Time Series Analysis of Water Quality of Ramgarh Lake of...
Transcript of Time Series Analysis of Water Quality of Ramgarh Lake of...
Time Series Analysis of Water Quality of Ramgarh Lake of Rajasthan
Published in:Rajesh Singh, F. Smarandache (Editors) STUDIES IN SAMPLING TECHNIQUES AND TIME SERIES ANALYSIS Zip Publishing, Columbus, USA, 2011ISBN 978-1-59973-159-9pp. 5 - 23
Rajesh Singh, Shweta Maurya Department of Statistics, Banaras Hindu University
Varanasi-221005, INDIA
Ashish K. Singh Raj Kumar Goel Institute of Technology, Ghaziabad, India.
Florentin Smarandache Department of Mathematics, University of New Mexico, Gallup, USA
5
Abstract
In this chapter an attempt has been made to study the effect of certain pollutants
on water of Ramgarh Lake of Rajasthan. Time series analysis of the observed data has
been done using trend, single exponential smoothing and double exponential smoothing
methods. Keywords: Pollutants, trend, single, double exponential smoothing, time series.
1. IntroductionSeventy percent of the earth’s surface is covered by water. Water is undoubtedly
the most precious natural resource that exists on our planet. Water is an important
component of the eco-system and any imbalance created either in terms of amount which
it is represent or impurities added to it, Can harm the whole eco-system. Water pollution
occurs when a body of water is adversely affected due to the addition of large amount of
pollutant materials to the water. When it is unfit for its intended use, water is considered
polluted. There are various sources of water pollution (for detail refer to Jain (2011))
Some of the important water quality factors are:
1) Dissolved oxygen (D.O.)
2) Biological oxygen demand (B.O.D.)
3) Nitrate
6
4) Coliform
5) P.H.
Chemical analysis of any sample of water gives us a complete picture of its
physical and chemical constituents. This will give us only certain numerical value but
for estimating exact quality of water a time series system has been developed known
as water quality trend, which gives us the idea of whole system for a long time (see
Jain(2011)).
In this chapter we are calculating the trend values for five water parameters of
Ramgarh lake in Rajasthan for the year 1995-2006 and for three parameters of Mahi river
for the year 1997-2008.these methods viz. trend analysis, single smoothing are used to
analyze the data.
2. Methodology After ensuring the presence of trend in the data, smoothing of the data is the next
requirement for time series analysis. For smoothing the common techniques discussed by
Gardner(1985) are trend, simple exponential smoothing (SES), double exponential
smoothing (DES), triple exponential smoothing (TES) and adaptive response rate simple
exponential smoothing (ARRSES). Jain (2011) used trend method to analyze the data.
We have extended the work of Jain (2011) and analyzed the data using SES and DES and
compared these with the help of the available information. The methods are described
below:
1. Fitting Of Straight line
The equation of the straight line is-
Ut=a+bt
where,
ut=observed value of the data, a=intercept value, b=slope of the straight line and
t=time (in years)
Calculation for a and b:
7
The normal equations for calculating a and b are
∑Ut=n a+b∑t
∑t Ut=at+b∑t**2
2. Single Exponential Smoothing
The basic equation of exponential smoothing is
St = α yt-1+ (1-α) St-1 , 0 ≤ α ≤1
and parameter α is called the smoothing constant.
Here, Si stands for smoothed observation or EWNA and y stands for the original
observation. The subscripts refer to the time periods 1,2,3…..,n.
The smoothed series starts with the smoothed version of the second observation.
3. Double Exponential Smoothing
Single smoothing does not excel in following the data when there is a trend. This
situation can be improved by the introduction of a second equation with a second
constant γ, which must be chosen in conjunction with α.
St=α yt+ (1-α) (St-1+bt-1) , 0≤α≤1
bt= γ (St-St-1) + (1- γ ) bt-1, 0≤ γ ≤1.
For forecasting using single and double exponential smoothing following method is used-
Forecasting with single exponential smoothing
St=αyt-1+(1-α)St-1 0≤ α ≤1
The new forecast is the old one plus an adjustment for the error that occurred in the last
forecast.
8
Boot strapping of forecasts
St+1=α yorigin +(1-α) St
This formula works when last data points and no actual observation are available.
Forecasting with double exponential smoothing
The one period-ahead forecast is given by:
Ft+1=St+bt
The m-periods-ahead forecast is given by:
Ft+m=St+mbt
( for detail of these methods refer to Gardner (1985)).
4. Results and Discussion
Table 1 shows the data on Dissolved Oxygen (D.O.) for Ramgardh Lake for the
years 1995-2006 and fitted values using trend, single and double exponential smoothing.
Table 1: Data on Dissolved Oxygen (D.O.) for Ramgardh Lake for the years 1995-2006 and fitted values using trend, single and double exponential smoothing
Observed Data
Trend values
Single exponential smoothing(alpha=0.1)
Double exponential smoothing(alpha=0.9
and gamma=0.1) 1995 5.12 5.3372 5.12 1996 5.75 5.3036 5.12 5.707 1997 5.26 5.27 5.183 5.32857 1998 5.72 5.2364 5.1907 5.698556 1999 4.64 5.2028 5.24363 4.765484 2000 5.14 5.1692 5.183267 5.110884 2003 4.23 5.1356 5.17894 4.329044 2004 6.026 5.102 5.084046 5.858346 2005 4.46 5.0684 5.178242 4.616965 2006 5.52 5.0348 5.106417 5.4327 Total 51.866 51.86 46.46824 51.96755
Ut = 5
with
of α a
the d
comb
tried
Figur
neces
adequ
levels
www
above
Tim
ese
riesv
alue
s
For trend,
5.1866-0.01
mean square
are tried and
ata, Holt’s d
binations of
and MSE =
re 1- Gr
expon
the ye
Adequate
ssary elemen
uate oxygen
s in water d
w.state.ky.us)
Form Tab
e the require
0
1
2
3
4
5
6
7
1995 1
Tim
e se
ries v
alue
s
, the fitted li
69 * t
ed error (MS
d minimum
double expon
α and γ bot
0.0094 was
raph of obse
nential smoo
ears 1995-20
e dissolved
nt to all fo
n levels in or
drop below
).
ble 1 and Fi
ed standard 5
1996 1997 199
ne is
SE) 0.3077.
MSE = 0.39
nential smoo
th ranging b
least for α =
erved data an
othing of Di
006
oxygen is n
orms of life
rder to prov
5.0 mg/l, a
igure 1, we o
5.0 mg/l, exc
98 1999 2000
Year
Disso
9
For single e
915 was obt
othing was fo
etween 0.1
= 0.9 and γ =
nd fitted val
issolved Ox
necessary fo
. Natural st
vide for aero
aquatic life
observe that
cept for the t
0 2003 2004 2
olved oxy
exponential s
tained for α
found to be m
and 0.9 with
= 0.1.
lues using t
xygen (D.O.)
or good wat
tream purifi
obic life form
is put unde
t level of D.
two years 19
2005 2006
ygen
smoothing v
= 0.1. For
most appropr
h increment
trend, single
) for Ramga
ter quality.
ication proc
ms. As disso
r stress ( fo
O. in Ramga
999 and 2005
Observed
Trend va
single exsmoothin
Double esmoothingamma=
various value
smoothing o
riate. Variou
s of 0.1 wer
e and doubl
ardh Lake fo
Oxygen is
esses requir
olved oxyge
or details se
arh Lake wa
5.
d Data
alues
xponential ng(alpha=0.1)
exponential ng(alpha=0.9 &
=0.1)
es
of
us
re
le
or
a
re
en
ee
as
&
Ta
Figur
Tim
ese
riesv
alue
s
able 2: Com
Per
23456789
11111111112
re 2: Graph o
0
2
4
6
8
1 2
Tim
e se
ries v
alue
s
mparison of f
riod Obs
D1 52 53 54 55 46 57 48 6.9 4
10 511 12 13 14 15 16 17 18 19 20
of forecasts
3 4 5
Co
Observed D
forecasts
served Data Fo5.12 5.75 5.26 5.72 4.64 5.14 4.23 .026
4.46 5.52
6 7 8
ompariso
Data Fo
10
orecast(sing
5.7437 5.25923
5.7147074.6460363
5.140432674.239489406.016580464.467182415.515864175.147775735.184998155.218498345.248648505.275783655.300205295.322184765.341966285.359769655.37579269
9 10 11
Period
on of fo
orecast(single)
gle) Foreca
65.7
6.513 5.017 5.81
03 4.363 7.51
6 4.6275 6.6831 558 542 508 557 91 662 686 657 692
12 13 14 1
recasts
) Fore
ast(double) 5.32
.23084 7869651
10832048 13406419 15207177 2655631
11558059 21632535 86376834 5.6266 5.7442 5.8618 5.9794 6.097
6.2146 6.3322 6.4498 6.5674 6.685
15 16 17 18
cast(double)
8 19 20
11
Table 3 shows the data on Nitrate for Ramgardh Lake for the years 1995-2006
and fitted values using trend, single and double exponential smoothing.
Table 3: Data on Nitrate for Ramgardh Lake for the years 1995-2006 and fitted values using trend, single and double exponential smoothing
Year(x) Observed Data
Trend values
Single exponential smoothing(alpha=0.1)
Double exponential smoothing(alpha=0.9 and gamma=0.1)
1995 0.32 0.588 0.32 1996 1.28 0.546 0.32 1.176 1997 0.38 0.504 0.416 0.46096 1998 0.08 0.462 0.4124 0.104883 1999 0.04 0.42 0.37916 0.028797 2000 0.92 0.378 0.345244 0.815205 2003 0.24 0.336 0.40272 0.300708 2004 0.25 0.294 0.386448 0.247331 2005 0.186 0.252 0.372803 0.184874 2006 0.3 0.21 0.354123 0.281431 Total 3.996 3.99 3.388897 3.920189
For trend, the fitted line is
Ut = 0.3996 - 0.0216 * t
with MSE= 0.3078. For single exponential smoothing various values of α are tried and
minimum MSE = 0.3412 was obtained for α = 0.1. For smoothing of the data, Holt’s
double exponential smoothing was found to be most appropriate. Various combinations
of α and γ both ranging between 0.1 and 0.9 with increments of 0.1 were tried and MSE =
0.0033 was least for α = 0.9 and γ = 0.1.
Nitrites can produce a serious condition in fish called "brown blood disease."
Nitrites also react directly with hemoglobin in human blood and other warm-blooded
animals to produce methemoglobin. Methemoglobin destroys the ability of red blood
cells to transport oxygen. This condition is especially serious in babies under three
months of age. It causes a condition known as methemoglobinemia or "blue baby"
disease. Water with nitrite levels exceeding 1.0 mg/l should not be used for feeding
babie
have
was b
Figur
Nitrit
anim
cells
mont
disea
babie
have
was b
time
serie
sval
ues
es. Nitrite/ni
no effect on
Form Ta
below the sta
re 3: Graphexpon
Nitrites c
tes also reac
als to produ
to transpor
ths of age.
ase. Water w
es. Nitrite/ni
no effect on
Form Ta
below the sta
00.20.40.60.8
11.21.4
1995
time
serie
s val
ues
itrogen level
n warm wate
able 3 and F
andard 1.0 m
h of observnential smoo
can produce
ct directly w
uce methem
rt oxygen. T
It causes a
with nitrite
itrogen level
n warm wate
able 3 and F
andard 1.0 m
1996 1997 19
ls below 90
r fish (for de
Figure 3, we
mg/l, except
ved data andothing of Ni
a serious c
with hemogl
moglobin. M
This conditi
a condition
levels excee
ls below 90
r fish ( for d
Figure 3, we
mg/l, except
998 1999 2000
year
12
mg/l and n
etails see ww
e observe th
for the year
d fitted valitrate for Ram
condition in
lobin in hum
ethemoglobi
ion is espec
known as m
eding 1.0 m
mg/l and n
details see ww
e observe tha
for the year
0 2003 2004 2
Nitrate
nitrate levels
ww.state.ky.u
hat level of N
1996.
ues using tmgardh lake
n fish called
man blood a
in destroys
cially seriou
methemoglo
mg/l should
nitrate levels
ww.state.ky
at level of N
1996.
2005 2006
below 0.5
us).
Nitrate in R
trend, singlee for the year
d "brown blo
and other w
the ability
us in babies
obinemia or
not be used
below 0.5
.us).
Nitrate in R
Observe
Trend va
single exsmoothi
Double esmoothigamma=
mg/l seem t
Ramgarh Lak
e and doublrs 1995-2006
ood disease
warm-bloode
of red bloo
s under thre
"blue baby
d for feedin
mg/l seem t
Ramgarh Lak
ed Data
alues
xponential ing(alpha=0.1)
exponential ing(alpha=0.9 &=0.1)
to
ke
le 6
."
ed
od
ee
y"
ng
to
ke
&
T
Figur
Tim
ese
riesv
alue
s
Table 4: C
P
re 4 : Graph
-0.5
0
0.5
1
1.5
1Tim
e se
ries v
alue
s
Comparison o
eriod Obs
D1 02 13 04 05 06 07 08 09 0.
10 011 12 13 14 15 16 17 18 19 20
h of forecast
2 3 4
C
Observed
of forecasts
served Data For0.32 1.28 0.38 0.08 0.04 0.92 0.24 0.25 .186 0.3
ts
5 6 7
Comparis
Data
13
recast(singl0.416
0.4124 0.37916
0.345244 0.40272
0.386448 0.372803 0.354123 0.34871
0.348711 0.34384
0.339456 0.33551
0.331959 0.328763 0.325887 0.323298 0.320968 0.318872 0.316984
8 9 10 1perio
son of fo
Forecast(singl
e) Forecas0
1.10.32-0.0-0.10.840.220.1
0.110.240.20.20.10.10.0
0.00.0-0.0-0.0
-0
11 12 13 14d
orecasts
le) For
st(double) 0.24 1896 28832 07203 12795 47085 23314 7474
14309 44291
24426 20712
6998 3284
0957 05856 02142 01572 05286 0.09
4 15 16 17
s
recast(double)
7 18 19 20
)
14
Table 5 shows the data on B.O.D. for Ramgardh Lake for the years 1995-2006
and fitted values using trend, single and double exponential smoothing.
Table 5: Data on Biological oxygen demand (B.O.D.) for Ramgardh Lake for the years
1995-2006 and fitted values using trend, single and double exponential smoothing
Year(x) Observed
Data Trend values
single exponential smoothing(alpha=0.1)
Double exponential smoothing(alpha=0.9
& gamma=0.1) 1995 1.96 5.122 1.96 1996 14.09 4.762 1.96 12.87167 1997 1.84 4.402 3.173 3.047483 1998 1.8 4.042 3.0397 1.920392 1999 1.46 3.682 2.91573 1.490847 2000 2.78 3.322 2.770157 2.633116 2003 2.76 2.962 2.771141 2.742563 2004 1.976 2.602 2.770027 2.049477 2005 2.78 2.242 2.690624 2.697155 2006 3.58 1.882 2.699562 3.489379 Total 35.026 35.02 24.78994 34.90208
Biochemical oxygen demand is a measure of the quantity of oxygen used by
microorganisms (e.g., aerobic bacteria) in the oxidation of organic matter. Natural
sources of organic matter include plant decay and leaf fall. However, plant growth and
decay may be unnaturally accelerated when nutrients and sunlight are overly abundant
due to human influence. Urban runoff carries pet wastes from streets and sidewalks;
nutrients from lawn fertilizers; leaves, grass clippings, and chapter from residential areas,
which increase oxygen demand. Oxygen consumed in the decomposition process robs
other aquatic organisms of the oxygen they need to live. Organisms that are more tolerant
of lower dissolved oxygen levels may replace a diversity of natural water systems contain
bacteria, which need oxygen (aerobic) to survive. Most of them feed on dead algae and
other dead organisms and are part of the decomposition cycle. Algae and other producers
in the water take up inorganic nutrients and use them in the process of building up their
organic tissues (for details refer to www.freedrinkingwater.com).
For tr
with
minim
doub
of α a
0.300
Figur
Tim
ese
riesv
alue
srend, the fitt
Ut = 3.502
MSE= 11.74
mum MSE =
le exponenti
and γ both ra
00 was least
re 5: Grapexpon
02468
10121416
1995
Tim
e se
ries v
alue
sted line is
26+0.18094
4366. For si
= 17.1093 w
ial smoothin
anging betw
for α = 0.
ph of observnential smoo
1996 1997 19
Bio
* t
ngle expone
was obtained
ng was foun
een 0.1 and
.9 and γ = 0.
ved data anothing of B.O
998 1999 2000
Year
ological
15
ential smooth
d for α = 0.1
nd to be mos
0.9 with inc
1.
nd fitted valO.D. for Ram
2003 2004 20
oxygen
hing various
1. For smoo
st appropriat
crements of 0
lues using tmgardh Lake
005 2006
demand
s values of α
othing of the
te. Various
0.1 were trie
trend, singlee for the yea
d Obse
Trend
singlesmoo.1)Doubsmoo.9 & g
α are tried an
e data, Holt’
combination
ed and MSE
e and doublars 1995-200
erved Data
d values
e exponential othing(alpha=0
ble exponentiaothing(alpha=0gamma=0.1)
nd
’s
ns
=
le 06
0
l 0
Tabl
Figur
Tim
ese
ries
valu
es
le 6: Com
P
re 6 : Graph
0
5
10
15
1 2
Tim
e se
ries
valu
es
mparison of
eriod Obs
D1 12 143 14 5 16 27 28 1.9 2
10 311 12 13 14 15 16 17 18 19 20
h of forecast
2 3 4 5
Com
Observed Dat
forecasts
served Data For1.96 4.09
1.84 1.8
1.46 2.78 2.76 .976
2.78 3.58
ts
6 7 8
mpariso
ta For
16
recast(singl
3.173 3.0397
2.91573 2.770157 2.771141 2.770027 2.690624 2.699562 2.787606
2.7871 2.86639
2.937751 3.001976 3.059778
3.1118 3.15862
3.200758 3.238683 3.272814
9 10 11 12
Period
n of fore
ecast(single)
e) Forecas1.9013.93.001.761.312.582.711.952.673.54
3.53.63.63.73.73.83.
3.94.04.0
2 13 14 15
ecasts
Foreca
st(double) 06667 91483 03915 68471 11164 85629 10768 51553 73792 47575 5474 6055 6636 7217 7798 8379 .896 9541 0122 0703
16 17 18 1
ast(double)
19 20
17
Table 7 shows the data on Total Colifirm for Ramgardh Lake for the years
1995-2006 and fitted values using trend, single and double exponential smoothing.
Table 7 : Data on Total Colifirm for Ramgardh lake for the years 1995-2006 and fitted
values using trend, single and double exponential smoothing
Year(x) Observed
Data Trend values
single exponential smoothing(alpha=0.6)
Double exponential smoothing(alpha=0.9
& gamma=0.1) 1995 1169 687.894 1169 1996 285 615.314 1169 339.2333 1997 840.75 542.734 638.6 751.5507 1998 144 470.154 759.89 173.7353 1999 65.33 397.574 390.356 42.47463 2000 121.5 324.994 195.3404 81.95854 2003 119 252.414 151.0362 87.21566 2004 720.6 179.834 131.8145 632.042 2005 86 107.254 485.0858 123.3548 2006 61.66 34.674 245.6343 47.21817 Total 3612.84 3612.84 4166.757 3447.783
Total coliform bacteria are a collection of relatively harmless microorganisms that
live in large numbers in the intestines of man and warm- and cold-blooded animals. They
aid in the digestion of food. A specific subgroup of this collection is the fecal coliform
bacteria, the most common member being Escherichia coli. These organisms may be
separated from the total coliform group by their ability to grow at elevated temperatures
and are associated only with the fecal material of warm-blooded animals. The presence of
fecal coliform bacteria in aquatic environments indicates that the water has been
contaminated with the fecal material of man or other animals. The presence of fecal
contamination is an indicator that a potential health risk exists for individuals exposed to
this water (for details see www.state.ky.us).
with
minim
doub
of α
MSE
Figur
For trend,
Ut = 361.2
MSE= 9989
mum MSE =
le exponenti
and γ both
E = 2432.458
re 7: Grapexpon1995-
0
200
400
600
800
1000
1200
1400
19
Tim
e se
ries v
alue
s
, the fitted li
284 - 36.2989
96.33. For si
= 205949.6 w
ial smoothin
h ranging be
8 was least fo
ph of observnential smoo-2006.
995199619971
ne is
9* t
ngle expone
was obtaine
ng was foun
etween 0.1 a
or α = 0.9 an
ved data anothing of T
199819992000
Year
Tota
18
ential smooth
d for α = 0.6
nd to be mos
and 0.9 wit
nd γ = 0.1.
nd fitted valTotal Colifor
020032004200
al colifor
hing various
6. For smoo
st appropriat
th increment
lues using trm for Ramg
052006
rm
s values of α
othing of the
te. Various
ts of 0.1 w
trend, singlegardh Lake
Observed Da
Trend values
single exponsmoothing(a
Double exposmoothing(agamma=0.1)
α are tried an
e data, Holt’
combination
ere tried an
e and doublfor the year
ata
s
nential alpha=0.6)
onential alpha=0.9 & )
nd
’s
ns
nd
le rs
19
Table 8: Comparison of forecasts
Period Observed
Data Forecast(single) Forecast(double) 1 1169 827.3333 2 285 638.6 -51.2433 3 840.75 759.89 441.3534 4 144 390.356 -163.224 5 65.33 195.3404 -273.915 6 121.5 151.0362 -198.843 7 119 131.8145 -164.98 8 720.6 485.0858 459.5482 9 86 245.6343 -82.7583 10 61.66 135.2497 -145.897 11 135.248 -145.897 12 91.0952 -339.012 13 73.43408 -532.127 14 66.36963 -725.242 15 63.54385 -918.357 16 62.41354 -1111.47 17 61.96142 -1304.59 18 61.78057 -1497.7 19 61.70823 -1690.82 20 61.67929 -1883.93
Figure 8: Comparison of forecasts
Tablevalue Tabl
Y
conce
range
e 9 shows thes using tren
e 9: Data ontrend, s
Year(x) Ob
1995 1996 1997 1998 1999 2000 2003 2004 2005 2006 Total 7
pH is a
entration of
e of 6.0 to 9.
-3000
-2000
-1000
0
1000
2000
Tim
e se
ries v
alue
s
he data on pHd, single and
n pH for Ramsingle and do
bserved Data
Tv
7.64 7.48 7.87 8.05 8.44 8.3
7.25 8.28 7.87
8.006 79.186
measure of
the hydroge
.0 appears to
1 2 3 4
Co
Obser
H for Ramgad double exp
mgardh Lakouble expon
Trend values
sismo
6.29 6.65 7.01 7.37 7.73 8.09 8.45 8.81 9.17 9.53 79.1
f the acidic
en ion [H+]
o provide pro
5 6 7
omparis
rved Data
20
ardh Lake forponential sm
ke for the yeanential smoot
ingle exponoothing(alp
7.64 7.608
7.66047.73832
7.8786567.9629257.82034
7.9122727.90381770.12473
or basic (
activity in
otection for t
8 9 10 11
Period
son of fo
Forecast(s
r the years 1moothing.
ars 1995-200thing
ential ha=0.2)
Dsm
2 6 5
4 2 7 3
alkaline) na
a solution d
the life of fr
1 12 13 14
d
orecasts
single)
995-2006 an
06 and fitted
Double expomoothing(al
& gamma7.64
7.50967.84498.042748.414178.327647.371508.19197.912928.003579.259
ature of a s
determines th
reshwater fis
15 16 17 1
Forecast(dou
nd fitted
d values usin
onential lpha=0.9 a=0.1)
67 63 46 77 45 03 54 23 57 14
solution. Th
he pH. A pH
sh and bottom
18 19 20
uble)
ng
he
H
m
dwell
syner
produ
Runo
amm
effect
effect
die (f
with
minim
doub
of α a
0.002
Figur
Tim
ese
riesv
alue
s
ling inverte
rgistic effect
uce effects
off from agr
onia, mercu
ts, if any, of
t at a pH of
for details se
For trend,
Ut = 7.918
MSE = 0.99
mum MSE =
le exponenti
and γ both ra
2735was leas
re 9: Grapexpon
0
2
4
6
8
10
12
1995
Tim
e se
ries v
alue
s
ebrates. The
ts. Synergy
greater than
ricultural, do
ury or other
f these subst
4.8. Howeve
ee www.state
, the fitted li
86 + 0.1845 *
995. For sin
= 0.1831 wa
ial smoothin
anging betw
st for α =
ph of observnential smoo
1996 1997 19
e most sign
involves the
n their sum
omestic, and
r elements.
ances. For e
er, as little a
e.ky.us).
ne is
t
ngle exponen
as obtained
ng was foun
een 0.1 and
0.9 and γ =
ved data anothing of pH
998 1999 2000
Year
21
nificant env
e combinatio
m. This proc
d industrial
The pH of
example, 4 m
as 0.9 mg/l o
ntial smooth
for α = 0.2
nd to be mos
0.9 with inc
0.1.
nd fitted valH for Ramgar
2003 2004 20
PH
ironmental
on of two or
cess is impo
areas may
the water w
mg/l of iron w
f iron at a pH
ing various
. For smoo
st appropriat
crements of 0
lues using trdh Lake for
005 2006
impact of
r more subs
ortant in su
contain iron
will determ
would not pr
H of 5.5 can
values of α
othing of the
te. Various
0.1 were trie
trend, singler the years 19
Obse
Trend
singlesmoo.2)Doubsmoo.9 & g
pH involve
stances whic
urface water
n, aluminum
ine the toxi
resent a toxi
n cause fish t
are tried an
e data, Holt’
combination
ed and MSE
e and doubl995-2006.
erved Data
d values
e exponential othing(alpha=0
ble exponentiaothing(alpha=0gamma=0.1)
es
ch
s.
m,
ic
ic
to
nd
’s
ns
=
le
0
l 0
Tabl
Figur
Tim
ese
riesv
alue
s
le 10: Comp
P
re 10: Graph
0
2
4
6
8
10
1 2
Tim
e se
ries v
alue
s
parison of for
eriod Obs
D1 72 73 74 85 86 87 78 89 7
10 8.11 12 13 14 15 16 17 18 19 20
h of forecast
2 3 4 5
Co
Observed Da
recasts
served Data For7.64 7.48 7.87 8.05 8.44 8.3
7.25 8.28 7.87 .006
ts
6 7 8
mpariso
ata For
22
recast(singl
6.732 7.083 7.245 7.596 7.47
6.525 7.452 7.083
7.2054 7.8368
7.85372 7.868948 7.882653 7.894988 7.906089 7.91608
7.925072 7.933165 7.940448
9 10 11 1
Period
on of for
recast(single)
e) Forecas7.777.617.978.188.578.467.398.297.988.078.078.1
8.218.2
8.358.4
8.498.5
8.648.
12 13 14 15
recasts
Foreca
st(double) 76667 19633 77463 81774 76446 65033 99539 99231 81569 74402 74395 4524
16085 28693 57775
42862 99465
57031 41155 .712
5 16 17 18
ast(double)
19 20
23
We observe from the calculations for the different parameters of pollutants that
double smoothing follows the data much closer than single smoothing. Furthermore, for
forecasting single smoothing cannot do better than projecting a straight horizontal line,
which is not very likely to occur in reality. So for forecasting purposes for our data
double exponential smoothing is more preferable.
Conclusion
From the above discussions, we observe that the various pollutants considered in
the article may have very hazardous effect on quality of water. Increase of pollutants in
water beyond a certain limit may be dangerous for aquatic animals. Also, according to
recent reports, most of the tap and well water in the India are not safe for drinking due to
presence of various pollutants in inappropriate percentage. Now, we have reached the
point where all sources of our drinking water, including municipal water systems, wells,
lakes, rivers, and even glaciers, contain some level of contamination. So, we need to
keep a routine check of the quality of water so that we can lead a healthy life.
References
Gardner, E. S. (1985) : Exponential smoothing- The state of the art. Jour. Of
Forecasting,4, 1-28.
Jain, Smita (2011): A Statistical study of effect of different pollutants on water (with
specific reference to Rajasthan). Unpublished thesis submitted to University of
Rajasthan, India.
www.state.ky.us
www.freedrinkingwater.com