On the Dynamics of Evaporative Losses from Penman-Monteith with Fixed and Variable Canopy Resistance During Partial and Complete Maize Canopy

The Penman-Monteith (PM) model is based on a one-dimensional aerodynamic and energy balance approach and provides powerful information and assessment regarding the interrelationships between vegetation surface properties, evaporative losses from the surface, and the microclimatic conditions of the boundary layer. However, some challenges still exist in terms of the practical applicability of the model to estimate actual evapotranspiration (ETc) rates from different vegetation surfaces using a one-step approach. Using field measurements of ETc, leaf area index (LAI), canopy height (h), leaf stomatal resistance (rL), net radiation (Rn), soil heat flux (G), and other microclimatic and plant physiological variables, we made an attempt to delve into and tackle the following topics related to the operational characteristics of the PM model: (1) estimating ETc using the one-step approach of the PM for a non-stressed maize (Zea mays L.) canopy from measured and scaled up canopy resistance (rc) values; (2) comparing PM one-step ETc values (ETc-PM) with variable rc and those estimated using the ASCE PM (ETc-ASCE-PM) two-step approach [i.e., reference ET adjusted with crop coefficients (ETref Kc) with fixed rc] with the Bowen ratio energy balance system-measured ETc (ETc-BREBS) and assessing the diurnal, daily, and seasonal pattern of all ETc values during partial and complete canopy conditions; and (3) investigating the data source impact [i.e., solving the PM using the near-reference weather station-measured microclimatic data (ETc-PM-WS) over the grass surface and using the BREBS-measured microclimatic data (ETc-PM-BREBS) over the maize canopy] on the performance and dynamics of the PM model. The seasonal average minimum canopy resistance (rc-min) was measured as 56 s m-1, whereas the seasonal average rc-max was 109 s m-1. There was a strong correlation [r2 = 0.88, root mean squared difference (RMSD) = 0.11 mm h-1, n = 768] between the ETc-PM-WS and ETc-BREBS. Overall, the ETc-PM-WS overestimated ETc-BREBS by 9%. Overestimation was larger at higher ETc rates (>0.8 mm h-1). The ETc-ASCE-PM with a fixed rc value (50 s m-1) performed similarly to the ETc-PM-WS (r2 = 0.88, RMSD = 0.10 mm h-1). When we solved the PM model using the BREBS-measured climatic data (including measured Rn and G) over the maize canopy (ETc-PM-BREBS), the performance of the PM improved significantly. The estimations were within 2% of the ETc-BREBS with a higher r2 (0.93) and lower RMSD (0.08 mm h-1) than the ETc-PM-WS and ETc-ASCE-PM. We found that when the weather station-measured climatic data were used to solve the PM model, a 7% error can be introduced by using estimated Rn and G relative to using measured Rn and G. Overall, the ETc-PM-WS and ETc-PM-BREBS daily estimates were close to those of ETc-BREBS at a wide range of LAI, with ETc-PM-BREBS performing better than the ETc-PM-WS in most cases. The ETc-PM-WS and ETc-PM-BREBS underestimated during early stages of plant development. The ETc-PM-WS usually overestimated towards the end of the season, with the magnitude of overestimations being greater from the maximum LAI (5.30) to the end of the season (LAI = 3.70). The ETc-ASCE-PM overestimated by approximately 10% in the 1.2 < LAI < 2.7 range and then consistently underestimated by 10% to 20% until LAI reached 4.6. We observed that some of the largest underestimation by the ETc-PM-BREBS occurred in the early season during partial canopy when 1.20 < LAI < 2.00, with underestimations reaching up to 30%. Unlike with the ETc-PM-WS and ETc-ASCE-PM, we did not observe a distinct pattern of over or underestimation by the ETc-PM-BREBS for the LAI range of 1.50 < LAI < 5.30. The PM model successfully tracked the measured ETc throughout the season for a wide range of LAI using scaled up variable rc.