L. Cavazza, Professor General Agronomy and A. Patruno, Associate Professor of Soil Physics, University of Bologna; P. Rossi, Professor of General Agronomy, University of Turin; M.O. De Seneen, candidate for Doctorate at the University of Bologna; and M, Gammino, Technician at the Sin. Ofanto Prof. Administration, Italy 
SUMMARYIn an irrigated area of southern Italy (Sinistra Ofanto, Foggia) an irrigation scheduling model was compared with the personal decisions of the farmers in water application during a twoyear period. The scheduling model was based on daily measurements of rainfall and evaporation from a class A pan, as well as parameters accounting for soil and climate conditions, crop (species, variety, and growth stage), and agronomic practices. The water applied by the farmers and the yields obtained were recorded.
The results were disappointing for trees (olives, peaches, grapevine, etc.) but very satisfactory for herbaceous crops. For tomatoes, the maximum yield was obtained by the farmers who applied a total water depth similar to the one recommended. Considering the number of applications, a much more complex relationship was obtained. This crop is very sensitive to the number of irrigations when the optimum seasonal water depth is applied. In this case, a greater number of applications could be more productive. This conclusion meets with the wellknown advantage of daily drip irrigation of tomatoes compared to larger irrigation intervals.
Many criteria have been proposed for scheduling irrigation. They are still being improved to meet the requirements of operating conditions. In areas where irrigation is traditionally practised, a comparison between the recommended technique and the irrigation empirically applied by the farmer may be appropriate. This would be useful for corroborating the superiority of the recommended way of irrigation compared to the farmer's practice and for providing indications for further improvement of the scheduling procedure.
This paper uses the observations from a survey of irrigation scheduling experiences carried out in the Apulia region (southern Italy).
METHODS
The data were collected in the irrigation project of 'Sinistra Ofanto', which covers approximately 38 000 ha in the province of Foggia (southern Italy). The area is characterized by a Mediterranean type climate with a mean annual temperature of 16.2°C and rainfall of 591 mm/year. In July the mean temperature is 32.6°C with only 24.9 mm of rainfall. The soils are fertile and usually ploughed to a depth of 35 cm for annual crops and 15 cm for trees. The texture of the ploughed layer varies from sandy loam to silty loam. Inceptisols are common as well as soils with hardpans overlaying calcareous friable sandstone which were ripped a few decades ago.
TABLE 1  Technical data for the crops under observation
Crop 
No. of fields under observation 
Total no. of seasons observed 
Irrigation system 
Sprinklers or drippers 

Spacing (m × m) 
No. of drippers per tree 
Discharge rate of distributors 

Tomato 
8 
16 
drip 
0.6 × 1.8^{1} 

2 l h^{1} 
Pepper 
1 
1 
drip 
0.6 × 0.9 
 
2 l h^{1} 
Water melon 
2 
1 
sprinkler 
15 × 15 

1 l s^{1} 


traveller spr. 
^{2} 



Sugar beet 
1 
2 
traveller spr. 
^{3} 
 
10 l s^{1} 
Asparagus 
1 
2 
sprinkler 
12 × 12 
 
0.75 l^{1} 
Grapevine 
3 
6 
drip 
2.5 × 2.5 
1 
4 l h^{1} 


calibr. orifices 
2.5 × 2.5 
1 
1.2 l mm^{1} 



traveller spr. 
^{2} 
 
1 l s^{1} 

Kiwi tree 
1 
2 
mist irrig. 
2 × 4 
2 
50 l h^{1} 
Peach tree 
3 
6 
drip 
3 × 6 
2 
8 l h^{1} 


drip 
 
3 
2 l h^{1} 



subsurface irr. 
1 × 6 
2 
4 l h^{1} 

Apricot tree 
1 
2 
subsurface irr. 
1 × 6 
2 
4 l h^{1} 
Olive tree 
3 
6 
drip 
3.5 × 7 
2 
8 l h^{1} 


drip 
3 × 5 
2 
4 l h^{1} 



calibr. orifices 
 
2 
60 1 h^{1} 
^{1} In one case 0.4 x 1.8.
^{2} Irrigating 1 ha in 1 day.
^{3} Irrigating 4 ha in 1 day; speed 15 m h^{1}.
The irrigation water is of good quality. It is drawn from the Ofanto River and delivered on demand through a pipe system at about 0.3 MPa and a 10 l s^{1} discharge rate. A number of farmers have their own wells. Two irrigation systems are commonly used in the area: sprinklers and drip irrigation; drip irrigation, as a rule, is not applied daily because of organizational reasons and, possibly, tradition.
For a few years an extension service has been operated by the 'Consorzio di Bonifica per la Capitanata' to help farmers schedule their irrigation. The scheduling used the 'Renana' model described by Giannerini (1993), available at the 'Consorzio della Bonifica Renana' (Bologna). It evaluates the theoretical need for time and depth of irrigation based on class A pan evaporation, rainfall readings, water table depth, soil parameters, crop species, and growth stage.
Corrections for the class A pan readings and crop coefficients are taken from FAO Irrigation and Drainage Paper 24 (Doorenbos and Pruitt, 1977). Some corrections (deficit coefficient, K_{D}) are introduced to reduce the water application depth in order to account for the effect on the quality of the fruit according to Jensen (1970) as modified by Mannini (1993). Other corrections are introduced on the basis of local experimental or empirical observations. A comparison between the 'Renana' and the 'Duett' model was carried out on peach trees by Vitali et al. (1995).
Eleven meteorological stations are distributed throughout the irrigated area. The farmers and the advisory centre are connected via the videotel net, which is also used for information other than irrigation.
In 1990 and 1991, a set of observations was established in order to evaluate the usefulness of this advisory service for irrigation. Nine farms were selected in the area. On each farm a number of fields were grown with different crops (altogether, ten crops in the area). A total of 44 combinations were observed. The farmers were advised on the calculated need for irrigation to each field, but they were left free to apply it at any chosen time and with any watering depth. The crop of each selected field was observed by technicians throughout the growing cycle twice a week during the irrigation season. Agricultural practices, water application (times and depths chosen by the farmer as read on a discharge meter at the hydrant), growing stage of the crop, and final yield were recorded for each field under the supervision of the technician responsible. Meteorological data were obtained from the nearest station. Data for each crop are given in Table 1.
At the end of each cropping season, the total seasonal water depth and the number of irrigations applied to each field by the farmer were compared with the recommended ones and the relationship with the yields was determined.
The information available for tomatoes (8 farmers × 2 years) irrigated by drip irrigation was sufficient to perform a regression analysis. The deviation of farmers' practice from the recommended one was seen as more convenient. Namely, the variables were the deviation in seasonal water depth (D V_{A}) and the deviation in number of water applications (D n). For the whole set of olive trees, peach trees, and grapevine a larger number of seasonal data were available, thus allowing for an analysis of variance (ANOVA), assuming a split block design with year as a random variable crossed with the combination of crops × scheduling, crops and scheduling being fixed effects.
RESULTS
The difference between recommended and applied seasonal water depth is often very disappointing for trees, especially in the second year (Figure 1). By contrast, good agreement was generally found for vegetable crops especially in the first year (Figure 1).
Considering the three main crops, olive trees, peach trees, and grapevine (Table 2), the ANOVA shows that most of the variability is accounted for by the interaction of year x difference between advised and applied seasonal water depth (significant at P 0.01). The mean water depth applied to these crops is 76.8% of the recommended depth in the first year and 27.6% in the second year. It is not possible to deduce from these data which of the two criteria is the best one. Apparently, there are farmers who apply more water for all crops and years, and others who apply lower irrigation depths in all cases.
Considering now tomatoes irrigated by drip irrigation, when the yields of tomato were plotted against the deviations from recommended water depths, a significant curvilinear regression (R = 0.759) was obtained (Figure 2). This shows a decline in tomato yield when too little or too much water was applied as compared to the recommended seasonal depth. The maximum of the regression curve corresponds to about 64 mm. Given that the mean recommended water depth was 411 mm, the latter result would suggest that the calculated water requirement is overestimated by about 15%.
When the deviation from the number of water applications recommended was considered, no significant regression of yield was found nor was there any correlation between seasonal water depth and number of applications.
A multiple regression of yield (P_{A}), as a function of both deviations of seasonal water depth and deviation of number of applications, gave a satisfactory relationship when the following type of equation was applied:
P_{A} = b_{0} + b_{1}D V_{A} + b_{2} (D V_{A})^{2} + b_{3} (D n)^{2}  b_{4} (D V_{A})(D n)^{2}, (1)
where P_{A} = yield; b_{0} = yield estimated when recommended seasonal water depth and number of waterings were applied; the bs are partial regression coefficients and D V_{A} and D n as stated above. The regression equation is:
P_{A} = 67.8  0.1476 D V_{A} + 0.000838 (D V_{A})^{2} + 0.7198 (D n)^{2}  0.0000791 (D V_{A})^{2} (D n)^{2}, (2)
significant at P 0.0006, with R = 0.900; P_{A} in t/ha, D V_{A} in mm. This equation is plotted in Figure 3. The good fit obtained is shown in Figure 4 and explains most of the residual variation from the multiple regression equation (2). The complex response surface of Figure 3 is better understood after examining some selected sections of this solid representation. In Figure 5 it appears that when the frequency of applications was close to the recommended one (3 < D n < +3) the effect of changing the seasonal water depth was relatively small, although in favour of reduced depths (negative D V_{A} values). When, however, the number of waterings exceeds the recommended one by more than three (curve at D n = +10), the yield response curve becomes highly responsive to the irrigation depth and shows a maximum yield exactly associated to the recommended irrigation depth. More precisely, at the recommended irrigation depth (D V_{A} = 0), higher yields could be achieved by increasing the number of waterings (e.g.. D n = +10) and reducing single watering depths, which is preferable with dripirrigation. The observed range of D n (see the plane D n, D V in Figure 3) does not extend far enough to permit reliable interpolations of D n in the negative range; Eq. 2 suggests, however, that higher yields could also be obtained by reducing the number of waterings (e.g., D n = 10) and, correspondingly, applying larger watering depths (this would be advisable when sprinkling or with furrow irrigation). At about 100 mm above or below the advised irrigation depth (D V_{A} = ±100), the yield is in any case lower and appears insensitive to the number of waterings.
TABLE 2  Recommended versus applied water depth (mm) for some tree crops in two years. Means of three completely randomized replications.

1990 
1991 

Advised 
Applied 
App./Adv. % 
Advised 
Applied 
App./Adv. % 

Olive tree 
372.0 
274.0 
73.7 
367.0 
97.7 
26.6 
Peach tree 
447.3 
393.3 
87.9 
371.3 
70.0 
18.9 
Grapevine 
429.0 
291.3 
67.9 
390.7 
144.0 
36.9 







Means 
416.3 
319.6 
76.8 
376.3 
103.9 
27.6 
The interaction scheduling × year is significant at P 0.01.
The effect of fractionating the water depth is shown in Figure 5. When the seasonal water depths are far from the recommended ones (e.g., D V_{A} = ±95.4), the effect of the number of waterings is negligible (horizontal lines), whereas at the recommended seasonal water depth, (D V_{A} = 0), a large increase in the number of waterings appears to be of relevance to increasing the yield. This last statement agrees with what is expected from drip irrigation on tomato. An increase in both seasonal water depth and number of waterings is detrimental (curve at D V_{A} = +110 in Figure 5).
FIGURE 2  Yield of tomatoes as a function of the deviation from the recommended seasonal water depth. R = 0.759. No significant difference between years.
CONCLUSIONS
The practical application of irrigation scheduling in an area of southern Italy was disappointing for trees, while it was encouraging for herbaceous crops. The response of tomatoes was satisfactory but complex. Results for the range of the tested variables suggest that the recommended seasonal water depth was only slightly overestimated, while the number of applications could be increased substantially to take full advantage of drip irrigation; otherwise, the number should possibly be reduced and the watering depth increased, by shifting to other irrigation systems (e.g., furrow irrigation or sprinkling). Appreciable deviations from the optimum seasonal irrigation depth reduce the yield and seem to suppress the role of the number of applications.
FIGURE 3  Response surface of the yield of tomatoes (P_{A}; t/ha) to deviations from both the recommended seasonal water depth (DV_{A}; mm) and the recommended number of waterings (Dn). The solid points on the plane DV_{A}  Dn indicate the observed combinations of these variables.
FIGURE 4  Observed yield of tomato as a function of the expectation on the basis of equation 2
FIGURE 5  Left: Effect on the yield of tomato, of deviations from the recommended water depth (D V_{A} in mm), at selected deviations from the recommended frequency of irrigation (D n). Right: Effect of deviations from the recommended frequency of irrigation (D n), at selected deviations from the recommended seasonal water depth (D V_{A} in mm).
More research is needed to understand the irrigation requirement for tree crops because much less than the recommended water depth is applied by the farmer.
REFERENCES
Doorenbos, J. and Pruitt, W.O. 1977. Crop water requirements. FAO Irrigation and Drainage Paper 24, revised 1977. FAO, Rome.
Giannerini, G. 1993. Renana model, a model for irrigation scheduling employed on a large scale. Proc. 2nd Workshop on Crop Water Models. XV Cong. ICID, The Hague.
Jensen, M.C. 1970. Irrigation system design capacity. Washington Agr. Exp. Sta. circ. 525.
Mannini, P. and Zinoni, F. 1993. Possibilità di applicazione dello stress idrico controllato sul pesco in EmiliaRomagna. Rivista di Frutticoltura, no. 4, 7378.
Vitali, G., Testi, L., and Rossi Pisa, P. 1995. Confronto tra due modelli di simulazione per lo studio delle condizioni idriche finalizzata all'irrigazione del pesco. Irrigaz. e drenaggio 42 (3): 1119.