18.2.3 Sequential Steps of the Exercise Implementation
1. Having chosen several starting CPA concentration vectors for different values of
the mean-square error, carry out the retrieval of the OAC concentration vector
making use of the code “lm”; choose the most accurate retrieval results. As it
ensues from above, the mean-square error according to the Eq. 18.4 is a criterion
for exiting the iterative procedure of the concentration vector calculation with
the Eq. 18.3. The closer its value to zero, the more accurate is the concentration
vector retrieval. However, for different starting concentration vectors (which are
necessary for initiating the iterative procedure) the mean-square error value
close to zero can either be unattainable or hardly achievable because of unreasonably long computational time required for that. It is difficult, if ever possible,
to specify/set up in advance the actual closeness of the mean-square error to zero
for any initial (starting) concentration vector. That’s why the students are
recommended to first set up a tentative mean-square error value for each initial
concentration vector and then consecutively decrease it after termination of each
run of the code “IM.pas.” Each step of this trail procedure should then be
followed by assessment of the accurateness of the concentration vector retrieval.
2. Analyze the retrieval accuracy variations for each of the CPA depending on the
concentrations of the rest CPAs given that the mean root-square error is fixed.
Table 18.1 Options of CPA concentration combinations
Option
C chl , mg/l
C sm , mg/l
C doc , mgC/l
1
0.5; 1; 5; 18
0.5
1
2
0.5; 1; 5; 18
0.5
5
3
0.5; 1; 5; 18
0.5
18
4
0.5
0.5; 1; 5; 18
1
5
0.5
0.5; 1; 5; 18
5
6
0.5
0.5; 1; 5; 18
18
7
1
1
0.5; 1; 5; 18
8
5
1
0.5; 1; 5; 18
9
1
5
0.5; 1; 5; 18
18
5
5
0.5; 1; 5; 18
Table 18.2 Input data and results of retrieval from R(À0,l) of the concentration of chlorophyll,
suspended matter and dissolved organics
Input concentration
vector
Initial concentration
vector
Retrieval results
Mean rootsquare error
C chl ,
mg/l
C sm ,
mg/l
C doc ,
mgC/l
C chl ,
mg/l
C sm ,
mg/l
C doc ,
mgC/l
C chl ,
mg/l
C sm ,
mg/l
C doc ,
mgC/l
3.0
3.0
3.0
1
1
1
3.37
3.34
3.78
0.0273
2
2
2
3.29
3.44
3.61
0.0881
18.2 Practice 17
179
1. Having chosen several starting CPA concentration vectors for different values of
the mean-square error, carry out the retrieval of the OAC concentration vector
making use of the code “lm”; choose the most accurate retrieval results. As it
ensues from above, the mean-square error according to the Eq. 18.4 is a criterion
for exiting the iterative procedure of the concentration vector calculation with
the Eq. 18.3. The closer its value to zero, the more accurate is the concentration
vector retrieval. However, for different starting concentration vectors (which are
necessary for initiating the iterative procedure) the mean-square error value
close to zero can either be unattainable or hardly achievable because of unreasonably long computational time required for that. It is difficult, if ever possible,
to specify/set up in advance the actual closeness of the mean-square error to zero
for any initial (starting) concentration vector. That’s why the students are
recommended to first set up a tentative mean-square error value for each initial
concentration vector and then consecutively decrease it after termination of each
run of the code “IM.pas.” Each step of this trail procedure should then be
followed by assessment of the accurateness of the concentration vector retrieval.
2. Analyze the retrieval accuracy variations for each of the CPA depending on the
concentrations of the rest CPAs given that the mean root-square error is fixed.
Table 18.1 Options of CPA concentration combinations
Option
C chl , mg/l
C sm , mg/l
C doc , mgC/l
1
0.5; 1; 5; 18
0.5
1
2
0.5; 1; 5; 18
0.5
5
3
0.5; 1; 5; 18
0.5
18
4
0.5
0.5; 1; 5; 18
1
5
0.5
0.5; 1; 5; 18
5
6
0.5
0.5; 1; 5; 18
18
7
1
1
0.5; 1; 5; 18
8
5
1
0.5; 1; 5; 18
9
1
5
0.5; 1; 5; 18
18
5
5
0.5; 1; 5; 18
Table 18.2 Input data and results of retrieval from R(À0,l) of the concentration of chlorophyll,
suspended matter and dissolved organics
Input concentration
vector
Initial concentration
vector
Retrieval results
Mean rootsquare error
C chl ,
mg/l
C sm ,
mg/l
C doc ,
mgC/l
C chl ,
mg/l
C sm ,
mg/l
C doc ,
mgC/l
C chl ,
mg/l
C sm ,
mg/l
C doc ,
mgC/l
3.0
3.0
3.0
1
1
1
3.37
3.34
3.78
0.0273
2
2
2
3.29
3.44
3.61
0.0881
18.2 Practice 17
179