Predicting the transpiration of lucerne

Abstract

Crop transpiration rate is often predicted as the minimum of the supply of water from the roots and the demand for water from the shoots. A common method for predicting transpiration demand is the transpiration efficiency approach, which uses daily dry matter growth, constant transpiration efficiency coefficient (ω ) and the vapour pressure gradient from the leaf to the atmosphere. The former is usually represented by daytime averaged atmospheric vapour pressure deficit (VPD′z), and assumes that the difference between air temperature and leaf temperature is small. In this paper we show that this assumption is likely to be incorrect in cool, moist environments where unstressed leaves may become substantially warmer than the surrounding air. We used VPD′z to calculate an apparent ω for field grown lucerne and found it increased from 1.5 to 3.5 Pa across a mean daily temperature range of 7–18 °C. We present an alternative method for calculating transpiration demand which we call the canopy conductance approach, which uses the product of VPD′z, intercepted radiation (Rᵢ) and a canopy conductance coefficient (ΘT). Experimental data showed transpiration demand of lucerne was linearly related to Rᵢ and VPD′z and, in contrast to the variability of the derived value of ω, a constant and stable ΘT of 0.45 mm MJ⁻¹ kPa⁻¹ was derived. Predictions of transpiration were compared with transpiration calculated from soil water balance measurements for dryland and irrigated lucerne crops grown over 3 years at Lincoln University, Canterbury, New Zealand. The canopy conductance approach accurately (±0.05 mmd⁻¹, RMSD = 26% of observation mean) predicted transpiration from an independent data set, confirming its simplicity and efficacy. Calculations of transpiration using the transpiration efficiency approach with a constant ω of 5.0 Pa underestimated transpiration by 1.4 mm d⁻¹ (RMSD = 126%). However, using ω and changing it in proportion to radiation use efficiency (which changed in response to temperature) substantially improved the accuracy of the growth approach (±0.10 mm d⁻¹, RMSD = 30%). The methods presented can be adopted to improve the generic applicability of transpiration predictions in crop simulation models.