A well-managed irrigation scheduling system needs a rapid, preCIse, simple, costeffective and non-destructive soil water content sensor. The PRl profile probe and Diviner 2000 were used to determine the timing and amount of irrigation of three cover crops (Avena sativa L., Secale cereale L. and Lolium multiflonlm Lam.), which were planted at Cedara, KwaZulu-Natal. The PRl profile probe was first calibrated in the field and also compared with the Diviner 2000. For the calibration of the PRl profile probe the factory-supplied parameters (aJ = 8.4 and ao = 1.6) showed good correlation· compared to the soil-estimated parameters (aJ = 11.04 and ao = 1.02). The factorysupplied parameters gave a linear regression coefficient (r2 ) of 0.822 and root mean square error (RMSE) of 0.062. The soil-estimated parameter showed a linear regression coefficient of 0.820 with RMSE of 0.085. The comparison between the soil water content measured using the PR1 profile probe and Diviner 2000 showed a linear regression coefficient of 0.947 to 0.964 with a range of RMSE of 0.070 to 0.109 respectively for the first 100 to 300 mm soil depths. The deeper depths (400, 600 and 1000 mm) showed a linear regression coefficient ofO.716to 0.810 with a range of 0.058 to 0.150 RMSE. These differences between the shallow and deeper depths could be due to soil variability or lack of good contact between the access tube and the surrounding soil. To undertake irrigation scheduling using the PRl profile probe and Diviner 2000, the soil water content limits were determined using field, laboratory and regression equations. The field method was done by measuring simultaneously the soil water content using the PR1 profile probe and soil water potential using a Watermark sensor and tensiometers at three depths (100, 300 and 600 mm) from a 1 m2 bare plot, while the soil dries after being completely saturated. The retentivity function was developed from these measurements and the drained upper limit was estimated to be 0.355 m3 m-3 when the drainage from the pre-wetted surface was negligible. The lower limit was calculated at -1500 kPa and it was estimated to be 0.316 m3m,3. The available soil water content, which is the difference between the upper and lower limit, was equal to 0.039 m3 m,3. In the laboratory the soil water content and matric potential were measured from the undisturbed soil samples taken from the edge of the 1 m2 bare plot before the sensors were installed. Undisturbed soil samples were taken using a core sampler from 100 to 1000 mm soil depth in three replications in 100 mm increments. These undisturbed soil samples were saturated and subjected to different matric potentials between -1 to -1500 kPa. In the laboratory, the pressure was increased after the cores attained equilibrium and weighed before being subjecting to the next matric potential. The retentivity function was then developed from these measurements. The laboratory method moved the drained upper limit to be 0.390 m3 m,3 at -33 kPa and the lower limit be 0.312 m3m-3 at -1500 kPa. The regression equation, which uses the bulk density, clay and silt percentage to calculate the soil water content at a given soil water potential, estimated the drained upper limit to be 0.295 m3m-3at -33 kPa and the lower limit 0.210 m3 m,3 at -1500 kPa. Comparison was made between the three methods using the soil water content measured at the same soil water potential. The fieldmeasured soil water content was not statistically the same with the laboratory and estimated soil water content. This was shown from the paired-t test, where the probability level (P) for the laboratory and estimated methods were 0.011 and 0.0005 respectively at 95 % level of significance. However, it showed a linear regression coefficient of 0.975 with RMSE of 0.064 when the field method was compared with the laboratory method. The field method showed a linear regression coefficient of 0.995 with RMSE of 0.035 when compared with the estimated method. The timing and amount of irrigation was determined using the PR1 profile probe and Diviner 2000. The laboratory measured retentivity function was used to define the fill (0.39 m3 m-3 ) and high refill point (0.34 m3 m-3 ). The soil water content was measured using both sensors two to three times per week starting from May 29 (149 day of year, 2002) 50 days after planting until September 20 (263 day of year) 11 days before harvesting. There were five irrigations and twenty rainfall events. The next date of irrigation was predicted graphically using, the PRl profile probe measurements, to be on 3 September (246 day of year) after the last rainfall event on 29 August (241 day of year) with 8 mm. When the Diviner 2000 was used, it predicted two days after the PRl profile probe predicted date. This difference appeared since the Diviner 2000-measured soil water content at the rooting depth was slightly higher than the PRl profile probe measurements. The amount of irrigation was estimated using two comparable methods (graphic and mathematical method). The amount of irrigation that should have been applied on 20, September (263 day of year) to bring the soil water content to field capacity was estimated to be 4.5 hand 23 mm graphically and 5.23 hand 20 mm mathematically. The difference between these two methods was caused due to the error encountered while plotting the correct line to represent the average variation in soil water content and cumulative irrigation as a function of time. More research is needed to find the cause for the very low soil water content measurements of the PRI profile probe at some depths. The research should be focused on the factors, which could affect the measurement of the PRl profile probe and Diviner 2000 like salinity, temperature, bulk density and electrical conductivity. Further research is also needed to extend the non-linear relationship between the electrical resistance of the sensor and soil water potential up to -200 kPa. This non-linear equation of the Watermark is only applicable within the range of soil water potential between -10 and -100 kPa. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 2003.

Wheat irrigations were scheduled using two computer models and an infrared thermometer using two critical threshold values. Yields from these plots were compared with plots scheduled by the farm manager. The highest yield was obtained by the computer model using evapotranspiration data taken from the local AZMET station; the plots scheduled by the infrared thermometer (using the low threshold value) ranked a close second.