Ordination and Classification

The SOM can be applied to community data collected from field observations (Figure 2a). The data used for training were from benthic macroinvertebrate communities collected across different levels of pollution ranging from oligo- to poly-saprobity in urban streams (Figure 2b). The sample sites showed different levels of organic pollution (Figure 2b). The trained SOMs produced groupings of computation nodes on a hexagonal map (Figure 3). By using the hexagonal map, distances between the target node and its surrounding nodes would be the same. The acronyms used in each unit of the SOM map stand for the samples. The first two letters relate to study sites (the Suyong (SY), Cheolma (CM), Hoedong (HD), and Soktae (ST) streams in the Suyong River in Korea), while the last three characters represent the sampling seasons in October 1989 and in January, May, and August 1990: SPR, spring; SUM, summer; AUT, autumn; and WIN, winter (e.g., ST1SPR; samples at ST1 in spring).

Grouping of patterned sample sites was arranged according to different pollution levels, with the level of pollution being in accordance with sample sites. Consequently, the patterned nodes were grouped according to sample sites, ST, HD, CM, and SY. Samples collected from CM, for instance, are mostly located in the lower area of the SOM, while those belonging to HD are concentrated in the upper right area of the SOM. The arrangement of the groups on the map also revealed the pollution gradient along the vertical axis: the upper area of the trained map correlates to the polluted sampling sites, whereas the lower area was associated with the relatively clean sites. Seasonal variations were also observed locally on the map (Figure 3). For instance, samples collected in summer at SY1—5 were either grouped together in the same unit or were located near each other (e.g., nodes (5 (row),3 (column)) and (6,2)).

In order to show the degree of association among the patterned nodes, the U-matrix algorithm can be applied to the trained SOM (Figure 4). The U-matrix calculates the distance of a weight vector to its neighbors in the SOM, and displays the cluster structure of the map units. Suppose the map has a size of m columns and n rows, the

Figure 2 Location of sampling sites (a) and their pollution states (b). Reproduced from Park Y-S, Chon T-S, Kwak I-S, and Lek S (2004) Hierarchical community classification and assessment of aquatic ecosystems using artificial neural networks. Science of the Total Environment 327: 105-122, with permission.

Suyong stream

SY1 SY2 SY3 SY4 SY5

Hoedong stream

Cheolma stream

CM1 CM2 CM3 CM4 CM5

CM1 CM2 CM3 CM4 CM5

ST1 ST2 ST3 ST4

Hoedong stream

HD1

1

(4.4)

HD2

1

(3.9)

HD3

(8.1)

1

HD4

1

(12.7)

HD5

Soktae stream

Suyong river

I I Oligosaprobity

0 ^-Mesosaprobity |~~1 a-Mesosaprobity

1 I Polysaprobity ( ) BOD

Figure 2 Location of sampling sites (a) and their pollution states (b). Reproduced from Park Y-S, Chon T-S, Kwak I-S, and Lek S (2004) Hierarchical community classification and assessment of aquatic ecosystems using artificial neural networks. Science of the Total Environment 327: 105-122, with permission.

ST1 ST2 ST3 ST4

following value (Mxy; U-matrix) for the node at (x, y) is calculated for all positions:

Was this article helpful?

0 0
Oplan Termites

Oplan Termites

You Might Start Missing Your Termites After Kickin'em Out. After All, They Have Been Your Roommates For Quite A While. Enraged With How The Termites Have Eaten Up Your Antique Furniture? Can't Wait To Have Them Exterminated Completely From The Face Of The Earth? Fret Not. We Will Tell You How To Get Rid Of Them From Your House At Least. If Not From The Face The Earth.

Get My Free Ebook


Post a comment