Category: Post 4: Sampling Strategies

Blog post #4: Virtual Forest Sampling Techniques Simulator

TRU Open Learning Leave a comment

In the virtual forest sampling simulator the random sampling was significantly faster than systematic and haphazard sampling.

Systematic –  12 hours, 35 minutes

Random –  4 hours, 24 minutes

Haphazard – 12 hours, 30 minutes

  Common Species Rare Species
Sampling Technique

 % error

 Eastern Hemlock Sweet Birch Striped Maple White Pine
Systematic 17.2% 30.0% 126.8% 138.1%
Random 16.6 28.2% 100.0% 25.0%
Haphazard 8.2% 31.2% 18.9% 99.0%

In comparing the percent error of the three sampling techniques with the exception of white pine in the random sampling,  the percent error increased with decreasing species abundance. Based on this sampling simulator, the haphazard sampling technique is the most accurate. Although, it should be noted that the random sampling technique had similar percent error values. Systematic sampling had notably higher percent errors for species with low abundances.

Blog Post 4, Sampling Strategies

TRU Open Learning Leave a comment

The 3 sampling strategies learnt in the virtual forests tutorial are systematic, random, and haphazard sampling. Throughout the virtual forest tutorial I learned which were the most efficient, fastest, and most accurate. I sampled the Snyder-Middleswarth Natural Area using area sampling technique rather than the distance sampling technique.

 

Systematic Sampling –

A combination of both random and haphazard sampling. It is easier than random but has less of a bias than haphazard. Quadrats were chosen in a specific pattern across the location, usually a gradient. On the virtual forest tutorial, systematic was slower than haphazard, but faster than random sampling. However it did have a high percent error as follows:

Eastern Hemlock – 14.0%

Red Maple – 5.4%

Striped Maple – 138.3%

White Pine – 147.6%

 

Random Sampling –

Done by labelling quadrats and choosing the numbered quadrats at random, every quadrat has an equal chance at getting chosen. This takes the longest time to do to ensure everything is chosen at random. Random sampling did have the lowest percent error, making it the most accurate way of sampling.

Eastern Hemlock – 6.4%

Red Maple – 5.4%

Striped Maple – 66.8%

White Pine – 100.0%

 

Haphazard Sampling –

Done by choosing areas that have samples which are readily available, and taking samples from the different variations in your testing area. Haphazard samples are never random but always available, and therefore also have a high percent error.

Eastern Hemlock – 23.3%

Red Maple – 43.65%

Striped Maple – 4.6%

White Pine – 1.2%

 

Sampling speed:

Haphazard > Systematic > Random

 

Accuracy in 2 most common species:

Random > Systematic > Haphazard

 

Accuracy in 2 most rare species:

Haphazard > Random > Systematic

 

The accuracy did change with abundance of the species. Random sampling was the most accurate for common species, and haphazard was the most accurate for rare species. Overall, if time allows, random sampling would be the most accurate.

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

Blog Post 4: Sampling Strategies

The results of the sampling strategies for the virtual forest tutorial are summarized in Table 1 at the bottom of this post.

The sampling strategies did not vary greatly in terms of time spent sampling, however, the most efficient was the systematic sampling method.  This method took 12 hours, 38 minutes as opposed to 12 hours, 43 minutes and 12 hours, 55 minutes for the other two sampling methods.

The accuracy of sampling varied for the most and least common species, depending on the sampling strategy used.  The most accurate sampling strategy for the most common species (i.e. eastern hemlock) was the systematic method.  The most accurate sampling strategy for the rarest species (i.e. white pine and striped maple) was the haphazard sampling method.  The accuracy was lower for the rare species across all sampling strategies.  The range of percent error across sampling strategies for eastern hemlock was 10.66-22.37 %, while for white pine this jumped to 42.86-52.38 %.

Overall, the haphazard sampling strategy was the most accurate as this had the smallest percent error averaged across the species (4.03), while the percent errors for the systematic and random sampling strategies were 17.92 and 29.87, respectively.

Table 1: Results of the virtual forest sampling strategies

Sampling Strategy: Systematic Random Haphazard
Species Actual Estimated % error Estimated % error Estimated % error
Eastern hemlock 469.9 520 10.66 575 22.37 540 14.92
Sweet birch 117.5 124 5.53 79.2 -32.60 108 -8.09
Yellow birch 108.9 92 -15.52 58.3 -46.46 116 6.52
Chestnut oak 87.5 92 5.14 62.5 -28.57 92 5.14
Red maple 118.9 144 21.11 150 26.16 144 21.11
Striped maple 17.5 0 -100.00 0 -100.00 8 -54.29
White pine 8.4 4 -52.38 4.2 -50.00 12 42.86

 

 

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

Three different sampling techniques were used for today’s blog assignment in the virtual forest tutorial. These three sampling techniques were; Systematic, Random & Haphazard. 24 samples were collected in each.

The sample results are as follows, with Systematic taking the lead as the fastest estimated sampling time, by over one hour.

 

Systematic: 4 hrs, 3 min

Random: 5 hrs, 18 min

Haphazard: 12 hrs, 25 min

 

Comparision of % error between 2 common & 2 rare tree species:

Eastern Hemlock (Common), Sweet Birch (Common), Striped Maple (Rare), White Pine (Rare);

Systematic: 1.2%, 11%, 26%, 10%

Random: 17%, 6%, 10%, 10%

Haphazard: 13.9%, 3.8%, 52.6%, 48.8%

The accuracy did change with species abundance. The sampling strategy that I found to be the most accurate was the systematic strategy, as the comparison % errors are low across the different samples, & the samples were spread out over a large area, allowing for more accurate comparisons.

 

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

In the virtual forest tutorial, the most efficient sampling method in terms of time-spent sampling was systematic sampling as seen in the results below, however the overall difference is marginal:

  • Systematic – 12 hours, 5 minutes
  • Random – 12 hours, 46 minutes
  • Haphazard – 12 hours, 29 minutes

Percent Error calculations for top two most common and most rare species:

Most Common Species Most Rare Species
Eastern Hemlock

(RF = 33.8%)

 Sweet Birch

(RF = 19.9%)

 Striped Maple

(RF = 2.8%)

 White Pine

(RF = 1.9%)
Systematic 4.23% 14.89% 100% 98.80%
Random 10.30% 25.53% 28.57% 50%
Haphazard 0.68% 17.02% 42.85% 1.19%

*RF = Relative Frequency

For the two most common species combined the most accurate sample strategy appears to be systematic sampling (i.e.: lowest percent error calculations for both Eastern hemlock and sweet birch at 4.23% and 14.89% respectively).  It should be noted that if just examining the single most common species that haphazard sampling appears to be the most accurate with Eastern hemlock receiving a remarkably low percent error of 0.68%, this is likely due to its high relative frequency in the forest. However, I would not conclude that haphazard is the best sampling strategy for common species in general since when examining the the top two most common species together systematic sampling appears to be the most accurate.

For the two most rare species combined the most accurate sampling strategy appears to be random sampling on average with striped maple having a percent error of 28.57% and white pine a percent error of 50%. Again it should be noted that haphazard sampling produced a remarkably low percent error for white pine at 1.19% but this accuracy was not reflected in the second least common species, striped maple. Overall haphazard sampling as the most accurate strategy for rare species makes very little sense since species with a very low relative frequency are statistically unlikely to be found identified with this sampling strategy and actually haphazardly sampling the correct plots to get such a low percent error repeatedly is very unlikely . That is why I would not conclude that haphazard sampling it is the best sampling strategy despite this low percent error.

Overall the accuracy declined with more rare species except for the outlier of white pine with the haphazard sampling strategy.

In total 7 species were captured with all species being captured in each sampling strategy expect for stripped maple which was not captured during the systematic sampling technique. As a result I would say that 24 samples were enough to capture the number of species in this community since almost all species were captured during all sampling techniques. However, I don’t think it was accurate enough to capture the abundance of species in each community as is shown in the percent error for white pine whose percent error varies widely from 1.19% to 50% to 98.8%.

 

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

The three sampling strategies that I used in the virtual forest tutorial were haphazard sampling (area), random sampling (area), and systematic sampling (area). The fastest estimated sampling time was the systematic sampling strategy (12 hours 36 minutes), as it only covered a small area within the region. With regards to the rare species, the fasted estimated sampling time was determined to be the haphazard sampling strategy, although it had a poor accuracy (1 hours 52 minutes).

With regards to the systematic sampling strategy, the common species had percentage errors of 1.8% and 5.3%. The rare species had percentage errors of 100% and 82%. With regards to the haphazard sampling strategy, the common species had percentage errors of 23.4% and 26.7%, and the rare species had percentage errors of 48.6% and 32.4%. Lastly, with regards to the random sampling strategy, the common species has percentage errors of 5.4% and 8.3% and the rare species had percentage errors of 32.7% and 48.6%.

Upon comparison of the percentage errors for the common and rare species based on the three different sampling strategies used, it can be concluded that the rare species had higher percentage errors than the common species in all three strategies. As such, it can be suggested that accuracy increases as abundance increases. Based on the results, it can also be concluded that the systematic strategy is the most accurate, whereas the haphazard strategy is the least accurate from the three.

Blog Post 4: Sampling Strategies

TRU Open Learning 1 Comment

The technique with the fastest sampling time was systemic (sampling along a gradient).

Systemic: 12hr 7m

Random: 12hr 40m

Haphazard: 13hr 1m

% Error respectively

Eastern Hemlock, Sweet Birch, Striped Maple, White Pine.

Systemic: 17.9%, 46.8%, 28.6%, 197.6%

Random: 14.0%, 68.1%, 100%, 98.8%

Haphazard: 2.15%, 2.13%, 54.3%, 42.9%

Accuracy decreases with decreasing species abundance. As seen above, the first two numbers represent common species accuracy whereas the second two numbers represent rare species accuracy.

Based on the data I collected, haphazard seems to be the most accurate. This was likely due to my selection of quadrants to sample. I chose quadrants that were spread out and sampled all regions of the area. Haphazard could have decreased in accuracy if I chose to select quadrants that did not represent the whole area.

Blog post #4 – Virtual sampling tutorial

TRU Open Learning Leave a comment

For my virtual forest sampling tutorial I used the area-based method. All my sampling times were similar in length and quite long:

Systematic: 12 hours 37 mins

Random: 12 hours 41 mins

Haphazard: 12 hours 26 mins

It is difficult for me to draw a conclusion on fastest time since they are all so similar. I assume the haphazard would’ve been faster if I had chosen a criteria rather than just randomly sampling without bias all over the whole map. This probably made it similar to the random technique time.

Here are the percent errors of the two most common and rarest species for each technique:

Systematic:

Common: 13.2% and 48.9%

Rare: 31.4% and 186.7%

Random:

Common: 3.5% and 45.3%

Rare:22.6% and 45.3%

Haphazard:

Common: 8.17% and 9.96%

Rare: 197.1% and 4.57%

When I calculated the mean error of each technique systematic had the highest mean error followed by haphazard, and the random technique had the lowest mean error.

As for accuracy related to species abundance, it does appear that the rare species had a higher rate of error.

Blog Post 4. Sampling

TRU Open Learning Leave a comment

In the virtual forest tutorial presented in the module 3 tutorial I chose Mohn Mill forest. The tree sampling techniques I used were Area: Sampling along a topographic gradient, Random sampling and Haphazard or subjective sampling. Sampling along a topographic gradient appears to be the fastest one a and took 12 hours, 36 minutes and Random sampling was the slowest and took 13 hours, 10 minutes. Even though, the time spent on sampling was almost equal Haphazard technique had shown significantly lower percent error for the most common species.

 

Species           Technique used             Estimated                Actual              Percent error

Red maple        Systematic sampling        420                        403.74             4.04%

Random                               416                        403.74             3.05%

Haphazard                     404                      403.74          0.075%

Which hazel     Systematic sampling        140                         142.4               1.7%

Random                                76                         142.4               45.2%

Haphazard                     149                       142.4               4.7%

 

Systematic technique appears to have the most precise data for the rarest species.

 

Species              Technique used             Estimated                Actual              Percent error

 

Downy              Systematic sampling       8                      9.9                  19.2%

juneberry         Random                                 20                        9.9                   102%

Haphazard                              4                         9.9                   59%

Black Tupelo   Systematic sampling     24                      35                    31.4%

Random                                  26.9                     35                    23.14%

Haphazard                             60                         35                    71.4%

 

 

 

Blog Post 4: Sampling Strategies

TRU Open Learning Leave a comment

In the virtual forest tutorial I used three different sampling techniques. All samples were collected using area mapping. The first was systematic sampling along a transect, the second was systematic sampling using random quadrant points and the third was haphazard sampling. Twenty-four samples were collected in each sampling.

Systematic sampling along the transect was the most efficient method and used the last amount of time. Densities were commonly higher than the actual densities; with transect sampling being the most accurate.

Transect sampling had the lowest percentage error for common and uncommon tree types. Haphazard sampling had the greatest error for uncommon tree types and random sampling had the greatest error for common tree types.

Twenty-four samples appears to be enough for transect sampling but not for the other two sampling methods.