MODELLING THE UPTAKE OF SOIL WATER AT DIFFERENTDEPTHS BY TEA (CAMELLIA SINENSIS) PLANTS

Amulti-layer soil water use model to simulate the extraction of soil water hy a tea crop was developed and tested with experimental data from Tanzania. The predictions of moisture contents a t different cleptbs in tllz soil profile a t different dates through the model agreed well with experimental data.


INTRODUCTION
Soil moisture content and its distribution with respect to both time and depth are key factors governing some hydrological processes that have important economic consequences in land and water management. For example, the efficient use of irrigation water, the effective operation of flood warning systems and the accurate assessment of ground water recharge are all dependent upon the availability of reliable soil moisture data, because they incorporate elements which are sensitive to soil moisture status. There is therefore a need for models that can accurately simulate soil moisture content and its distribution for any set of meteorological, soil and plant conditions1.
Improved understanding of any of the processes of the tea crop is important to Sri Lanka, because tea (Camellia sinensis) is one of the major export crops bringing in much needed foreign exchange to the country. Tea plants, particularly some vegetat~vely propagated clones are prone to drought which occurs annually specially under low country conditions. As a resu1.t there have been reduced crop yields and in some cases the failure of the crop altogether prompting the authorities to think of irrigating the tea crop as in some foreign countries. Therefore, a study on understanding the extraction of soil water at different depths by a tea crop was ~mdertaken. This paper describes the development of a multi-layer soil water use model (capable of predicting water use a t different depths in the soil profile by a tea crop) that is tested with field data from Tanzania.

Models to simulate soil water u p t a k e in a Tea Crop
Although water use/extraction of many of the plants have been modelled, not much work has been done on the water uselextraction by a tea crop. Willat" CoopeP and Stephens & Carr4 have modelled the water use of tea which gives the amount of water use by the whole plant rather than the extraction pattern of the root system. Willat2 working in Malawi, using gravimetric sampling, found a hysteretic relation between the actual (ETa) and potential (ETp) evapotranspiration (0.85 -0.90) and a critical soil moisture deficit (200 mm), above which t h e ratio ETa/ETp fell sharply. CooperVound this ratio to be 0.56 and little effect of water stress on the transpiration rate from his work in Kenya. Stephens & Carr4, from their work in Tanzania have come up with a single layer water use model, which predicts the water use very well. Also they have identified the critical soil water deficit for clone '618' as 60 mm.
From these few instances of modelling the water use of tea, it is clear that simulating the water use of tea has been studied only superficially up to now. Further studies on simulating the water use by a tea crop are therefore warranted.

METHODS AND lVWTERIALS
The location chosen to develop and test the model was Ngwazi (8"33'S, 35"10'E, altitude 1840 m) in Tanzania. The annual rainfall ranges from 800 to 1100 mm, with most rain falling between mid November and May. Pan evaporation rates range from approximately 3 mm/d from June to 5 mm/d in October. The soil a t Ngwazi is intensely weathered, well drained & acidic with a low base status typical of the grasslands in the area. In general the soils up to 5.5 m (maximum rooting depth of tea) can be classified as clayey. A detailed description of the climate and soil a t Ngwazi is given by Stephens & Carr.4 The model: The model is based on the approach used 5;. Carr et aL5 for predicting the water use by potatoes and sugarbeet with improvements a n d suitable modifications for the tea plant. These improvements and modifications are described and discussed later in this paper. A brief description of the model developed is given in the Appendix. A detailed description (and a flowchart) is found in de Silva." The calculations to determine the volumetric soil water content a t a particular depth in the soil profile a t a given day (or indirectly, water extraction by plant roots) is given below.  With the proposed method, where a small quantity is extracted from the layer (with highest SWP), this problem does not arise, because after extracting this very small quantity, SSWP of each layer is recalculated to identify the layer with highest SWP.

2.
Also, the transpiration restriction functions Fw and Fr are more easily calculated with the proposed method; since they (especially Fw) change with the changing moisture content in the profile.
Testirrg o f th,e model: A computer program written in Fortran 77 was used to simulate the daily water contents down the root zone. profile, with the modified method outlined in the previous section. Performance of the model was tested with data from work carried out by Stephens and Carr4 in Ngwazi Tea Research Unit, in the Mufindi District of Tanzania.
The maximum rooting depth used was 5.0 m and the rate of root growth was assumed to be zero, since the testing was done for mature, fully grown tea. The root zone was divided into 25 layers with a layer thickness of 200 mm each; The ratio between actual soil evaporation and potential soil evaporation of each layer was talcen as 1.0, 0.5, 0.1 for the top three layers and zero for the other layers7 The root zone consisted of five different soil horizons. Field capacity, permanent wilting point and moisture content a t -200 kPa CB, ) for each soil horizon is shown in Table 1. An irrigation treatment (10) was used for testing of the model. Fig. 2 shows the rainfall, pan evaporation and the irrigation (10) during the period of testing he., from 1 April 1989 to 1 December 1989).

3.
The redistribution of water once an amount is extracted from a soil layer. Redistribution does not seem to affect the total soil water extracted for the season (since the measured & calculated SWD tally very well), but could have an effect when it comes to the extraction from the individual layers. However, this is not very likely since extraction patterns agree on some other days like 3 May.
The results for 10 November agrees quite well with measured values except for depths between 0.4 m and 2.0m. This could well be due to the values specified for permanent wilting point (higher values were used in the model) apart from the reasons mentioned above.

DISCUSSION
The model predicted moisture contents on G different days are sl~own in Fig. 5. It, is seen that initially the tea crop seem to extract water from the sudace layers of the soil profile (as there is no change in moisture contents a t depths below 2.5 m on 10"' May). However, when the soil profile dries further, it is seen that the tea crop has extracted water from all layers up to the depth of 5.0 m, which is the maximum root depth. This finding will have important implications in irrigating a tea crop and also in the application of fertiliser to a tea crop (2s nutrients are likely to be absorbed with soil water).
In the approach used in this model, a number of simplifying assumptions were made, which have to be reconsidered and rectified if the accuracy of the predictions are to be very high. Instantaneous water movement down the profile was assumed in calculating drainage through the layers. For the conditions tested, this is unlilrely to cause a significant error, because of the small amounts of water inputs (i.e., rain $ irrigation). However, for a season where a large quantity of water is applied to the root zone, predictions could be inaccurate. Therefore, it is recommended that the rate of water movement through the layers be considered and built into the model for i t to be more accurate.
Redistribution of soil water a t different layers in the soil profile was assumed to be negligible. For improved accuracy a mecl~anism must be built in to the model which carries out the redistribution. Also the upward movement of soil water from depths greater than 5.0 m (as the soil layers above this depth dries out) was assumed negligible. This assumption can also be treated as redistribution of soil water and must be talren into consideration for improved accuracy. Some work also has t,o be done in the area where soil evaporatioil was calculated. The approach used would depend largely on the soil condition and the climate of the testing site. If this method is to be used for a particular site, the ratio between actual and potential evaporation has to be determined (also considering the layer thickness) for each layer. However, since the amount, of. evaporation from substrate in a crop like tea is very small, an inaccuracy in the soil evaporation will not be reflected in the predictions significantly.
The amount of evaporation due to interception was calculat,ed according to the information available as at present. It is felt that there is room for improvement, here as well, although the figures used seem to give an accurate prediction over the dry season when there was little or no rain.
The conclusions of this study are; 1. It is possible to successfully model the water use by a tea crop using a simple, multi-layer water balance model.

2.
The model developed performs well in predicting the soil water use a t different depths in a soil profile by a tea crop.

APPENDIX
The evaporative demand of the atmosphere (potential evapotranspiration ETp) can be cor~sidered to be satisfied by the crop evapotranspiration (Ecrop) and by the substrate evaporation (Esub) as in equation (1).

ETp= Ecrop + Esub
The potential evpoatranspiration (ETp) is also given by where Kc = crop factor and ETo = reference crop evapotranspiration In order to estimate substrate evaporation (Esub) and crop evaporation (Ecroy) lndependently, it is necessary to partition the available energy for evaporation between these two sources. This is done as in Aslyng

A.l Substrate Evaporatior~ (Esub)
Water loss by evaporatjon from the substrate or the soil surface is derived from two sources.

A. 1.1 Evaporation due to Te,-?tporary Storage (Efl
This is the evaporation of water which is temporarily stored on the surface of the soil or moving down through the larger pores of coarse textured soils. This is assumed only to occur within 24 hours of a rainfall or an irrigation event which was greater than the existing soil water deficit, and before free drainage has ceased.
To calculate evaporation due to temporary storage, a temporary storage zone (TSZ) is defined as the depth of water which moves down the profile in a day. This can be considered as equal to the vertjcal saturatscl hydraulic conductivity of the soil layer. The depth of water (Wfl temporarily stored in the temporary storage zone is determ~ned as follows. Where E*sub = potential evaporation rate of the substrate. E's = potential evaporation rate of the soil d = drainage

A.l.2 Evaporation Due To Soil Water (Es)
Here the root zone is divided into a number of layers and evaporation from each layer js talren as a factor of the potentia1 soil water evaporation as shown in Table  A.l.7 These factors depend on the required layer thickness, soil texture, soil structure etc. Water lost by evaporation from the crop (Ecrop) is derived from two sources.

A.2.1 Evaporatio~z due to Intercepted Water (Ew)
Thls IS the rarn or irrigation water which is intercepted by and wets the crop canopy.
Sqwre & Callander" reports a maximum canopy storage of 2 mm and an average of only 0.8 mm. Further, they suggest a tentative estimate of canopy storage of 1 mm for 'Assarn' type tea in the absence of a rellable value. However, since the variation of amount intercepted with rainfall is given by Cooper', his figures are used in this model and the potentjal evaporation due to intercepted water (E"w) is determined as follows.
. .  Where E*t is the potential transpiration.

A.2.2 Transpirntion (Et)
This is water which diffuses through the stomata1 pores as vapour into the surrounding air and wjll be equal to its potential value under certain ideal conditions only. At all other times it is less than the potential value. The point a t which the actual transpiration falls below the potential rate is not fixed and i t depends on many variables including soil hydraulic characteristics, rooting density and the potential evaporation rate.1° " Thus the transpiration is calculated as follows. Where 0, : is the critical soil water content a t which transpiration falls below its potenti a1 value.

Et
Available water function js determined as given below.