You are on page 1of 24

Agricultural Systems 47 (1995) l-24 1994 Elsevier Science Limited Printed in Great Britain 0308-521X/95/$09.

50

Combined Use of Watershed, Aquifer and Crop Simulation Models to Evaluate Groundwater Recharge through Percolation Ponds
M. Selvarajan
Water Technology Centre, Tamil Nadu Agricultural Nadu, India University, Coimbatore, Tamil

A. K. Bhattacharya
Water Technology Centre, Indian Agricultural Research Institute, New Delhi, India

F. W. T. Penning de Vries
Centre for Agrobiological Research, Wageningen, The Netherlands (Received 28 August 1992; accepted 25 May 1993) ABSTRACT Percolation ponds are constructed in many parts of the tropics to augment fast-depleting groundwater resources. Increased agricultural productivity due to additional well irrigation water is the direct benefit from the pond schemes. Usually, location-spectfic empirical relationships are adopted for hydrologic and economic design of ponds. In this paper, an approach is presented to evaluate hydrologic pond design, by combining watershed, aquifer and crop simulation models of the pond system. A watershed model (CAPSIM) was developed based on the state variable approach describing the dynamic catchment-pond hydrologic processes such as runox runofl due to interjow, evapotranspiration, direct percolation, pumping and irrigation. CAPSIM operates on a daily basis, using rainfall and pan evaporation data. Based on the Alternating Direction Implicit Algorithm (ADIA), a finite d@erence Aquifer Simulation model (ASIM) was developed to simulate two-dimensional recharge mounds beneath the pond, Predictability of CAPSIM and ASIA4 was vertjied with observed field data from one pond site, from which the models were improved. The models were further validated separately for three pond locations and for dtferent years of weather data, by comparing observed pond and well water levels. The crop model, SAHEL, was adapted to the rice-growing environments of pond command areas, by representing water

M. Selvarajan et al. stress, runofl and well irrigation. Total Additional well Irrigation made possible Due to recharge (AID-recharge) in two space dimensions during a crop season was predicted by linking CAPSIM and ASIM. AIDrecharge thus computed was input to the crop model for predicting the spatial distribution of crop response to direrent well irrigation scenarios. A series of distance-capacity-yield curves were developed, which quanttfy the benefits associated with a modtjication in the pond structure. The batch of models thus linked could serve as a decision support tool for pond designers for evaluating desired hydrologic economic strategy.

INTRODUCTION Due to increased utilization of groundwater resources for intensive irrigation and domestic needs, a fast-declining trend in the groundwater table is observed in many parts of the world. Overdraft has significantly lowered the groundwater status of California (Nightingale & Bianchi, 1973). During the past five decades, India has witnessed a three-fold increase in the number of wells coupled with a 120-fold expansion in the number of pumpsets. Observations revealed that in many parts of India, there had been a consistent drop in groundwater table during the past 18 years (Subbiah & Bhavani, 1988). On the other hand, regional water budgets indicate that considerable surface flows are lost to the sea (Michael, 1978). Artificial groundwater recharge assumes greater importance as this would enable the replenishment of the fast-depleting groundwater sources by impounding the parts of avoidable wastes of surface flows to the sea. Thus, groundwater recharge enhances the role of aquifers as potential storage reservoirs. Percolation ponds (hereafter referred to as ponds are convenient ) artificial recharge structures. These ponds consist of a bund constructed across the watercourse, and detain a portion of runoff from the upstream catchment, facilitating seepage for a longer duration of time from the ponded water spread (Fig. 1). Huge investments are made in constructing pond structures in many parts of India. The direct benefit from these structures is an improvement in agricultural production. This is due to the augmented water supplies to the wells. The total agricultural area benefited due to a pond scheme is usually referred to as the command area Direct enquiries with the farmers . (the primary beneficiaries of pond schemes) indicated that significant improvements in yield were obtained after pond construction in many command areas of Tamil Nadu state, India (Selvarajan, 1990). However, some farmers reported low agricultural productivity due to inadequacy of available well irrigation supplies. Some indirect benefits are also de-

Watershed, aquifer and crop models of groundwater recharge

Fig. 1. Schematic representation of a typical pond recharge system. Ponded water fluctuations influence the quantum of recharge to the water table in the command area. Farmers near the pond are able to withdraw more water from their irrigation wells than those at far reaches.

rived from these structures, such as improved flood control, development of aquaculture, and restoration of environmental balance in the region. The problem At present, in India the organizations involved in constructing pond structures adopt empirical relationships to design and evaluate ponds (PWD, 1982; DAE, 1983a,b,c; GOI, 1984). As these relationships are derived from specific locations, their use as general tools of estimation is limited. For example, for the purpose of pond design in the Tamil Nadu state in India the effective zone of pond recharge (usually referred as the zone of influence is considered as the area within 945 m radius from ) the centre of the pond. However, sand tank experiments (Ramaprasad & Gangaraju, 1983) revealed that the zone of influence ranges from 2000 to 5000 m. GO1 (1984) recommended that for design purposes 50% of the gross storage could be taken as the annual recharge from ponds. Rao and Shanmugasundaram (1985) reported that a well near the pond recouped with 61% recharge from the pond. A simple water balance approach revealed that an aquifer could receive 47-69% of the stored volume (Thiruvenkatasamy & Sakthivadivel, 1969). The wide range of deviations in these recommendations indicates that there exists an amount of uncertainty associated with the use of empirical tools as means for recharge quantification.

M, Selvarajan et al.

Field experiences revealed that ponds with larger Water Spread Area (WSA) have better recharge (Executive Engineer, DAE, pers. comm.). The design engineers and pond executive agencies are in need of more dependable approaches to evaluate strategies such as increasing bund height, deepening the ponds, increasing the number of wells commanded by a pond, adjusting the allowable number of ponds in a region, etc. This study was undertaken with a view to developing an explanatory modelling approach describing the basic physical processes of pondaquifer-crop system behaviour. It was also envisaged to determine the elements of such an approach relating pond construction schemes in alluvial aquifers of southern India. A sample strategy evaluation is presented in which the productivity increase associated with increasing pond capacity is quantified, for a typical well irrigation system recharged by a pond scheme. The approach It is evident that several factors associated with the surface and sub-surface formations influence the behaviour of the pond system. Exclusive assessments (i.e. by elimination of the effect of other factors) on the following chain of processes are required in order to develop a comprehensive approach to evaluate recharge (Fig. 1) including: (1) pond responses (storage, WSA) to the runoff from the catchment; (2) aquifer responses (storage, spatial distribution) to recharge from the pond; (3) crop responses to additional irrigation made possible due to recharge. A model based on the dynamic system simulation approach was developed to simulate pond responses and a two-dimensional transient finite difference groundwater model was developed to simulate aquifer responses These two models were coupled through an interface. Another . interface was developed to link the predictions of these models to a crop model which simulates crop responses to water-limited environments. As a result of this, the spatial distribution of productivity increase around the pond was assessed, for different pond storage capacities. The details of these models are discussed in the subsequent sections.

THE MODELS CAtchment, Pond and Aquifer Simulation Model (CAP-ASIM) The state variable approach used in dynamic system simulation, in

Watershed, aquifer and crop models of groundwater recharge

RAINFALL

EVAPORATTION __----ty----____

w p 0

Rote orlable Rote changlnp s,me Mrlabler

the UOfF

_-

Fig. 2.

System relational diagram of the combined model CAP-ASIM showing various stateand ratevariables influencing the pond-aquifer-recharge system.

simple terms, could be summarized by the following equation (Rabbinge & De Wit, 1981): State (t + 8t) = State (at) + Rate (t + at)
X St (1)

where the term rate implies the quantity by which the state variable changes during a specified time interval. A system relational diagram comprising the entire catchment-pond-aquifer processes, their state and rate variables is presented in Fig. 2. This conceptual model (hereafter called CAP-ASIM) was developed with rainfall and evaporation as the chief climatologic inputs. PCSMP-a Personal computer Continuous Simulation Modelling Program (a FORTRAN-based simulation software)-was used in developing the model. For more details on this software see IBM (1975). Since PCSMP is a non-procedural language, the components describing the different processes of the recharge model could be in slices with their respective input data and run indepencut dently. Such sub-programs will be referred to hereafter as modules A . summary of various modules and their functions in the model are presented in Fig. 3. Essentially the CAP-ASIM is a combination of a one-dimensional CAtchment Pond Simulation Model (CAPSIM) and a two-dimensional finite difference Aquifer Simulation Model (ASIM). The former is a PCSMP main program and the latter is a FORTRAN sub-program.

h4. Selvarajan

et al.

Module Inltlal (0) (b) Ict (d) (1) (0) (hl (1) (iI

0 (MO) condition9

rnd

parametera

(el

curve number Seaacn IWlCPE ratio Catchment area Saturated thickneaa MaxImum pond capacit Recharge rate Irrigation depth Aquiler prcpertiea Well dimenaiona

I
[MO - g.h.k.ll

rI

(81 (b) (c) (d)

Depth-area-8tcrage AMC Converrion Direct recharge fraction Root cowtant w time rx

- e.

r J
-r
modul

Module 7(M7) Daily data (a) (b) Rainlall Pan evaporation MO - e,g.i.i,t

*aI

Two dimensional recharge mound

c
Input nicdules
Update tat tb)
L

Time atap *pecificationa Output apecitication~

Fig. 3.

Summary diagram of the model CAP-ASIM.

CAtchment Pond Simulation Model (CAPSIM, modules 1,2,3) From simple linear equations to complex routing models, a vast number of methods are available to estimate water yields from catchments (Wu et al., 1964; Sopper & Lall, 1965; Blank & Beer, 1968; Krishnaswamy, 1976; Ryan & Pereira, 1978; Rao & Minikou, 1983). Neither simple nor complex models are free from failures (Naef, 1981). Several attempts have focused on the general use of the hydrologic soil cover complex method as a general tool of runoff estimation, especially to small catchment tanks, under Indian soil conditions (Verma, 1987; WTC, 1989). Of the modules of Fig. 3, module 1 simulates daily runoff volume based on hydrologic soil cover and antecedent moisture content. Curve number, the season (growing or dormant), and certain minimum rainfall only above which runoff would occur, are inputs to this module. Based on the Antecedent Moisture Conditions (AMC), the curve number is further modified, using the non-linear function generators (module 6). Runoff 2, the interflow component, is computed using water balance of the root zone. For steady state conditions,

Watershed, aquifer and crop models of groundwater recharge

Runoff 2 = Rainfall - Runoff 1 - Actual ET - Direct percolation

(2)

Total runoff is obtained by adding runoff 1 and runoff 2. A function switch decides whether runoff would take place or not, based on the minimum rainfall criterion. Direct recharge from precipitation is common in shallow aquifers with loose soils (Senarath, 1981; Rushton and Senarath, 1983; GOI, 1984). Module 2 determines the quantum of direct recharge from precipitation. Direct recharge is computed as different fractions of rainfall which depends on the amount of rainfall. This is generated by a linear function in module 6. A fraction of irrigation from the paddyfields in the region contributes to recharge (Michael, 1978). Thus the total direct recharge from the surface was computed by adding the fraction of irrigation to the previously computed direct recharge. This fraction of irrigation was dynamically simulated using computed pumpage from the aquifer; i.e. a fraction of water pumped would be fed back to the aquifer, on the days when pumping occurs. Representation of pumpage was based on the recommended irrigation practices (TNAU, 1988). Depth of irrigation required, previous rainfall, desired ratio of irrigation depth to cumulative pan evaporation (IWCPE ratio) and the actual cumulative pan evaporation from the day of previous irrigation were considered to represent a decision on pumpage during a particular day; i.e. if the IW/CPE ratio of the required depth of irrigation to the cumulative pan evaporation exceeds a pre-assigned value on a day, pumpage would occur. This would be true only when previous rainfall is less than a specific value. The equivalent depth of pumpage thus obtained by module 8 is input to the main program for every time step of simulation. Daily changes in WSA of the pond are simulated in module 3. Computation of the WSA within this module involves an implicit loop as described below. The volume of the ponded water (the state variable) of the present time step is influenced by several rate variables of water balance. These variables are (Fig. 2): (1) (2) (3) (4) A Total runoff (i.e. Runoff 1 + Runoff 2) Evaporation from the water surface Recharge Domestic consumption of these variables could be summarized as (3)

daily water balance

(1) - (2) - (3) - (4) = Change in pond storage


Using storage-area step is

first time

function (module 6) generators, the WSA of the computed, as the initial storage is input to the model.

M. Selvarajan et al.

Function switches of PCSMP were used to simulate pond overflow, due to excess runoff. During the second time step the simulated WSA of the present time step is used to compute the rate variables (2), (3) and (4). These updated rate values are used to compute the WSA of the second time step and so on. Thus the implicit feedback loop is formed. The WSA at any given time is approximated to a rectangle. From the available observed data on depth-area-storage relationships the equivalent length and width of the pond are derived by linear interpolation, through function generators in module 6. Length and width values are supplied as inputs to the pond-aquifer interface in module 4. Aquifer system Simulation Model (ASIM, module 9) A schematic representation of the pond aquifer interface is presented in Fig. 4. Module 9 is developed based on a Finite Difference Scheme (FDS) which solves the Boussinisque equation of the form
1

(K,i+h;)

1 $

(4)

where

H= h2 - h; h= Height of water table from the impervious layer h, = Initial height of water table R= Constant recharge rate K= Hydraulic conductivity s= Specific yield &Y = Space coordinates t= Time since start of recharge

with respect to boundary conditions presented in Fig. 4. A set of assumptions as given by Marino (1975) forms a prelude in deriving eqn (4). Typical overlay of finite grids is shown in Fig. 5 for a pond area in Tamil Nadu state in India. The matrices of this FDS are solved based on the Alternating Direction Implicit Algorithm (ADIA) of the central difference formula of second-order derivatives, and is unconditionally stable (Smith, 1969; Remson et al., 1971; Rushton & Redshaw, 1979). The Gauss elimination process is used to solve the system of tridiagonal matrices (Smith, 1969), which contain the nodal elevations of the groundwater mound. This system of mathematical solutions is different from those developed by Prickett and Lonnquist (1971) and Marino (1975), which are based on the forward difference formula (described as Modified ADIA or MADIA by Rushton dz Redshaw, 1979). The MADIA requires a careful formulation on the number of iterations. The ADIA, in the

Watershed, aquifer and crop models of groundwater recharge

NODE (11.11) p. I \ 4 ) CONSTANT BOUNDARY HEAD

---+

SOLUTION DOMAIN

ORIGIN

AT (1,l

) NODE

i
I
I
1 -3 I / CONSTANT

I
RECHAR t E RATE, R

,&++

L Y&g+.

7,//j

,,,,,,,,
IMPERVIOUS

j//J,,
LAYER

Fig. 4. Pond-aquifer interface as conceptualized in module 9. For each time step the water table profile is predicted for a given WSA, using the finite difference scheme ADIA (see Fig. 5).

10

A4. Selvarajan

et al.

ram

water

and

qther bodles wells

0 0 PANNRUTI TALUK

Simulated

Other wells

ERoad e Streams

==m nmR T&N%-=-_=__ -_--------- ---___--__---_----_--------

Fig. 5. Typical overlay of finite grids in the pond 1 study area (Table 1).
Observed water table data on wells 1, 2 and 3 were used to evaluate the performance of the model CAP-ASIM. present case, was developed as a subroutine written in FORTRAN. This subroutine can be used as an independent program when detached from the PCSMP main program. In fact, this program was tested for several grid sizes independently, before attaching it as a subroutine to the

PCSMP main program. For ponds evaluated in the present study a grid

Watershed, aquifer and crop models of groundwater recharge

11

TABLE 1

Details of Pond Sites Used to Evaluate the Performance of the Model CAP-ASIM
Pond ref: Location (village in Tamil Nadu, India) Catchment area (ha) Hydraulic conductivity (m/d) Maximum water spread area (ha)

Pond 1 Krishnankuppam, Cuddalore Taluk Pond 2 Peranakkavur, Uthiramerur Taluk Pond 3 Sethupattu, Sriperumpudur, Taluk

100 52 60

5.4 48.0 624.0

16 2.2 2.6

size of 100 m X 100 m was found adequate. The main function of this module is to develop a groundwater head profile in two space dimensions for the selected time steps and for a given set of aquifer parameters and boundary conditions. For every time step the simulated length and width of the pond are input from the main program to this module. In order to meet convergence requirements, module 9 operates at much smaller time steps than the time steps of the main program. Thus, for a given time step of the main program (which is one day) the computations within this sub-program are iterated over 10-20 time steps. For more details on these aspects see Selvarajan (1990). Linker interface (module 4) Inputs for each time step to this module are derived from several modules. These include (a) the rate of change of groundwater mound for every node from module 9, (b) direct percolation from module 2, (c) pumpage from module 8, and (d) specific yield and initial well water levels of the aquifer from module 0. Of the above inputs, all except (d) are water balance rates of the current time step, which decides the position of the water table at a given time. An algebraic sum of these rates yields the net rate of change in the water table on a specific day. This net rate is integrated to give daily updated values of water table elevation. From the updated values of water table positions, the daily rates of flow to well at different radial distances are computed within this module using the steady-state flow equation (Michael, 1978). These daily rate values when integrated over a season give the total volume of flow to a well in that season. In the present approach, the computed flow to a well due to pond recharge is treated as additional irrigation water available

12

M. Selvarajan

et al.

for crop production in the area commanded by a specific well. Consequently, these values are input to a crop simulation model to depict benefits associated with different well irrigation scenarios. The crop model-SAHEL The pond command areas of southern India in general are characterized by rice cultivation. Some attempts were made in the past to quantify yield responses of rice for different irrigation regimes. By adopting the procedure of Wickham (1973), Palanisamy and Flinn (1988) reported the use of a linear regression equation. This equation was validated through observed yields in farmers fields. Deviations of the observed and predicted yields varied from 3% to 33%. Crop models of Penning de Vries et al. (1989) were adapted to several rice-growing environments of Tamil Nadu (Budhar et al., 1989; Selvarajan & Palanisamy, 1991). These modules simulate various crop production environments such as potential production, water limited production, etc. Of these, the one based on SAHEL (Soils in semi Arid Habitats that Easily Leach) was used to simulate crop responses for Additional Irrigation Due to recharge (hereafter referred as AID-recharge). A brief description of this concept follows. The modules of MACROS LlD + L2C + L2SU + (CROP + SOIL + WEATHER DATA) were used in simulating crop responses to irrigation during the wet season of the region. This concept is well suited for the soils of pond areas which have a free drainage. The module LlD simulates potential growth and development of the crop through explanatory representations on photosynthesis, respiration, phenological development, carbohydrate partitioning, leaf area and senescence, from sowing or transplanting to harvest with a time increment of one day. The modules L2C and L2SU simulate water balance processes in crop and soil with one-day time increments. Potential transpiration is computed using the approach of Penman (1948). Actual crop water uptake is a function of the potential transpiration, a water stress sensitivity factor, and the available soil water. Water stress influences the different crop development processes mentioned above. The root zone is compartmentalized into layers and daily water balance is computed to assess the moisture contents available for growth. In the present application of this model, crop data on rice variety IR 50 (Selvarajan et al., 1991), observed soil data in the region, and weather data for the year 1984 were used for analysis. Modifications were made to SAHEL to represent runoff in the same way as in CAP-ASIM. A subroutine was added to the model which permits reruns for different well irrigation scenarios (Fig. 6). Thus, the outputs in terms of total AID-

Watershed, aquifer and crop models of groundwater recharge

13

Ramfall

adequate?

L-

Crop

stage

within

krigation

domain?

u AID-recharge available

Irrigation

R
Y

Yes NO
OK?

interval

Irrigate

Do not Irrigate

Update

R AID-recharge Y crop

Update

growth

Switch Fii.

t to next

day

Flow diagram depicting the subroutine added to the SAHEL model in order to simulate well irrigation scenarios resulting from different AID-recharge levels.
6.

recharge from CAP-ASIM were input to the SAHEL model to predict crop responses for the same. Summary of the approach The AID-recharge from the pond is predicted by CAP-ASIM in terms of the available total irrigation flow to wells at different radial distances from the pond centre. The total flow to each nodal well is input to the SAHEL model to predict yields for different irrigation regimes at different radial distances from the pond. This procedure, if repeated for an alternative improvement strategy (such as increasing bund height, desilting, deepening the pond, etc.), would generate a chain of information on the spatial distribution of productivity in the pond command area. From such an approach, the total production and the associated cost of an improvement strategy can be worked out. A suitable improvement strategy could be chosen from such an analysis. In this paper a sample analysis is presented for the case of heightening the bunds as a strategy option.

14

M. Selvarajan et al.

RESULTS AND DISCUSSION


Evaluation of the models

Among the various modules of CAPSIM and ASIM, module 9 of ASIM was developed based on explanatory theoretical treatment of flow through porous media. Hence, the prediction of this module would be reasonably accurate as long as the aquifer properties are reliable. However, CAPSIM requires inputs such as curve number, minimum rainfall, crop coefficients, etc., which are values averaged over the pond region. The descriptive nature of these modules adds some uncertainty to their predictions. Model calibration was a term used by some previous workers (Rao & Sarma, 1981) to mean adjustment of certain parametric inputs such that the observed and predicted system responses become similar. Rushton and Redshaw (1979) explained the need to exercise caution in the use of calibration to validate models. It was emphasized that a combination of unrealistic parametric inputs may result in a close fit of observed and predicted variables. Three pond sites were chosen for evaluation of CAP-ASIM in the alluvial formations of Tamil Nadu (Table 1). Observed data on fortnightly pond and well water levels (40 well locations, three years), daily climatologic data, data on pond structure, infiltration, and hydraulic properties of the aquifer were collected from different research and executive agencies. Data on area commanded by wells, cropped area and farmers appraisal on benefits from pond schemes were collected through direct enquiry. CAP-ASIM was validated in two phases, firstly against observed pond levels and secondly against observed well levels. For each pond location three well locations were chosen for validation at head, middle and tail reaches (represented by well numbers 1, 2 and 3, respectively, in Fig. 5). Typical curves representing the observed and predicted behaviour of pond and well levels are shown in Fig. 7. In all the 23 simulation runs made for different pond locations/years, 19 runs were found statistically acceptable by the Wilcoxon matched pair signed rank test, which tests both magnitude and direction of the deviations in observed and predicted series (Siegel, 1956). The general performance of the model was found satisfactory for these pond locations, although there was considerable spatial variability in the general hydrologic features (such as rainfall distribution and aquifer properties) of these areas. The structure of the crop model has been validated extensively (Penning de Vries et al., 1989). As mentioned earlier, this model has been validated for several rice-growing situations of Tamil Nadu state in the past. A modification made with respect to runoff in the model has been evalu-

Watershed, aquifer and crop models of groundwater recharge

15

Groundwater

elevation

from

MSL,

40 - + +

Hpred - Near Hobs Hobs - Near - Middle Hprad - Middle

36 180

1
200

I
220 240

1
260

1
280 300

I
320 340 360

Julian

Day

Fig. 7. Sample figure of observed and predicted water table elevations for pond 2 during the irrigation season of the year 1986. Generally recharge is affected during two different monsoons. In the study, 23 such figures (for a combination of different seasons and ponds) were analysed for validating CAP-ASIM.

ated independently for parametric modules of CAP-ASIM. Depth-area-storage relations

inputs,

within

the pond

runoff

Typical depth-area-storage curves (observed) reveal that a small increase in the pond bund height (i.e. maximum depth of storage) results in an exponential increase in the storage and WSA of the pond (Fig. 8). Further, this exponential increase in WSA is associated with a small increase in pond capacity (Fig. 9). This behaviour, however, is due to the mild slope topographic conditions of the pond area. A study of the toposheets of southern India revealed that a vast tract of land over the alluvial formations falls under this category. These inferences indicate that in most locations of pond schemes of southern India, a relatively small increase in water harvest of the catchment through heightening the pond bund would result in considerable increases in the WSA of the pond. A comparison of predicted runoff and the changes in pond storage indicated that frequently the catchment runoff is in excess of the pond capacity. PWD (1985) reported that usually a full pond storage depletes completely after a period of 20 days. Thus, it is possible to improve catchment water harvest and duration of pond storage simultaneously, by increasing pond capacity, which would enhance recharge efficiency. Having checked the validity of CAP-ASIM, simulation runs were made

16

h4. Selvarajan et al.


Water spread area, lOOOm2 30 -Water 25 spread area + capacity

Capacity,

lOOOm3 25

20

15

10

0.5

1.5

0 2.5

Depth ,m

Fig. 8.

Depth-WSA-storage relationships of a typical pond.

to compute cumulative annual recharge (module 4) in all the study sites. A comparison of these results (Fig. 10) indicates that the area of the catchment has little influence over the recharge from the pond, as in pond 2 and pond 3 the amount of recharge has increased despite a reduction in the catchment area when compared to pond 1. Consequently, it is
WSA, ha a SV, ha m 3 Slope of WSA 8 SV curves

12.5 _.WSA, SV. ha ha m CrB,


CUWB,

2.5 - -+2-

Sbpe-WSA
SlOpe-SV

ha,,,,
,,a

j
1.5

1.5 -1 l-

0.5 -

- 0.5

____ .+g/__.~ __.__----- < ___-,01 0


0.5

1 Depth ,m

1.5

0 2.5

Fig. 9. Effect of increasing pond depth on WSA and SV.

Watershed, aquifer and crop models of groundwater recharge

17

10

12

Pond1

Pond2

pond3
,

i
14

6 Annual recharge.

8 ha m

10

12

Fig. 10.

Simulated annual groundwater recharge at different pond locations for two years.

inferred that an increase in recharge in pond 2 and pond 3 during both years is attributed either to WSA or to the hydraulic conductivity of the formation (Fig. 10). Since the hydraulic conductivity of the pond area is fixed, any attempt to maximize recharge in a pond system should opt to increase the WSA of the pond. Thus, it is now qualitatively understood that given a location, the maximum WSA is a predominant factor influencing the amount of recharge from a pond. Further, from the preceding discussion it is evident that increase in the WSA is possible in practice by marginally heightening the existing bund (Figs 8 and 9) of the pond. Evaluating capacity options Simulation runs were made with CAP-ASIM to assess the effects of increasing pond capacity. Four hypothetical pond capacities above the existing pond capacity of pond 1 were considered. Thus, CAP-ASIM was run for 10000, 20000, 30000, 40000 and 50000 m3 capacity. Simulated WSA and water table elevation increased substantially for initial

18

M. Selvarajan

et al.

Simulated 6

WSA,

ha

Capacity, -1 4-2

ha

m --3 ----4

:I
3-

2-

t-

0 240

I
260 300 Julian day

Fig. 11. Simulated WSA for different hypothetical SV values for a sample case. in the capacity (Figs 11 and 12), but further increase beyond 30000 m3 yielded a comparatively slower increase in WSA and water table elevation. This is attributed to the following factors: (a) topographic conditions usually do not permit increase of WSA beyond a certain limit; and (b) as the water spreads further, the distance between pond and constant head boundary decreases. Consequently outflow from the recharge system is increased. Spatial distribution of predicted well levels (Fig. 12) indicates that the water table would rise for higher pond storage. Nearer to the constant head boundary (1000 m from the centre of the pond in the present case) the water table converges with the original water table due to discharge effects. Rise in well levels also follows a similar trend as compared to the WSA for different pond capacities. This also corroborates the influence of WSA on recharge. A continuous plot of simulated well levels over a rice-growing season reveals (Fig. 13) that the net increase in recharge realized in the nearer reaches is more than the farther reaches of the command area. Typical simulation results on the cumulative flows to a well at middle reach during a season indicates that the possible augmentation to irrigation above 3 ha m capacity level is limited. increases

Watershed, aquifer and crop models of groundwater recharge

19

Elevation

from

MSL.

Capacity. -1 -2

ha

m --3 ---4

3s.t

36.4 I-

35.: t-

1
300 600 700
Of

100 Fig. 12.

800

Distance

from

Cbntrb

pond, m

Simulated water table elevations for the case in Fig. 11.

Elevation 34.5

from MSL,

Capacity,
-1 -3

ha

m ---5

33.5 -

32.5 -

30.5 ---i 245 255

I
265

I
275

/
285

I
315

1
325 335 345

295 305 Julian day

Fig. 13.

Simulated continuous water table elevations for the irrigation season during 1984, in pond 1 location.

20

hf. Selvurujan et al.

AID-recharge,

mm Capacity, ha m

300 250 -

-1

-2

---3

-----4

.<_ +..

300

400

500

600

700

600

900

Distance from pond centre, m

Fig. 14.

Spatial

distribution

of AID-recharge

for the case.

The distance-capacity

curves

Predicted total quantities of flow during a season, to wells at different radial distances, were converted into total depth of available AID-recharge. A series of curves (Fig. 14) were obtained which depict the position of a well and AID-recharge for varying pond capacities during a full growing season (distance-capacity-recharge curves). These parametric values of AIDrecharge were input to the crop model, for different well irrigation scenarios. Several simulation runs were made to predict the water-limited yields, for various zones in the pond command area. The total flows nearer to the pond centre are almost equal to those at the locations in the middle reach, due to a relatively flat hydraulic gradient at the head reach. A steep decrease in recharged irrigation at the far reach is observed due to (a) spatial distribution of the mound, and (b) outflows through the constant head boundary. The distance-capacity-yield curves in Fig. 15 reveal that the water-limited yields attainable at different reaches of the pond area vary with the changes in the pond capacity. For an initial increase in pond size (i.e. 2 ha m) from the present capacity, a relatively huge increase in the yields is observed, even at farther distances from the pond. For subsequent improvements in the capacity the associated additional yields are small at the nearer reaches of the pond. However, at farther ends productivity is considerably improved for higher pond capacities when compared to nearer reaches.

Watershed, aquifer and crop models of groundwater recharge

21

Simulated yield, kg/ha

3060 -

3020 --__ 2960 --__ -_

300

400

500

600

700

800

900

Distance from centre of pond, m Fii.

15. Simulated water-limited yield potentials and their distribution for the case.

The distance-capacity curves presented in this section pertain to one pond location for a wet season during 1984. Thus, it is clear at this stage that generations of such information over historic and expected weather conditions could be used to compute the direct benefits associated with a modification in the hydrologic design of the pond structure, such as raising the bund and/or increasing the WSA of the pond. The spatial variability of these benefits can also be assessed by means of weighted averages over different zones. Having estimated the total productivity changes and associated costs in a pond command area, the designer would be able to take a decision on the implementation of the hydrologic strategy under consideration.

CONCLUSIONS Artificial groundwater recharge has become a necessity in many tropical areas. Huge investments are made in constructing percolation ponds which are small and convenient recharge structures. In designing and evaluating these ponds several general empirical procedures were being adopted. These descriptive rules limited their use due to the spatial variability of real-world pond systems. In this study a methodology was devised to

22

M. Selvarajan

et al.

evaluate ponds by linking watershed, aquifer and crop system models. This could be more effectively used in place of previously used empirical relationships, as these models incorporate the basic processes of the agroeco system. The performances of the different models were verified and validated using observed and predicted data at various levels. The complex of models thus tested was used to evaluate the performance of the system behaviour. It is inferred from the simulated results that at any given location the WSA plays an important role in influencing the pond recharge. Thus, for a given micro catchment water harvest, ponds over mild slope terrain would yield better results. In such areas, moderately increasing the bund height would result in a steep increase in the WSA. Increasing the pond storage effectively improves the water table position. However, this improvement tends to decline beyond a certain dimension of the pond bund. Local boundary conditions play an important role in deciding the amount of recharge. By linking a crop growth model to the pond model (CAP-ASIM), the productivity responses to different pond storage options were evaluated. The upstream areas of the pond command benefit more from the initial increase in the pond storage. However, for further increases in the pond storage the downstream areas of the command showed higher productivity responses than the upstream areas. Usually, designers are interested in evaluating such trends as the farmers at the downstream locations are generally affected due to poor volumes of recharge. Therefore, it is necessary to evaluate these trends at each pond location, in order to design and execute suitable pond schemes. The combination of models used in the present study would serve as a decision support tool for designing and evaluating ponds.

ACKNOWLEDGEMENTS The authors wish to thank the Indian Agricultural Research Institute, New Delhi, and the SARP project of the International Rice Research Institute, Manila, Philippines for providing fellowship and facilities respectively, for conducting this study. REFERENCES
Blank, D. & Beer, C. E. (1968). Factors affecting water yields from small watersheds in Iowa. Trans. ASAE., 11, 7014. DAE (1983~). Report to accompany the estimate for percolation tank at Sethupattu village of Sriperumpudhur Talu. Department of Agricultural Engineering, T.N., India.

Watershed, aquifer and crop models of groundwater recharge

23

DAE (1983b). Report to accompany an estimate for forming a pond at S.F. No. 204-5 of Perannakkavur village, Uthiramerur Taluk, Chengalpattu District of Tamil Nadu State. Department of Agricultural Engineering, T.N., India. DAE (1983~). Report to accompany the estimate for the construction of percolation pond in Krishnakuppam village of Cuddalore Taluk. Department of Agricultural Engineering, T.N., India. GO1 (1984). Groundwater estimation methodology. Report of the groundwater estimation committee. Ministry of Irrigation, Government of India. Krishnaswamy (1976). Prediction of monthly runoff. Irrigation and Power, 33(4),
503-74.

Marino, M. A. (1975). Mathematical models of artificial recharge systems. Paper No. 2005, Dept. of Water Science and Engineering, University of California, Davis, CA. Michael, A. M. (1978). Irrigation Theory and Practice. Vikas Publishing House, New Delhi, India. Naef, F. (198 1). Can one model rainfall runoff process today? Hydrological Service Bulletin, 26(3), 281-9.

Nightingale, H. I. & Bianchi, W. C. (1973). Groundwater recharge. The Leaky Acres Project. Groundwater, 11(6), 3644. Palanisamy, K. & Flinn, J. C. (1988). Evaluating the performance of tank irrigation systems. Agric. Syst., 28, 161-77. Penman, H. L. (1948). The dependence of transpiration on weather and soil conditions. J. Soil Sci., 1, 74-89. Penning de Vries, F. W. T., Jansen, D. M., ten Berge, H. F. M. & Bakema, A. (1989). Simulation of ecophysiological processes of growth in several annual crops. Simulation monographs, PUDOC, Wageningen, The Netherlands. Prickett, T. A. & Lonnquist, C. G. (1971). Selected digital computer techniques for groundwater resource evaluation. Illinois State Water Survey Division, Urbana, IL. PWD (1982). EWS Study-SMA calculation by Dry-DampWet MethodStrange table in metric units. Communication: PWD (Groundwater), T.N., s India. PWD (1985). Report on the recharge study conducted in the pond at KettureddiPatti village. Groundwater Investigation Division, Dharmapuri, T.N., India. Rabbinge, R. & De Wit, C. T. (1981). Theory of Modelling and Systems Management. Simulation reports, Centre for Agro Biological research, Wageningen, The Netherlands. Ramaprasad & Gangaraju, S. A. (1983). Groundwater recharge through percolation tanks in coastal Karnataka. Case Study Report, Department of Civil Engineering, IISc., Bangalore, India. Rao, A. R. & Minikou, M. (1983). Regional Monthly rainfall runoff model. J.
Water Res. Planning and Mgmt, 109(l), 75-93.

Rao, K. A. & Shanmugasundaram, K. (1985). Integrated planning in water scarcity areas. Unpublished paper, Department of Agricultural Engineering, IIT, Kharagpur, India. Rao, N. H. & Sarma, P. B. S. (1981). Groundwater recharge from rectangular areas. Groundwater, 19(3), 2714. Remson, I. G. M., Hornberger & Molz, F. J. (1971). Numerical Methods in Subsurface Hydrology. John Wiley, New York.

24

h4. Selvarajanet al.

Rushton, K. R. & Redshaw, S. C. (1979). Seepage and Groundwater Flow. John Wiley, New York. Rushton, K. R. & Senarath, D. C. H. (1983). A mathematical model study of an aquifer with significant dewatering. J. Hydrol., 62, 143-58. Ryan, G. J. & Pereira, M. (1978). Deviation of empirical models for the prediction of runoff on small agricultural watersheds in the semi-arid tropics. Paper presented at the International Workshop on the Agro-climatological Needs of Semi Arid Tropics, ICRISAT, Hyderabad, India. Selvarajan, M. (1990). Studies on groundwater recharge through percolation ponds. PhD thesis, Indian Agricultural Research Institute, New Delhi. Selvarajan, M., Palanisamy, K. & Ramasamy, S. (1991). Evaluation tank irrigation systems for future groundwater conditions. Case study report, Third Int. SARP Course, IEEI, Manila. Senarath, D. C. H. & Rushton, K. R. (1984). Groundwater, 22(2), 142-7. Siegel, S. (1956). Non Parametric Statistics. McGraw-Hill, New York. Smith, G. D. (1969). Numerical Solution of Partial Dt#erential Equations. Oxford University Press, London. Sopper, W. E. & Lall, H. W. (1965). Stream flow characteristics of physiographic units in the North East. Water Res., l(l), 115-24. Subbiah, S. & Bhavani, R. (1988). Water resources and crop pattern: an emerging disharmony in the wake of Green Revolution. Zndo-Soviet Seminar on
Rational Utilization of Natural Resources and Regional Development, IQ-12

October, 1988, New Delhi. Thiruvenkatasamy, K. R. & Sakthivadivel, R. (1969). Groundwater recharge due to percolation ponds. Report, Centre for Water Resources, College of Engineering, Anna University, Madras, India. Wickham, T. (1973). Predicting yield benefits in lowland rice through a water balance model. In Water Management in Philippine Irrigation Systems, ed. IRRI, Los Banos, Philippines. WTC (1989). Sixth Trainer S training programme proceedings. Water Tech. Centre, Indian Agricultural Research Institute, New Delhi. Wu, I. P., Delleur, J. W. & Deskin, M. H. (1964). Determination of Peak Discharge and Design Hydrographs for Small Watersheds in Indiana. School of Civil Engineering, Purdue University, Lafayette, IN.

You might also like