You are on page 1of 7

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/245287452

Precipitation Runoff Modeling Using Artificial


Neural Network and Conceptual Models

Article in Journal of Hydrologic Engineering · April 2000


DOI: 10.1061/(ASCE)1084-0699(2000)5:2(156)

CITATIONS READS

236 504

2 authors, including:

Momcilo Markus
University of Illinois, Urbana-Champaign
39 PUBLICATIONS 704 CITATIONS

SEE PROFILE

All content following this page was uploaded by Momcilo Markus on 29 November 2015.

The user has requested enhancement of the downloaded file.


PRECIPITATION-RUNOFF MODELING USING ARTIFICIAL NEURAL
NETWORKS AND CONCEPTUAL MODELS
By A. Sezin Tokar1 and Momcilo Markus2

ABSTRACT: Inspired by the functioning of the brain and biological nervous systems, artificial neural networks
(ANNs) have been applied to various hydrologic problems in the last 10 years. In this study, ANN models are
compared with traditional conceptual models in predicting watershed runoff as a function of rainfall, snow water
equivalent, and temperature. The ANN technique was applied to model watershed runoff in three basins with
different climatic and physiographic characteristics—the Fraser River in Colorado, Raccoon Creek in Iowa, and
Little Patuxent River in Maryland. In the Fraser River watershed, the ANN technique was applied to model
monthly streamflow and was compared to a conceptual water balance (Watbal) model. The ANN technique was
used to model the daily rainfall-runoff process and was compared to the Sacramento soil moisture accounting
(SAC-SMA) model in the Raccoon River watershed. The daily rainfall-runoff process was also modeled using
the ANN technique in the Little Patuxent River basin, and the training and testing results were compared to
those of a simple conceptual rainfall-runoff (SCRR) model. In all cases, the ANN models provided higher
accuracy, a more systematic approach, and shortened the time spent in training of the models. For the Fraser
River, the accuracy of monthly streamflow forecasts by the ANN model was significantly higher compared to
the accuracy of the Watbal model. The best-fit ANN model performed as well as the SAC-SMA model in the
Raccoon River. The testing and training accuracy of the ANN model in Little Patuxent River was comparatively
higher than that of the SCRR model. The initial results indicate that ANNs can be powerful tools in modeling
the precipitation-runoff process for various time scales, topography, and climate patterns.

INTRODUCTION knowledge, and understanding of the model and the watershed


characteristics. In conceptual models, precipitation, air tem-
The precipitation-runoff relationships are among the most perature, and/or evaporation data are usually employed as in-
complex hydrologic phenomena to comprehend due to the tre- put data. In addition, streamflow data are required for the cal-
mendous spatial and temporal variability of watershed char- ibration of the model. Although well-calibrated conceptual
acteristics, snowpack, and precipitation patterns, as well as a models provide reasonable forecast accuracy, their uses are
number of variables involved in modeling the physical pro- limited only to a small number of watersheds due to the dif-
cesses. For many years, hydrologists have attempted to un- ficulties discussed.
derstand transformation of rainfall and snow to runoff in order
Artificial neural networks (ANNs) have been in existence
to forecast streamflow for water supply, flood control, irriga-
since the 1940s, but since current algorithms have overcome
tion, drainage, water quality, power generation, recreation, and
fish and wildlife propagation. The transformation of precipi- the limitations of those early networks great interest in the
tation to watershed runoff involves many highly complex com- practical applications of ANNs has arisen in recent decades
ponents, such as interception, depression storage, infiltration, (Wasserman 1989; Muller and Reinhardt 1990). Various ANN
overland flow, interflow, percolation, evaporation, and tran- algorithms have an objective to map a set of inputs to a set of
spiration. Conceptual models provide daily, monthly, or sea- outputs. An ANN is described as an information processing
sonal estimates of streamflow for short- and long-term fore- system that is composed of many nonlinear and densely in-
casting on a continuous basis. The entire physical process in terconnected processing elements or neurons. ANNs have been
the hydrologic cycle is mathematically formulated in the con- proven to provide better solutions when applied to (1) complex
ceptual models. Thus, they are composed of a large number systems that may be poorly described or understood; (2) prob-
of parameters. The Sacramento soil moisture accounting lems that deal with noise or involve pattern recognition, di-
(SAC-SMA) model is defined by 22 parameters in addition to agnosis, abstraction, and generalization; and (3) situations
12 parameters required by the potential evaporation (‘‘Na- where input is incomplete or ambiguous by nature. It has been
tional’’ 1996). The number of water balance (Watbal) model reported that an ANN has the ability to extract patterns in
parameters can be much larger and can often exceed 50, or phenomena and overcome difficulties due to the selection of
even 100, for larger basins with a large number of computa- a model form such as linear, power, or polynomial. An ANN
tional units (Markus and Baker 1994). The simple conceptual algorithm is capable of modeling the rainfall/snowmelt-runoff
rainfall-runoff (SCRR) model has seven fitting coefficients and relationship due to its ability to generalize patterns in noisy
two storage elements (McCuen and Snyder 1986). Since there and ambiguous input data and to synthesize a complex model
are numerous model parameters, and the interaction of these without prior knowledge or probability distributions. The ANN
parameters is highly complicated, the optimization of the
model is calibrated using automatic calibration techniques.
model parameters is usually accomplished by a trial-and-error
Thus, an ANN model eliminates subjectivity and lengthy cal-
procedure. Therefore, the accuracy of the model predictions is
very subjective and highly dependent on the user’s ability, ibration cycles.
In the literature some of the hydrologic applications of
1 ANNs include modeling of the daily rainfall-runoff process
Hydro., Consultant, Nat. Weather Service, 1325 East-West Hwy., Sil-
ver Spring, MD 20910-3283. and snowmelt-runoff process, assessment of stream ecological
2
Sr. Hydro., Michael Baker, Jr., Inc., Alexandria, VA. and hydrological responses to climate change, rainfall fore-
Note. Discussion open until September 1, 2000. To extend the closing casting, sediment transport prediction, pier scour estimation,
date one month, a written request must be filed with the ASCE Manager and ground-water remediation (French et al. 1992; Trent et al.
of Journals. The manuscript for this paper was submitted for review and
possible publication on October 5, 1999. This paper is part of the Journal
1993a,b; Roger and Dowla 1994; Hsu et al. 1995; Markus et
of Hydrologic Engineering, Vol. 5, No. 2, April, 2000. 䉷ASCE, ISSN al. 1995; Poff et al. 1996; Tokar 1996; Shamseldin 1997; Tokar
1084-0699/00/0002-0156–0161/$8.00 ⫹ $.50 per page. Paper No. 22049. and Johnson 1999).
156 / JOURNAL OF HYDROLOGIC ENGINEERING / APRIL 2000
REVIEW with a preassigned value, and combines these weighted inputs.
The combination of the weighted inputs is represented as
Artificial Neural Networks
An ANN is an information processing system that is com-
netj = 冘 wij xi (1)
posed of a number of processing elements (or neurons) anal- where netj = summation of weighted input for the jth neuron;
ogous to biological neurons and interconnections (or weights) wij = weight from the ith neuron in the previous layer to the
between these elements that imitate the synaptic strength in a jth neuron in the current layer; and xi = input from the ith to
biological nervous system (Fig. 1). In an ANN architecture, the jth neuron. The netj is either compared to a threshold or
the neurons are arranged in groups called layers. Each neuron passed through a transfer function to determine the level of
in a layer operates in logical parallelism. Information is trans- activation (Fig. 1). If the activation of a neuron is strong
mitted from one layer to another in serial operations (Hecht- enough, it produces an output that is sent as an input to other
Nielsen 1990). A network can have one or several layers. The neurons in the successive layer. The hyperbolic tangent and
basic structure of a network usually consists of three layers— sigmoid are often employed as transfer functions in the train-
the input layer, where the data are introduced to the network; ing of networks.
the hidden layer(s), where data are processed; and the output In this study, the training of ANNs was accomplished by a
layer, where the results for given inputs are produced (Fig. 2). back-propagation algorithm. Back-propagation is the most
The most distinctive characteristic of an ANN is its ability commonly used supervised training algorithm in the multi-
to learn from examples. Learning (or training) is defined as layer-feedforward networks. In back-propagation networks, in-
self-adjustment of the network weights as a response to formation is processed in the forward direction from the input
changes in the information environment. When a set of inputs layer to the hidden layer(s) and then to the output layer (Fig.
is presented, a network adjusts its weights in order to approx- 2). The objective of a back-propagation network is to find the
imate the target output (observed or measured output) based weights that approximate target values of output with a se-
on a certain algorithm. Learning in ANNs consists of three lected accuracy. The least-mean-square-error method, along
elements—weights between neurons that define the relative with the generalized-delta rule, is used to optimize the network
importance of the inputs, a transfer function that controls the weights in back-propagation networks. The gradient-descent
generation of the output from a neuron, and learning laws that method, along with the chain rule of the derivative, is em-
describe how the adjustments of the weights are made during ployed to modify the network weights. Detailed information
training (Caudill 1987). During learning, a neuron receives on learning rules and back-propagation is available (Rumelhart
inputs from the input or previous layer, weights each input et al. 1986; Hecht-Nielsen 1990).

Conceptual Models

Watbal Model
The Watbal model is a physically based model used for
forecasting water yields in areas where runoff is dominated by
snowmelt (Leaf and Brink 1973, 1976; Leaf and Alexander
1975; Markus and Baker 1994). It simulates winter snow ac-
cumulation, shortwave and longwave radiation balance, snow-
pack conditions, snowmelt, evapotranspiration, and subsequent
runoff in time and space.
Shortwave and longwave radiation represents the energy
components available for snowmelt. In the Watbal model,
shortwave radiation to snow or the ground surface beneath the
forest canopy is controlled by a transmissivity coefficient,
which varies with forest cover characteristics. The incident
FIG. 1. Artificial Neuron (xn = nth Input; w In j = Weight from n th shortwave radiation as measured on a horizontal surface is
Input to jth Neuron in I th Layer; Netj = Weighted Inputs of j th adjusted based on the slope and aspect of each hydrologic
Neuron; f ( ) = Transfer Function; and outj = Output of j th Neuron) response unit. Longwave radiation is computed by the Stefan-
Boltzmann equation (Leaf and Brink 1973, 1975).
Snowpack reflectivity varies with precipitation form, air
temperature, and the energy balance. During the winter
months, temperatures within the snow cover are simulated us-
ing the unsteady heat flow theory. The snowpack will yield
snowmelt only when it becomes isothermal at 0⬚C, and its free
water holding capacity is reached.
Evapotranspiration is derived from the Hamon equation
(Leaf and Brink 1973, 1975) for potential evapotranspiration
reduced in proportion to the radiation actually received. The
soil water content is computed for each time increment. The
critical point, at which the soil water content begins to limit
evapotranspiration, varies with time and forest tree species.
Forest cover density plays an important role in the simulation
model. It is the major descriptive parameter of the form, struc-
ture, and arrangement of forest stands, and therefore controls
the energy balance, interception, and evapotranspiration.
Reflectivity is described as a function of the forest stand
FIG. 2. Schematic of Back-Propagation Network reflectivity, forest opening reflectivity, and natural forest cover
JOURNAL OF HYDROLOGIC ENGINEERING / APRIL 2000 / 157
density. Interception is given as a function of forest cover. The parameter that distributes the surface runoff. The schematic of
combined evaporation from snow surfaces and intercepted the model is presented in Fig. 3. Daily precipitation, air tem-
snow is a function of potential evapotranspiration, intermedi- perature, and streamflow data are employed for the calibration
ate forest cover density, and forest cover density. Many other of the model. The objectives are to approximate a combination
processes, such as snow redistribution, tree cutting, ground- of the following: total measured runoff, peak discharges, tim-
water flow, etc., that are significant factors influencing runoff ing of peak discharges, soil moisture state at the beginning of
are also modeled in the Watbal model. the record, recession of the storm events, and bias in the flow
during dry periods. Since there are multiple objective func-
SAC-SMA Model tions, the calibration can be accomplished by the subjective
optimization technique.
The SAC-SMA model is a conceptually based, lumped,
rainfall-runoff model that represents spatially heterogeneous CASE STUDIES
runoff processes for river basins at scales ranging from a few
hundred to thousands of square kilometers. The SAC-SMA The numerical algorithm based on ANNs was applied to
model is an operational rainfall-runoff model that is used in three watersheds in the continental United States (Table 1).
the National Weather Service river forecast system (‘‘United Historical measurements of precipitation (P), temperature (T),
States’’ 1972). The model has six soil moisture states and 16 snow-water equivalent (SWE), and stream discharge (Q) are
parameters, not counting the 12 monthly adjustment factors of available for the Fraser River, the Raccoon River, and the Lit-
potential evaporation. Most of the parameters have to be cal- tle Patuxent River watersheds.
ibrated using historical hydrometeorological data.
In this model, the watershed consists of two areas—a per- Fraser River
vious area that produces runoff only when precipitation ex-
ceeds a specific amount, and an impervious area that produces The ANN technique was used to develop a one-step,
runoff regardless of the amount of precipitation. In pervious monthly flow forecast model for the Fraser River in central
areas, the soil profile is divided into two vertical zones—an Colorado. The Fraser River is a mountainous, subalpine river,
upper zone and a lower zone. The upper zone represents the with steep slopes, coniferous vegetation, and large snowpack
amount of water held in the interception and upper-soil layer, accumulation in wintertime. The discharge of the Fraser River
and the lower zone represents the amount of water retained in is dominated by snowmelt in the late spring and early summer.
the lower zone and ground-water storage. Both zones are di- The Fraser River flows into the Colorado–Big Thompson res-
vided into two types of storage-—tension water and free water ervoir system. Therefore, timely and accurate forecasts of
storage. Tension water represents the field capacity, which is monthly river volumes result in significant economic benefits.
the amount of water held in the soil zones due to binding with In the Fraser River watershed, data are available for the
soil particles. Once the tension water is satisfied, water starts periods of 1951–83 and 1987–93. In calibration of the ANN
accumulating in free water storage, and from there it moves and Watbal models, only data for high-flow months, May,
either laterally or vertically. In the upper zone, free water June, and July, were employed. The Watbal model was cali-
moves laterally to the streams as interflow and vertically to brated using the data for the entire record, and tested using
the lower zone as percolation. The free water from the lower data for the period from 1981 to 1983 and 1987–93. However,
zone either constitutes the base flow or feeds to the ground- the ANN model was trained using only data extending from
water storage. 1951 to 1980, and then tested for the period from 1981 to
1983 and 1987–93. A standard back-propagation algorithm
SCRR Model and the sigmoid transfer function were applied in training of
the networks. A one-hidden-layer network with two hidden
The conceptual rainfall-runoff model, SCRR, that was de- neurons was used, and the number of iterations for the training
veloped by McCuen and Snyder has three storage components varied from several hundred to several thousand, depending
—surface, ground water, and the unsaturated zone (McCuen on the particular month. The watershed runoff was modeled
and Snyder 1986). The rainfall first enters to the surface and as follows using ANN:
ground-water systems depending on the model parameter that
defines the fraction of infiltrated water into the ground. Water Q(t) = f {Q(t ⫺ 1), P(t ⫺ 1), SWE(t ⫺ 1), SWE(t ⫺ 2), T(t ⫺ 1)}
that enters to the surface storage contributes to the stream sys- (2)
tems using a unit hydrograph with a time base of four time
where Q(t) and Q(t ⫺ 1) = streamflows at times t and t ⫺ 1,
intervals. The unit hydrograph ordinates are based on a model
respectively; P(t ⫺ 1) = precipitation at time t ⫺ 1; SWE(t ⫺
1), SWE(t ⫺ 2) = snow water equivalents at times t ⫺ 1 and
t ⫺ 2, respectively; T(t ⫺ 1) = air temperature at time t ⫺ 1;

TABLE 1. General Information for Basins Used in Case Stud-


ies
Middle Little
Fraser Raccoon Patuxent
Parameter River River River
(1) (2) (3) (4)
Station Near Granby, Near Bayard, Near Guilford,
Colo. Iowa Md.
Latitude 40⬚08⬘49⬙ 41⬚46⬘43⬙ 39⬚10⬘04⬙
Longitude 105⬚56⬘31⬙ 94⬚29⬘33⬙ 76⬚51⬘07⬙
Drainage area (km2) 458.2 960.0 98.4
Mean annual discharge
(m3/s) 6.2 7.4 1.32
FIG. 3. Flowchart for Simple Conceptual Rainfall-Runoff Period of record 1948–94 1978–93 1968–93
Model, SCRR (McCuen and Snyder 1986)

158 / JOURNAL OF HYDROLOGIC ENGINEERING / APRIL 2000


and t = one month. The testing results for the ANN and Watbal TABLE 2. Statistical Indices for Testing ANN and Watbal Mod-
models are presented in Table 2. els Developed Using Monthly Streamflow in Fraser River, Colo.
The ANN model provided higher streamflow forecast ac- May June July Average
curacy compared to the forecasts of the Watbal model for high- 2 2 2
Model Se /Sy R Se /Sy R Se /Sy R Se /Sy R2
flow months based on the ratio of root-mean-square error (Se)
(1) (2) (3) (4) (5) (6) (7) (8) (9)
to the standard deviation of the observed data (Sy), and coef-
ficient of determination (square of the correlation coefficient) Watbal 0.95 0.10 0.69 0.52 0.69 0.52 0.78 0.38
ANN 0.77 0.41 0.68 0.54 0.45 0.80 0.63 0.58
between the predicted and observed discharges (R 2). The ratio
of Se /Sy for testing the ANN model was significantly lower for Note: Se and Sy are root-mean-square error and standard deviation of
May and July, and about the same for June, as illustrated in observed discharge, respectively. R 2 represents coefficient of determina-
tion between predicted and observed discharges.
Table 2. These results indicate that the ANN model has a
higher accuracy than that of the Watbal model for forecasting
monthly streamflows in the Fraser River, despite the fact that TABLE 3. Statistical Indices for Calibration of ANN and SAC-
the testing data set was included in the calibration of the Wat- SMA Models for Middle Raccoon River, Iowa
bal model. The ANN model is also simpler, and takes less
Model Sy Se Se /Sy R2
time to calibrate compared to the Watbal model.
(1) (2) (3) (4) (5)
Observed 16.0 — — —
Middle Raccoon River Model 1 11.7 10.5 0.66 0.57
Model 2 12.2 10.8 0.67 0.55
Various ANN models were trained to simulate daily rainfall-
Model 3 12.0 10.2 0.64 0.60
runoff relationships for the Raccoon River near Bayard, Iowa. Model 4 14.7 6.4 0.40 0.84
The loss of life and property in this region due to the Great SAC-SMA 16.7 6.7 0.42 0.84
Flood of 1993 led to a need for improved predictions to sup- Note: Se and Sy are root-mean-square error and standard deviation of
port flood/drought management and damage mitigation (‘‘The observed discharge, respectively. R 2 represents coefficient of determina-
Great’’ 1994). The SAC-SMA model is currently used by the tion between predicted and observed discharges.
National Weather Service for rainfall-runoff calibration and
operational forecasting (‘‘National’’ 1996). The calibration ac-
curacy of the ANN was compared to the calibration accuracy
of the SAC-SMA model. TABLE 4. Statistical Indices for Training Average, Dry, and Wet
Years Using ANN and SAC-SMA for Middle Raccoon River, Iowa
In this study, four different ANNs were trained using the
back-propagation algorithm. In training of networks, the hy- Average Dry Wet
perbolic tangent was used as an activation function. The net- Model Se /Sy R2 Se /Sy R2 Se /Sy R2
works were trained and tested using NeuralWorks software (1) (2) (3) (4) (5) (6) (7)
(Professional 1993). A one-hidden-layer network was used for
Model 1 0.50 0.79 1.09 0.39 0.48 0.81
training of each model. The number of neurons in the hidden Model 2 0.52 0.76 1.07 0.36 0.52 0.79
layer was 10 for models 1–3 and 25 for model 4. The ANN Model 3 0.53 0.80 1.21 0.37 0.44 0.87
architecture for each model was determined by a trial-and- Model 4 0.49 0.77 0.59 0.68 0.23 0.96
error process. The models are represented by (3)–(6) SAC-SMA 0.38 0.86 0.78 0.57 0.30 0.93
Note: Se and Sy are root-mean-square error and standard deviation of
Q(t) = f {P(t ⫺ 1), P(t ⫺ 2), P(t ⫺ 3), T(t)} (3) observed discharge, respectively. R 2 represents coefficient of determina-
tion between predicted and observed discharges.
Q(t) = f {P(t), P(t ⫺ 1), P(t ⫺ 2), P(t ⫺ 3), T(t)} (4)

Q(t) = f {P(t), P(t ⫺ 1), P(t ⫺ 2), P(t ⫺ 3), P(t ⫺ 4), T(t)} (5)
Three years were selected based on the mean annual daily
Q(t) = f {P(t ⫺ 1), P(t ⫺ 2), P(t ⫺ 3), T(t), Q(t ⫺ 1)} (6) discharge values—average, dry, and wet years. As illustrated
in Table 4, the SAC-SMA model provided slightly higher ac-
where Q(t) = streamflow at time t; P(t ⫺ 1) = precipitation at curacy for the average year compared to the ANN models.
time t ⫺ 1; and T(t) = temperature at time t; note that t is in However, the SAC-SMA and models 1–3, with high Se to Sy
days. The calibration period for both the ANN and the SAC- ratios, fail to approximate the dry-year data as well as model
SMA extends from 1978 to 1993. The results are presented 4. In addition, model 4 provided the lowest Se to Sy ratio for
for the entire calibration period and for average-, dry-, and the wet year. As shown in Table 4, model 4 also had the high-
wet-year data sets, separately shown in Tables 3 and 4. est accuracy for the three years compared to the SAC-SMA
Model 4 has a slightly lower Se to Sy ratio, about 0.40, when and models 1–3. These results are explained by the addition
compared to the SAC-SMA model. Models 1–3 had higher Se of Q(t ⫺ 1). In model 4, the values of Q(t ⫺ 1) provided
to Sy ratios, ranging from 0.64 to 0.67, compared to both information that was not contained in precipitation and tem-
model 4 and the SAC-SMA model. As shown in Table 3, the perature.
SAC-SMA model reproduced observed standard deviation
fairly well, while the ANN models provided less variation of Little Patuxent River
streamflow values.
The SAC-SMA model underestimated the minimum and The Little Patuxent River is located in Maryland. It is a
maximum streamflows in the record, at about 69% and 11% rainfall-dominated stream exhibiting the behavior of a stable
of their observed values, respectively. While the ANN models ground-water and perennial runoff stream type (Poff et al.
closely approximated the maximum streamflow in the record, 1996). The streamflow regime in the Little Patuxent River con-
they significantly overestimated the low flow values. A poor sists of high flows from December to May. The highest pre-
performance of the ANNs in forecasting low flows is consis- cipitation variations are observed during the dry months, June
tent with the results reported in recent literature (Tokar 1996; through September, that are characterized by flash flood
Markus 1997). events. The ANN technique was used to model daily rainfall-
The sensitivity of model training accuracy to the content of runoff in the watershed. Networks defined by various combi-
data was assessed using three data sets within the training data. nations of rainfall, temperature, and discharge at present and
JOURNAL OF HYDROLOGIC ENGINEERING / APRIL 2000 / 159
TABLE 5. Statistical Indices for Training and Testing Using CONCLUSIONS
ANN and SCRR Models for Little Patuxent River, Md.
In the literature, the ANN methodology has been reported
Se /Sy Se /Sy R2 R2 to provide reasonably good solutions for circumstances having
Model Training Testing Training Testing complex systems that may be poorly defined or understood
(1) (2) (3) (4) (5) using mathematical equations, problems that deal with noise
ANN 0.46 0.42 0.78 0.83 or involve pattern recognition, and input data that are incom-
SCRR model 0.66 0.67 0.56 0.55 plete and ambiguous by nature. Because of these characteris-
Note: Se and Sy are root-mean-square error and standard deviation of tics, it was believed that ANN could be applied to model pre-
observed discharge, respectively. R 2 represents coefficient of determina- cipitation-runoff relationships. The ANN rainfall-runoff
tion between predicted and observed discharges. models exhibit the ability to extract patterns in the training
data.
The ANNs were applied to forecast monthly streamflows in
TABLE 6. Statistical Indices for Training and Testing Average,
the Fraser River basin, and daily streamflows in the Raccoon
Dry, and Wet Years Using ANN (Model 4) and SCRR Models for River basin (calibration only) and the Little Patuxent River
Little Patuxent River Watershed, Md. basin (calibration and testing). For the Fraser River, the ANN
provided more accurate monthly streamflow forecasts than did
Training Testing a physically based model. Similarly, for the Raccoon River,
Parameter Wet Dry Average Wet Dry Average the best-fit ANN model has reasonable calibration accuracy
(1) (2) (3) (4) (5) (6) (7) when compared to that of a conceptual model. Such perfor-
(a) SCRR Model mance is a result of adding the streamflow at t ⫺ 1 as an input
Se /Sy 0.68 0.64 0.63 0.68 0.74 0.63
to the ANN model. Finally, in the Little Patuxent River basin,
R2 0.54 0.59 0.60 0.54 0.45 0.60 the best-fit ANN model provided higher training and testing
accuracy compared to that of the simple conceptual rainfall-
(b) ANN Model
runoff model.
Se /Sy 0.44 0.84 0.47 0.37 0.66 0.39 These examples demonstrated that ANNs can accurately
R2 0.83 0.54 0.79 0.86 0.56 0.85
model nonlinear relationships between hydrologic inputs, rain-
Note: Se and Sy are root-mean-square error and standard deviation of fall, snow water equivalent, temperature, and output stream-
observed discharge, respectively. R 2 represents coefficient of determina- flow. The ANN models provided a systematic approach and
tion between predicted and observed discharges.
shortened time spent on training of models compared to the
conceptual models.
Although the models were applied to only three watersheds,
previous time periods were trained and tested to simulate the results presented here were encouraging, and demonstrated
streamflow using different ANN configurations. Based on av- a high potential for the application of neural networks to var-
erage precipitation, three data sets representing average, wet, ious precipitation-runoff modeling scenarios.
and dry years were selected for training and testing in order
to identify the rainfall-runoff behavior in the Little Patuxent APPENDIX. REFERENCES
River watershed. For training of the networks, 1979, 1980, and Caudill, M. (1987). ‘‘Neural network primer: Part I.’’ AI Expert, (Decem-
1984 were selected as the wet, dry, and average years. Simi- ber), 46–52.
larly for testing, 1989, 1991, and 1992 were chosen as the French, M. N., Krajewski, W. F., and Cuykendall, R. R. (1992). ‘‘Rainfall
wet, dry, and average years. The testing data were not directly forecasting in space and time using a neural network.’’ J. Hydro., 137,
1–13.
used in the training; however, in order to avoid overtraining
‘‘The Great Flood of 1993.’’ (1994). NOAA Natural Disaster Survey Rep.,
of the networks, networks were tested using a test set. For National Weather Service, Office of Hydrology, Silver Spring, Md.
different combinations of the input variables, networks were Hecht-Nielsen. (1990). Neurocomputing. Addison-Wesley, Reading,
trained using a one-hidden-layer network (Tokar 1996). The Mass.
training of networks was accomplished using the back-prop- Hsu, K, Gupta, V. H., and Sorooshian, S. (1995). ‘‘Artificial neural net-
agation algorithm of NeuralWorks (Professional 1993). The work modeling of the rainfall-runoff process.’’ Water Resour. Res.,
hyperbolic tangent was selected as an activation function in 31(10), 2571–2530.
Leaf, C. F., and Alexander, R. R. (1975). ‘‘Simulating timber yields and
training of the networks. The selection of the best-fit model hydrologic impact resulting from timber harvest on subalpine water-
was accomplished by a trial-and-error process (Tokar 1996). sheds.’’ USDA Forest Service Res. Paper RM-133, U.S. Department of
The best-fit model was the network with 10 hidden neurons, Agriculture, Washington, D.C.
and is given below Leaf, C. F., and Brink, G. E. (1973). ‘‘Hydrologic simulation model of
Colorado subalpine forest.’’ USDA Forest Service Res. Paper RM-107,
Q(t) = f {P(t), P(t ⫺ 1), T(t)} (7) U.S. Department of Agriculture, Washington, D.C.
Leaf, C. F., and Brink, G. E. (1975). ‘‘Land use simulation model of the
subalpine coniferous forest zone.’’ USDA Forest Service Res. Paper
where t is in days. The SCRR model for the Little Patuxent RM-135, U.S. Department of Agriculture, Washington, D.C.
River watershed (Tokar 1996) was calibrated under the super- Markus, M. (1997). ‘‘Application of neural networks in streamflow fore-
vision of R. H. McCuen, who originally developed the SCRR casting,’’ doctoral dissertation, Colorado State University, Fort Collins,
model (McCuen and Snyder 1986). For the conceptual model, Colo.
the Se /Sy values are 0.66 and 0.67 for calibration and verifi- Markus, M., and Baker, D. (1994). ‘‘The Fraser River: Streamflow fore-
casting and simulation computer package.’’ Tech. Rep., Northern Col-
cation, respectively. As indicated in Table 5, the forecast ac- orado Water Conservancy District, Loveland, Colo.
curacy for SCRR is relatively poor compared to that of the Markus, M., Salas, J. D., and Shin, H. (1995). ‘‘Predicting streamflows
ANN model, with low Se /Sy values of 0.46 and 0.42 for train- based on neural networks.’’ 1st Int. Conf. on Water Resour. Engrg.,
ing and testing, respectively. As shown in Table 6, the ANN ASCE, New York.
model provided a significantly higher training accuracy for the McCuen, R. H., and Snyder, M. W. (1986). Hydrologic modeling: Statis-
tical methods and applications. Prentice-Hall, Englewood Cliffs, N.J.
average and wet years compared to that of the SCRR, whereas
Muller, B., and Reinhardt, J. (1990). Neural networks: An introduction.
SCRR illustrated better training accuracy for the dry year. Springler, New York.
However, the ANN model demonstrated significantly higher ‘‘National Weather Service river forecast system.’’ (1996). National
testing accuracy for the three years. Weather Service, Office of Hydrology, Silver Spring, Md.

160 / JOURNAL OF HYDROLOGIC ENGINEERING / APRIL 2000


Poff, L. N., Tokar, A. S., and Johnson, P. A. (1996). ‘‘Stream hydrological Tokar, A. S., and Johnson, P. A. (1999). ‘‘Rainfall-runoff modeling using
and ecological responses to climatic changes assessed with an artificial artificial neural networks.’’ J. Hydrologic Engrg., ASCE, 4(3), 232–
neural network.’’ Limnology and Oceanography, 41(5), 857–863. 239.
Professional II/PLUS and NeuralWorks Explorer. (1993). NeuralWare, Trent, R., Molinas, A., and Gagarin, N. (1993a). ‘‘An artificial neural
Pittsburgh. network for computing sediment transport.’’ Proc., ASCE Hydr. Conf.,
Roger, L. L., and Dowla, F. U. (1994). ‘‘Optimization of groundwater ASCE, New York.
remediation using artificial neural networks with parallel solute trans- Trent, R., Molinas, A., and Gagarin, N. (1993b). ‘‘Estimating pier scour
port modeling.’’ Water Resour. Res., 30(2), 457–481.
with artificial neural networks.’’ Proc., ASCE Hydr. Conf., ASCE, New
Rumelhart, D. E., McClelland, J. L., and PDP Research Group. (1986).
Parallel distributed processing—Volume I: Foundations. MIT Press, York.
Cambridge, Mass. ‘‘United States Department of Commerce, National Oceanic and Atmos-
Shamseldin, A. (1997). ‘‘Application of a neural network technique to pheric Administration, National Weather Service, and National Weather
rainfall-runoff modeling.’’ J. Hydro., 199, 272–294. Service River Forecast System Procedures.’’ (1972). Tech. Memo. NWS
Tokar, A. S. (1996). ‘‘Rainfall-runoff modeling in an uncertain environ- HYDRO-14, National Weather Service, Silver Spring, Md.
ment,’’ doctoral dissertation, University of Maryland, College Park, Wasserman, P. D. (1989). Neural computing theory and practice. Van
Md. Nostrand Reinhold, New York.

JOURNAL OF HYDROLOGIC ENGINEERING / APRIL 2000 / 161

View publication stats

You might also like