New sensing methods for scheduling variable rate irrigation to improve water use efficiency and reduce the environmental footprint : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Soil Science at Massey University, Palmerston North, New Zealand

Abstract

Irrigation is the largest user of allocated freshwater, so conservation of water use should begin with improving the efficiency of crop irrigation. Improved irrigation management is necessary for humid areas such as New Zealand in order to produce greater yields, overcome excessive irrigation and eliminate nitrogen losses due to accelerated leaching and/or denitrification.

The impact of two different climatic regimes (Hawkes Bay, Manawatū) and soils (free and imperfect drainage) on irrigated pea (𝘗𝘪𝘴𝘶𝘮 𝘴𝘢𝘵𝘪𝘷𝘶𝘮., cv ‘Ashton’) and barley (𝘏𝘰𝘳𝘥𝘦𝘶𝘮 𝘷𝘶𝘭𝘨𝘢𝘳𝘦., cv. ‘Carfields CKS1’) production was investigated. These experiments were conducted to determine whether variable-rate irrigation (𝘝𝘙𝘐) was warranted. The results showed that both weather conditions and within-field soil variability had a significant effect on the irrigated pea and barley crops (pea yield - 4.15 and 1.75 t/ha; barley yield - 4.0 and 10.3 t/ha for freely and imperfectly drained soils, respectively).

Given these results, soil spatial variability was characterised at precision scales using proximal sensor survey systems: to inform precision irrigation practice. Apparent soil electrical conductivity (𝘌𝘊ₐ) data were collected by a Dualem-421S electromagnetic (𝘌𝘔) survey, and the data were kriged into a map and modelled to predict 𝘌𝘊ₐ to depth. The 𝘌𝘊ₐ depth models were related to soil moisture (θᵥ), and the intrinsic soil differences. The method was used to guide the placement of soil moisture sensors.

After quantifying precision irrigation management zones using 𝘌𝘔 technology, dynamic irrigation scheduling for a 𝘝𝘙𝘐 system was used to efficiently irrigate a pea crop (𝘗𝘪𝘴𝘶𝘮 𝘴𝘢𝘵𝘪𝘷𝘶𝘮., cv. ‘Massey’) and a French bean crop (𝘗𝘩𝘢𝘴𝘦𝘰𝘭𝘶𝘴 𝘷𝘶𝘭𝘨𝘢𝘳𝘪𝘴., cv. ‘Contender’) over one season at the Manawatū site. The effects of two 𝘝𝘙𝘐 scheduling methods using (i) a soil water balance model and (ii) sensors, were compared. The sensor-based technique irrigated 23–45% less water because the model-based approach overestimated drainage for the slower draining soil. There were no significant crop growth and yield differences between the two approaches, and water use efficiency (𝘞𝘜𝘌) was higher under the scheduling regime based on sensors.

To further investigate the use of sensor-based scheduling, a new method was developed to assess crop height and biomass for pea, bean and barley crops at high field resolution (0.01 m) using ground-based 𝘓𝘪𝘋𝘈𝘙 (Light Detection and Ranging) data. The 𝘓𝘪𝘋𝘈𝘙 multi-temporal, crop height maps can usefully improve crop coefficient estimates in soil water balance models. The results were validated against manually measured plant parameters.

A critical component of soil water balance models, and of major importance for irrigation scheduling, is the estimation of crop evapotranspiration (ETc) which traditionally relies on regional climate data and default crop factors based on the day of planting. Therefore, the potential of a simpler, site-specific method for estimation of ETc using in-field crop sensors was investigated. Crop indices (𝘕𝘋𝘝𝘐, and canopy surface temperature, Tc) together with site-specific climate data were used to estimate daily crop water use at the Manawatū and Hawkes Bay sites (2017-2019). These site-specific estimates of daily crop water use were then used to evaluate a calibrated FAO-56 Penman-Monteith algorithm to estimate ETc from barley, pea and bean crops. The modified ETc–model showed a high linear correlation between measured and modelled daily ETc for barley, pea, and bean crops. This indicates the potential value of in-field crop sensing for estimating site-specific values of ETc.

A model-based, decision support software system (𝘝𝘙𝘐–𝘋𝘚𝘚) that automates irrigation scheduling to variable soils and multiple crops was then tested at both the Manawatū and Hawkes Bay farm sites. The results showed that the virtual climate forecast models used for this study provided an adequate prediction of evapotranspiration but over predicted rainfall. However, when local data was used with the 𝘝𝘙𝘐–𝘋𝘚𝘚 system to simulate results, the soil moisture deficit showed good agreement with weekly neutron probe readings. The use of model system-based irrigation scheduling allowed two-thirds of the irrigation water to be saved for the high available water content (𝘈𝘞𝘊) soil.

During the season 2018 – 2019, the 𝘝𝘙𝘐–𝘋𝘚𝘚 was again used to evaluate the level of available soil water (threshold) at which irrigation should be applied to increase WUE and crop water productivity (WP) for spring wheat (𝘛𝘳𝘪𝘵𝘪𝘤𝘶𝘮 𝘢𝘦𝘴𝘵𝘪𝘷𝘶𝘮 L., cv. ‘Sensas’) on the sandy loam and silt loam soil zones at the Manawatū site. Two irrigation thresholds (40% and 60% 𝘈𝘞𝘊), were investigated in each soil zone along with a rainfed control. Soil water uptake pattern was affected mainly by the soil type rather than irrigation. The soil water uptake decreased with soil depth for the sandy loam whereas water was taken up uniformly from all depths of the silt loam. The 60% 𝘈𝘞𝘊 treatments had greater irrigation water use efficiency (𝘐𝘞𝘜𝘌) than the 40% 𝘈𝘞𝘊 treatments, indicating that irrigation scheduling using a 60% 𝘈𝘞𝘊 trigger could be recommended for this soil-crop scenario.

Overall, in this study, we have developed new sensor-based methods that can support improved spatial irrigation water management. The findings from this study led to a more beneficial use of agricultural water.