You are on page 1of 11

Name : Czarina Mae B.

Macaraeg Score:
Course/Section : BSABE 3-1
Date : December 09, 2021
Schedule (Day/Time) : Th 10:00-1:00

---------------------------------------------------------------------------------------------------------------------

Laboratory Exercise
Test of Homogeneity/Double Mass Curve

INTRODUCTION

Double mass curve analysis is one of the methods used to analyze the homogeneity of the

rainfall data. It requires two sets of data from different station, whereas the other data is set

as homogeneous set, while the other one is the area of interest, whether consistent or not.

H. Clayton and J. Searcy describe the cumulative rainfall data of the two stations will be plot

as scatter plot, inserting trend line to determine the regression line and correlation equation.

The breakpoint of the data is the area of interest of the researcher. These breaks may relate to

various consequences such as faulty instruments and unreliable observer. They also added

that poor correlation between the variables can prevent detection of inconsistencies in a

record, but prolong the offset relation of the data that effect the estimations. The residual

mass, curve which is a modification of the double mass curve, magnifies imperceptible

breaks in the double mass curve for detailed study.

MATERIALS AND METHOD

1. Table below shows the rainfall data for the month of January average from the 6 nearby

weather station from station X from year 1997-2018.


2. Station X have missing rainfall data, thus needed to estimate. However, a test of

consistency or homogenous test is needed before the estimation of the missing data.

3. Perform a homogeneity test and if found inconsistent, perform the double mass curve

technique in correcting the data. State all your assumptions.

Average January Precipitation January


Year from 6 nearby Station with the Precipitation at
same climatic condition, mm Station X, mm
1997 34.36 35.01
1998 42.50 41.50
1999 25.87 30.30
2000 40.60 39.04
2001 40.61 37.80
2002 40.06 43.50
2003 33.02 30.20
2004 34.19 36.40
2005 36.75 40.64
2006 37.58 40.40
2007 36.87 27.25
2008 29.91 21.50
2009 39.23 31.04
2010 30.01 -
2011 28.02 20.40
2012 31.97 23.31
2013 31.51 23.01
2014 33.38 24.20
2015 31.20 -
2016 32.00 25.10
2017 30.21 22.50
2018 27.39 20.40

RESULTS AND DISCUSSION/CONCLUSION

The missing data was estimated using the normal ratio method, and the calculation was made in
the previous lab exercises. Therefore, the value of the missing data in the previous lab exercise
was also the same value of this exercise.
REGRESSION LINE OF THE DATA
50

45

40

35 y = 1.2893x - 13.436
R² = 0.6323
30

25

20

15

10

0
0 5 10 15 20 25 30 35 40 45

Figure 1. The regression line of the data

Year Average Station X, mm


January
Precipitation
from 6 nearby y = 1.2893x - 13.436 𝑬𝒊
Station with
the same
climatic
condition, mm
1997 34.36 35.01 30.864348 -4.145652
1998 42.5 41.5 41.35925 -0.14075
1999 25.87 30.3 19.918191 -10.381809
2000 40.6 39.04 38.90958 -0.13042
2001 40.61 37.8 38.922473 1.122473
2002 40.06 43.5 38.213358 -5.286642
2003 33.02 30.2 29.136686 -1.063314
2004 34.19 36.4 30.645167 -5.754833
2005 36.75 40.64 33.945775 -6.694225
2006 37.58 40.4 35.015894 -5.384106
2007 36.87 27.25 34.100491 6.850491
2008 29.91 21.5 25.126963 3.626963
2009 39.23 31.04 37.143239 6.103239
2010 30.01 26.62 25.255893 -1.364107
2011 28.02 20.4 22.690186 2.290186
2012 31.97 23.31 27.782921 4.472921
2013 31.51 23.01 27.189843 4.179843
2014 33.38 24.2 29.600834 5.400834
2015 31.2 27.67 26.79016 -0.88984
2016 32 25.1 27.8216 2.7216
2017 30.21 22.5 25.513753 3.013753
2018 27.39 20.4 21.877927 1.477927
Table 1. Calculation of the cumulative residual to test the homogeneity

RESIDUAL
60

50

40

30

20

10

0
0 5 10 15 20 25
-10

Figure 2. Scatter plot of the calculated residual in Table 1

As we can see from above, the breakpoint of the data where they divide into groups. It started at

point 10, year 2007. I have decided that the succeeding years, 2008-2018, will be the one that

needed correction since there are two missing values fall on those years2010 and 2015. It also

prove that the precipitation data of station x was not homogenous with the nearby stations.

Since we test the homogeneity of the station X and found out, that it is not consistent with the

nearby station. To correct the data, we used double mass curve technique.
Year Average Station X, mm Cumulative sum Cumulative
January for nearby sum for
Precipitation stations station X
from 6 nearby
Station with
the same
climatic
condition, mm
1997 34.36 35.01 34.36 35.01
1998 42.5 41.5 76.86 76.51
1999 25.87 30.3 102.73 106.81
2000 40.6 39.04 143.33 145.85
2001 40.61 37.8 183.94 183.65
2002 40.06 43.5 224 227.15
2003 33.02 30.2 257.02 257.35
2004 34.19 36.4 291.21 293.75
2005 36.75 40.64 327.96 334.39
2006 37.58 40.4 365.54 374.79
2007 36.87 27.25 402.41 402.04
2008 29.91 21.5 432.32 423.54
2009 39.23 31.04 471.55 454.58
2010 30.01 26.62 501.56 481.2
2011 28.02 20.4 529.58 501.6
2012 31.97 23.31 561.55 524.91
2013 31.51 23.01 593.06 547.92
2014 33.38 24.2 626.44 572.12
2015 31.2 27.67 657.64 599.79
2016 32 25.1 689.64 624.89
2017 30.21 22.5 719.85 647.39
2018 27.39 20.4 747.24 667.79
Table 2. The cumulative data of the homogeneous station and station x

The missing data was fulfilled in the previous exercises, and the data in 2010 and 2015 were
26.62 and 27.67, respectively. On the other hand, the cumulative at both stations were calculated
as follows;
Cumulative sum for 6 nearby stations,
𝑖=1
𝑁𝑖 = 𝑁𝑖 + ∑ 𝑁𝑗
𝑗=1
𝑖=1
𝑋𝑖 = 𝑋𝑖 + ∑ 𝑋𝑗
𝑗=1

SCATTER PLOT OF THE DATA


800

700
CUMULATIVE AT STATION X

600 y = 0.9346x
R² = 0.9911
500

400

300

200

100

0
0 100 200 300 400 500 600 700 800
CUMULATIVE AT NEARBY STATIONS

Figure 3. Scatter Plot of the cumulative at both stations

From 6 nearby station, the mean is 35.58 while for station X, 31.5. In MS excel we plot the

rainfall data as scatter plot and add the trendline to determine the regression equation, coefficient

relation, and breakpoint of the data. Where the non- homogeneity of the data starts.

COMPUTATION USING REGRESSION LINES

𝑦 ′ = 𝑏𝑥𝑖

∑(𝑥𝑖 − 𝑥̅ )((𝑦𝑖 − 𝑦̅)


𝑏=
∑(𝑥𝑖 − 𝑥̅ )2

𝜺𝒊
4.541962903
3.813982055
7.360299194
3.03876874
1.789901441
7.97760285
0.920180881
6.082706976
8.052678601
7.076692838
-5.44372898
-5.022089335
-3.746411388
-1.010762318
-4.446169949
-5.038752793
-4.930857069
-5.399041859
-4.165970821
-3.275354688
-4.288108285
-3.887530153

Table 3. The residual data of using Double mass curve technique

Since we know the regression equation and the residual is calculated and tabulated above. The

residual data revealed the inconsistency of the data where the breakpoint is, the non

homogeneous data represented, which is visually assessed by plotting it.


Chart Title
800

700 y = 0.7723x + 91.13


R² = 0.9997

600

500
Axis Title

400 y = 1.0151x
R² = 0.9996
300

200

100

0
0 100 200 300 400 500 600 700 800
Axis Title
Figure 4. The scatter plot of the homogeneous and non-homogeneous set of precipitation data

The regression line of the homogeneous subset was y = 1.0151x. On the other hand, for non

homogeneous was y = 0.7723x + 91.13. The difference of the both set was used to recalculate

the data in order to match the set in the nearby stations.

Cumulative Rainfall at 6 Cumulative Rainfall at Station 𝒚𝒄,𝒊 = 𝒚𝒊 + ∆𝒚′


Station X

34.36 35.01 35.01


76.86 76.51 76.51
102.73 106.81 106.81
143.33 145.85 145.85
183.94 183.65 183.65
224 227.15 227.15
257.02 257.35 257.35
291.21 293.75 293.75
327.96 334.39 334.39
365.54 374.79 374.79
402.41 402.04 408.615148
432.32 423.54 437.377296
471.55 454.58 477.94234
501.56 480.18 511.848768
529.58 500.58 539.052024
561.55 523.89 570.12434
593.06 546.9 600.784968
626.44 571.1 633.089632
657.64 594.6 668.344992
619.7 689.64 701.214592
642.2 719.85 731.04958
662.6 747.24 758.099872
Table 4. The corrected version of the data

CORRECTED DATA
800

700
y = 1.0151x
R² = 0.9999
600

500

400

300

200

100

0
0 100 200 300 400 500 600 700 800

Figure 5. The homogeneous plot of the data in both stations


Year Average January Precipitation
from 6 nearby Station with the
same climatic condition, mm Corrected Data at Station X, mm

1997 34.36 35.01


1998 42.5 41.5
1999 25.87 30.3
2000 40.6 39.04
2001 40.61 37.8
2002 40.06 43.5
2003 33.02 30.2
2004 34.19 36.4
2005 36.75 40.64
2006 37.58 40.4
2007 36.87 33.825148
2008 29.91 28.762148
2009 39.23 40.565044
2010 30.01 33.906428
2011 28.02 27.203256
2012 31.97 31.072316
2013 31.51 30.660628
2014 33.38 32.304664
2015 31.2 35.25536
2016 32 32.8696
2017 30.21 29.834988
2018 27.39 27.050292
Table 5. The corrected data at station X

Through series of calculations, the data presented at station x was found to be inconsistent with

the nearby stations. So that, double mass curve analysis was used to estimate the data in order to

correct the inconsistent part of the precipitation. The corrected data at station x were presented

above. We can see the correlation relation is equal to 0.9999, approximately 1, which describe as

having an precise relationship with the nearby station. It is now consistent with the time series of

precipitation data from nearby 6 stations.


Conclusion

Therefore, I can conclude that using double mass curve to analyze homogeneity of precipitation

data is accurate. I can conclude that having a correlation equal to one means the techniques we

use to assess the different weather parameter is reliable. We check the consistency of the data in

order to know if there were missing data that affect the estimation of the precipitation on a

specific area.

Guide Questions:

1. What is the importance of performing homogeneity test in rainfall data analyses?

The importance of performing homogeneity test in the rainfall data analysis is to know its

reliability to estimate precipitation for a specific period. The inconsistency of the data can

due to:

1. The environment of the instrument, whether the trees, air, and houses.

2. Lack of commitment of the observer

3. Mechanical malfunction such as tipping bucket rain gauge, by broken wires.

References

Hardison, J. S. (2960). Double Mass Curves. UDSpace.

You might also like