Elsevier

Agricultural Water Management

Volume 181, February 2017, Pages 94-107
Agricultural Water Management

Estimation of transpiration fluxes from rainfed and irrigated sugarcane in South Africa using a canopy resistance and crop coefficient model

https://doi.org/10.1016/j.agwat.2016.11.024Get rights and content

Highlights

  • A Jarvis-Stewart model calibrated for sugarcane can be applied for the prediction of hourly and daily transpiration.

  • Canopy resistance for optimal conditions is constant due to the increase of minimum stomatal resistance with LAI.

  • An analytical function was derived to express the response of sugarcane transpiration to soil water conditions.

  • The Jarvis-Stewart model outperformed the classical FAO56 type crop coefficients when compared to field measurements.

Abstract

The area under sugarcane is rapidly growing worldwide. The consequences of such growth on basin scale water consumption and competing water resources need to be understood. Conventional models for sugarcane evapotranspiration have shown limitations for different environmental conditions. To improve current estimations of sugarcane water consumption, hourly and daily transpiration of rainfed sugarcane in Kwazulu-Natal (South Africa) and daily transpiration for irrigated sugarcane in Mpumalanga (South Africa) were calculated by using the Penman-Monteith equation (TPM) with a variable canopy resistance. Canopy resistance was calculated with the Jarvis-Stewart model from calibrated environmental stress functions. The classic FAO56 crop coefficient approach (TFAO56) was also investigated and crop coefficient values, crop basal coefficient and water stress coefficient were derived. There were differences between derived crop coefficient values and FAO56 weather-adjusted values. Derived crop basal coefficient (Kcb) was 0.9 for rainfed and irrigated sugarcane, which was lower than FAO56 weather-adjusted values of 1.19 for rainfed and 1.15 for irrigated at mid-stage. The reduction of the crop basal coefficient with the water stress coefficient resulted in an underestimation of transpiration for rainfed sugarcane. This indicates that water uptake under stress conditions is a complex process, not easy to model as water can be extracted from considerable depths. Daily estimates obtained from TPM outperformed those obtained from TFAO56 when compared to Bowen ratio and Surface Renewal system field measurements. For rainfed sugarcane with water-stressed conditions the TFAO56 RMSE was 1.55 mm day−1 compared to 0.3 mm day−1 for TPM. For rainfed sugarcane with water-unstressed conditions the TFAO56 RMSE was 0.5 mm day−1 and the TPM RMSE was 0.22 mm day−1 for TPM. For irrigated sugarcane the TPM RMSE of 0.47 mm day−1 was slightly lower than the TPM RMSE of 0.49 mm day−1, and TPM showed better correlation with an R2 of 0.85 compared to an R2 of 0.64 for TFAO56. This suggests that calibrated variables of the Jarvis-Stewart model for sugarcane proved to be suitable for both rainfed and irrigated sugarcane in South Africa. More research is needed to verify the validity of the calibrated stressed functions in other regions with high intensity of sugarcane plantations.

Introduction

Sugarcane is grown in tropic and sub-tropic zones, where water is often scarce. Due to increasing ethanol and sugar demands, the area under sugarcane is rapidly growing worldwide (Rudorff et al., 2010). Sugarcane water consumption is relatively high. Large scale increases in sugarcane farming compared to other crops may increase overall catchment evapotranspiration (ET) and reduce streamflow. Efficient irrigation systems and good on-farm water management practices are crucial elements to optimize its water consumption. The extent of commercial rainfed and irrigated sugarcane systems in South Africa provides an opportunity to evaluate current methods for computing sugarcane ET(ETc) over a wide range of conditions, and to derive a general model for this purpose. A valuable data set for evaluation of existing ETc methodologies was derived from two studies in commercial rainfed (Jarmain and Everson, 2002) and irrigated (Jarmain et al., 2014) sugarcane fields in South Africa.

ETc can be measured by means of a Bowen ratio system (e.g. Grantz and Meinzer, 1991, Inman-Bamber and McGlinchey, 2003, Jarmain and Everson, 2002), eddy covariance (e.g. Cabral et al., 2012, Pakoktom et al., 2013), a Surface Renewal system (Jarmain et al., 2014), weighing lysimeters (Olivier and Singels, 2012) or other systems. Although these methods are inappropriate for routine measurements, they are used for evaluation of indirect ETc estimates. For example, Inman-Bamber and McGlinchey (2003) compared ETc measurements from a Bowen ratio system with estimates by the direct Penman Monteith (PM) equation using a constant value for sugarcane surface resistance and fixed crop height for deriving aerodynamic resistances. They indicated that prediction errors of ETc in a version of the direct PM equation could be rectified by changing the surface resistance. They compared their ETc measurements with grass reference ET(ET0) estimates derived from the standardized PM equation by the United Nations Food and Agriculture Organization (FAO56) and the combination of crop coefficient (Kc) for sugarcane as proposed by Allen et al. (1998). They confirmed the Kc values of 0.4 at the initial stage and 1.25 at mid-stage in FAO56 to be correct. The value of 0.7 for the end stage was not supported but they suggested a value of 1.25 for an adequate water supply throughout crop development. Olivier and Singels (2012) evaluated the effect of two crop residue layers and no residue cover on ETc, measured with weighing lysimeters in South Africa. They calculated Kc values from ETc measurements and ET0 estimates. Averaged Kc values for the initial stage were 0.31 for no residue cover (Bare), 0.25 for soil covered by a light layer of cane tops (Tops) and 0.18 for soil covered by a heavy layer of tops and dead leaves (Trash). These values were lower than the Kc value of 0.4 proposed in FAO56. Averaged Kc values for mid-stage were 1.12 for Bare, 1.01 for Tops and 1.13 for Trash, which were lower than the FAO56 Kc value of 1.25. Averaged Kc values for the end stage were 1.1 for Bare, 0.8 for Tops and 0.78 for Trash, which were higher than the Kc value of 0.7 proposed in FAO56. Thus, in the absence of locally measured Kc coefficients, generic values for Kc are often used in hydrological studies and water management (ignoring the specific environmental conditions to which the generic Kc applies).

Estimation of ETc may be improved by decoupling the transpiration component (T) from the soil evaporation (E) component, as transpiration is disconnected from the soil physical conditions related to E. Allen et al. (1998) in their standard FAO56 approach also provide a procedure based on a two-layer crop coefficient model, which incorporates a crop basal coefficient (Kcb) and a coefficient for soil evaporation (Ke). This method is limited to well-watered conditions, and measurement of T for crops under water stress conditions (including rainfed) are likely to be overestimated. Allen et al. (1998) offer a method for calculating transpiration of water stressed crops by applying a water stress coefficient (Ks). This requires soil moisture to be determined for computation of T under dynamic growing conditions. A simple soil water balance such as CropWat (Smith, 1992) can be applied to describe the dynamics of soil moisture following rainfall and irrigation applications. Alternatively, soil moisture needs to be measured in situ with sensors (Dorigo et al., 2013) or by satellite (Scott et al., 2003).

Alternative to generic two-layer crop coefficient model, T can be predicted by the one-layer PM equation with a variable canopy resistance (rc) that responds to all environmental conditions (i.e. temperature, vapor pressure deficit, solar radiation, soil moisture) and to crop development (Dolman, 1993, Teixeira et al., 2008). The latter could potentially be more accurate under varying environmental and water limiting conditions than the generic FAO56 basal coefficient method with sugarcane crop coefficients. This is at least a hypothesis worth testing. The widely accepted Jarvis-Stewart model (Jarvis, 1976, Stewart, 1988) was used in this study to estimate rc in the PM equation. This model has the drawback that it needs to be calibrated for different types of forests (Matsumoto et al., 2008) and crop types (cotton and wheat: Garatuza-Payan et al., 1998; grass and millet: Hanan and Prince, 1997; maize and vineyard: Li et al., 2014). We believe, however, that once the Jarvis-Stewart model variables are calibrated for a crop such as sugarcane, it could be implemented for irrigated and rainfed fields over a range of climatic conditions. Another great advantage of resistance type of models over crop coefficients is that they help to compute instantaneous energy balance fluxes, which are used in climate, numerical weather prediction and remote sensing studies.

The idea of parameterizing sugarcane T fluxes with resistances is not entirely new. Grantz and Meinzer (1991) calculated for instance the relationship between rc and environmental conditions (photon flux density and humidity) for irrigated sugarcane. McGlinchey and Inman-Bamber (1996) calculated the surface resistance from the relation of the bulk stomata resistance and leaf area index (LAI) for irrigated sugarcane. The current study combined environmental stress functions and LAI to calculate an hourly variable rc value. We calibrated the parameters of the environmental stress functions and LAI development at both hourly and daily intervals. Since we are looking for a general method for predicting sugarcane water demand and crop water productivity for rainfed and irrigated fields, the accuracy of T estimates derived from a direct one-layer PM equation were compared with that of the more classical FAO56 basal coefficient method.

Section snippets

Description of sites

Field data for this study were collected in South Africa in a commercial rainfed sugarcane field in the province of KwaZulu-Natal and in a commercial irrigated sugarcane field in the province of Mpumalanga (Fig. 1).

Water consumption of sugarcane

The FAO56 specifies the generic Kc of sugarcane of initial stage (0.4), mid-stage (1.25) and end stage (0.75) for water-unstressed conditions (Ks = 1). The generic Kc values of mid and end stage were adjusted for specific weather conditions as described by Allen et al. (1998). The weather-adjusted Kc values of mid-stage were 1.24 for rainfed sugarcane and 1.21 for irrigated sugarcane. The weather-adjusted values of end stage were 0.72 for rainfed sugarcane and 0.70 for irrigated sugarcane. Fig. 3

Conclusions

Because of the well tested bio-physical parameterization under a range of climatic conditions and landscapes, the Jarvis-Stewart model is widely used in climate and remote sensing models for calculating rc in different ecosystems. The FAO56 methodologies are very common due to their simplicity. While the FAO56 methods are sound, crop ET can be predicted more accurately with a variable canopy resistance instead of a variable crop coefficient. Moreover, the ability to compute T is important for

Acknowledgments

The authors would like to thank the University of KwaZulu-Natal (UKZN), the South African Sugarcane Research Institute (SASRI), the Department of Water Affairs and Forestry and the Water Research Commission (WRC) and the Department of Agriculture, Forestry and Fisheries of South Africa for contributing funding to the field measurements associated with the work. Results presented here where generated as part of the WRC project K5/2079//4 “Water use efficiency of irrigated agricultural crops

References (50)

  • S.E. Park et al.

    Decline in the growth of a sugarcane crop with age under high input conditions

    Field Crops Res.

    (2005)
  • P. Rochette et al.

    Estimation of maize (Zea mays L.) canopy conductance by scaling up leaf stomatal conductance

    Agric. For. Meteorol.

    (1991)
  • D.M. Smith et al.

    Growth and function of the sugarcane root system

    Field Crops Res.

    (2005)
  • J.B. Stewart

    Modelling surface conductance of pine forest

    Agric. For. Meteorol.

    (1988)
  • A.H.d.C. Teixeira et al.

    Crop water parameters of irrigated wine and table grapes to support water productivity analysis in the São Francisco river basin, Brazil

    Agric. Water Manage.

    (2007)
  • A.H.d.C. Teixeira et al.

    Energy and water balance measurements for water productivity analysis in irrigated mango trees, Northeast Brazil

    Agric. For. Meteorol.

    (2008)
  • ASCE

    Hydrology Handbook, American Society of Civil Engineers. Task Committee on Hydrology Handbook

    (1996)
  • Allen, R.G., Pereira, L.S., Raes, D., 1998. Crop evapotranspiration: guidelines for computing crop water requirements....
  • R.G. Allen et al.

    FAO-56 dual crop coefficient method for estimating evaporation from soil and application extensions

    J. Irrig. Drain. Eng.

    (2005)
  • R.G. Allen

    Assessing integrity of weather data for reference evapotranspiration estimation

    J. Irrig. Drain. Eng.

    (1996)
  • E. Bappel et al.

    Assimilation in a sugarcane yield forecasting model of biophysical parameter estimated by remote sensing using SPOT4&5 data

    Proc. Int. Conf.

    (2005)
  • W.G.M. Bastiaanssen et al.

    Surface energy balance and actual evapotranspiration of the transboundary Indus basin estimated from satellite measurements and the ETLook model

    Water Resour. Res.

    (2012)
  • M.B. Ben-Mehrez et al.

    Estimation of stomatal resistance and canopy evaporation during the HAPEX-MOBILHY experiment

    Agric. For. Meteorol.

    (1992)
  • J.A. Businger et al.

    Introduction to Obukhov's paper on ‘turbulence in an atmosphere with a non-uniform temperature'

    Bound.—Layer Meteorol.

    (1971)
  • O.M.R. Cabral et al.

    Water use in a sugarcane plantation

    GCB Bioenergy

    (2012)
  • Cited by (0)

    View full text