Page 165 - ASBIRES-2017_Preceedings

STUDY ON OPTIMIZATION OF TROPICAL FISH HARVESTING OF A SRI LANKAN FISH FARM
3 METHODOLOGY
In order to develop a model, the chosen
fish varieties were analyzed through the key
factors that assumed to be affecting on the
fish population. Mainly the data related to
these key factors have been analyzed for
five consecutive years.
In this study the population dynamics of
each fish farm was considered with key
factors which include deaths on house, local
purchasing quantity, breeding, import,
deaths on arrival and shipment.
It is noteworthy to mention that there is
no import process in fresh water farm and
no breeding process in marine water farm.
Beside those, deaths on arrival from the
import process are out of the scope of this
research. It is obvious that since the study is
carried out for a farm and not for a natural
resource.
Therefore, the conditions such as
predators etc were disregarded. Deaths on
the house (DOH) had been examined
through four key factors, temperature,
salinity, pH level and alkalinity in order to
reduce the deaths on house. Alkalinity was
measured by the amount of acid needed to
bring the water sample to pH level 4.2. The
result was recorded as milligrams per liter
(mg/l) of calcium carbonate.
The mathematical software was used for
the analysis. Initially, import the data set to
the workspace. To avoid the overload only
the monthly fish population in the farm
during 5 years was selected for further
analysis. Then the graph of monthly fish
population with respect to days were plotted
to examine a pattern.
Curve fitting toolbox was used to fit the
polynomial equation. Auto fit option; center
and scale were selected to avoid the errors
of the model. In addition, option Robust
Bisquare, which gives the minimum weight
to the points away from the prediction
bound or the points that considered as
outliers were selected. Best fitted model was
obtained by giving different degrees to the
155
polynomial until it gives considerable R
square value.
Then correlation between the dependent
variable (DOH) and independent variables
(temperature, salinity, pH level and
alkalinity) were checked. Finally, the model
was constructed by selecting the significant
factors that affect for deaths on house.
4 DATA COLLECTION AND
ANALYSIS
As the initial step of this study, secondary
data were collected from each type of farm.
From collected data fish variety is randomly
selected from each farm because the analysis
has to be made for a selected fish variety to
avoid the effect of different mortalities.
Then data from five consecutive years had
been investigated to check whether there
was a pattern in the monthly population of
the farm.
5 RESULTS AND DISCUSSION
The monthly fish population was found
from the population of the last day of the
month as it was the population of the farm.
Figure 1: Monthly Wreck fish population
for the year 2016
Figure 1 illustrates the monthly Wreck
fish population in the farm with respect to
time. There is a high population during the
month September and low population in
April and July. It is necessary to notice that
the reasons for low population might be the
shipments beside the high DOH. The fitted
polynomial model is shown given by