the Yellow Birch and White Ash that was not detected with all methods. The most abundant species were more accurate while the least accurate was the haphazard sampling technique for White Ash which was at 52.8%.

Based on the results, all three sampling techniques showed fairly consistent results although random sampling appeared to be the most effective for the most abundant tree type. For the Yellow Birch, the random sampling method was the most effective with only a 5.6% percentage error. The White Ash tree had a percentage error of 52.8% with the haphazard sampling method which was the highest percent error for the rarest trees even though this method was the fastest method.