Sie sind auf Seite 1von 4

www.tjprc.org editor@tjprc.

org
International Journal of Civil, Structural,
Environmental and Infrastructure Engineering
Research and Development (IJCSEIERD)
ISSN(P): 2249-6866; ISSN(E): 2249-7978
Vol. 4, Issue 2, Apr 2014, 15-18
TJPRC Pvt. Ltd.

RAINFALL RUNOFF MODELING USING MODEL TREE TECHNIQUES
SEEMA A. JAGTAP
1
& S. K. UKARANDE
2

1
Assistant Professor, Department of Civil Engineering, Yadavrao Tasgaonkar College of Engineering,
Bhivpuri, Karjat, Maharashtra, India

2
Principal, Yadavrao Tasgaonkar College of Engineering, Bhivpuri, Karjat, Maharashtra, India


ABSTRACT
Earth is known as a blue planet due to presence of water on it. Near about 71% of earth is occupied by water.
Hydrology is considered as the study of various processes of the life cycle of water. One of the most important processes in
the cycle is transformation of rainfall into runoff. Predicting one-day a head runoff in rivers demands careful attention both
to reliability of forecasts and safety of decisions made on the basis of results. This paper presents the application of M5
model tree (MT) for predicting one day ahead runoff at Nigoje, Velholi and Paud in river basin namely Krishna River in
Maharashtra state, India. MT is a relatively new technique but has shown promise as shown by Solomatine and Dulal
(2003) particularly in modeling runoff.
KEYWORDS: M5 Model Trees, Runoff, and Rainfall
INTRODUCTION
Since rainfall records of longer duration than runoff data are normally available for basin many methods can be
adopted for synthetic generation of runoff data. Earlier models using large amount of heterogeneous as well as exogenous
hydrological data were not able to give a quick and accurate solution to the problems. In this paper M5 model tree
technique is discussed. M5 model tree is a method developed by Quinlan (1992) for inducing trees of linear regression
models. In the present paper runoff-runoff model and rainfall- runoff model to forecast runoff one day in advance were
developed at three rain gauge stations namely, Velholi, Nigoje and Paud in Krishna river basin, which are along the
tributary Bhima and along streams Indrayani, Kundalika and Mula. The data was taken from Hydrology Department, Meri,
Nashik. For this daily rainfall and runoff data from 1994 to 2007 subsets are prepared by dividing the data used for training
and testing for development of models to predicting the one-day ahead runoff.
The present work the predictions were carried out using the relatively advanced technique of Model Tree, which
is piecewise linear models as ARIMA and truly nonlinear model as ANN. Model Tree is a linear regression method based
on an assumption of linear dependencies between input and output.
MODEL TREE
A model tree (MT) is an extension of the concept of classification tree and regression tree in which the
computational process is represented by a tree structure consisting of a root node (decision box) branching out to numerous
other nodes and leaves. In M5 model tree a step towards non-linearity is made since it built a model that locally linear but
overall is non-linear. In M5 model tree after examining all possible spits. M5 chooses the one that maximize the expected
error reduction by applying pruning or smoothing.
16 Seema A. Jagtap & S. K. Ukarande

Impact Factor (JCC): 5.7179 Index Copernicus Value (ICV):3.0
MODEL DEVELOPMENT
Total fourteen years data for rainfall and discharge (runoff) from the year 1994 to 2007 was available for
developing the runoff models .It was decided to develop runoff models particularly for the monsoon season will hereafter
be called as NgjJune, NgjJuly, NgjAug, NgjSept and NgjOct for Nigoje station, and VelJune, VelJuly, VelAug, VelSept
and VelOct for Velholi station and PaudJune, PaudJuly, PaudAug, PaudSept and PaudOct for Paudstation.
For the two locations using 70% of available data for training and 30% data for testing purpose develops models.
First two i.e. Qt-1 (discharge for the previous day) and Qt (discharge for the day) were chosen as inputs and Qt+1
(discharge for the next day) was the output. Then next was Qt-2 (discharge for 2 day previous), Qt-1, Qt was the input and
Qt+1 was output. Next was Qt-3 (discharge for 3 day previous), Qt-2, Qt-1, Qt was the input and Qt+1 was output.
Thus 3 models were developed for each monsoon month at each station using previous values of discharge alone.
Rainfall data was used for model formation i.e. in addition to runoff; rainfall data was used for training and testing the
model for prediction of runoff. Model formed with Qt-1, Qt, Rt(rainfall for the day) as input and Qt+1 as output. Next
model was Qt-1, Qt, Rt-1 (rainfall for the previous day) and Rt as input and Qt+1 as output. Next model was Qt-2, Qt-1,
Qt, Rt as input and Qt+1 as output. Next model was Qt-2, Qt-1, Qt, Rt-1 and Rt as input and Qt+1 as output.
Next model was Qt-3, Qt-2,Qt-1, Qt, Rt as input and Qt+1 as output. Next model was Qt-3, Qt-2, Qt-1, Qt, Rt-1
and Rt as input and Qt+1 as output. Thus in total 9 models (3 with previous values of discharge and 6 with previous values
of both discharge and rainfall) were developed for each month at each station. From these different trials the best model
was selected for each month for each station depending on the statistical properties such as coefficient of correlation, root
mean squared error as well as by plotting hydrographs and scatter plots.
Table 1: Average Discharges (HQC) Obtained from Fourteen Years
Data for Station Velholi, Nigoje and Paud
Average of
HQC_(m_sec)
Station or Site Name
Month Velholi Nighoje Paud
June 3.053170334 19.03308222 6.484597875
July 19.28918916 103.7131933 38.59987895
August 22.1896802 105.4542076 45.41793195
September 7.696626724 39.38973067 15.36215402
October 1.246045642 11.00894314 2.081788074


Figure 1: Daily Average Discharge
Rainfall Runoff Modeling Using Model Tree Techniques 17

www.tjprc.org editor@tjprc.org

Table 2: Average Rainfall (MPS) Obtained from Fourteen Years
Data for Station Velholi, Nigoje and Paud
Average of MPS (mm) Station or Site Name
Month Velholi Nighoje Paud
June 15.86714 6.417179 12.1269
July 31.97996 6.595782 21.84585
August 23.02488 6.158313 16.79892
September 8.62119 4.753077 6.590476
October 2.418433 2.700496 2.297696


Figure 2: Daily Average Rainfall
RESULTS AND DISCUSSIONS
The above-developed nine different models for each monsoon month that is June, July, August, September, and
October for each station were tested for their performance using statistical parameters and plotting hydrographs and scatter
plots.
Table 3: Correlation Coefficient between the Observed and Predicted Discharge
Values for All the Models Developed Using Model Trees
Sr.No Model Linear Models RMSE Correlation Coefficient r
01 NigJune2 2 54.07 0.91
02 NigJuly2 1 51.82 0.88
03 NigAug 7 3 96.44 0.80
04 NigSept 9 3 35.03 0.80
05 NigOct 9 4 3.69 0.94
06 VelJune 4 1 6.55 0.93
07 VelJuly 9 4 12.35 0.84
08 VelAug 4 7 22.60 0.79
09 VelSept 6 1 3.08 0.77
10 VelOct 9 5 0.21 0.96
11 PudJune 7 1 4.07 0.97
12 PudJuly 9 1 34.84 0.83
13 PudAug 7 1 35.24 0.79
14 PudSept 2 1 8.08 0.65
15 PudOct 9 1 1.09 0.96

18 Seema A. Jagtap & S. K. Ukarande

Impact Factor (JCC): 5.7179 Index Copernicus Value (ICV):3.0

Figure 3: NgjOct Hydrograph

Figure 4: Nig Sept Scatter Plot
CONCLUSIONS
Initially the models were developed using only the previous discharge values. After that rainfall was also added as
input. The effect of rainfall improves the results obtained from the model. Inclusion of other influential parameters like
evaporation, infiltration may improve the model prediction. After various trials it was found that the testing accuracy has
not increased with more than five inputs, both in precipitation and runoff. The work is definitely help practically to predict
the runoff within few seconds.
REFERENCES
1. Solomatine, D.P. and Dulal, K. N., Model tree as an alternative to neural network in rainfall-runoff modeling,
Hydrological Sc. J., Vol.48(3), (2003), pp 399-411.
2. Quinian, J.R. (1992), Learning with continuous classes. In: Proc. A/92 (Fifth Australian Joint Conf. on Artificial
Intelligence) (ed.by A.Adams & L Sterling), 343-348. Word scientific, Singapore.

Das könnte Ihnen auch gefallen