Modelling LEAF stomatal resistance for common REED, peach‐leaf willow and cottonwood riparian plant communities

Abstract

AbstractAccounting for the impact of riparian plant communities on water balances is usually ignored due to the lack of data and information on water use of these plant communities. This is especially important for areas with limited water resources and in areas where water resources are shared between different users (agriculture, riparian zones, other ecosystems/agro‐ecosystems, municipalities, recreational, mining, agro‐ecology, etc.). Modelling stomatal resistance (rL) of riparian vegetation species is a complex process, but is also one of the limited processes to estimate evaporative losses for these plant communities for more inclusive, complete and robust water balance and hydrologic analyses for forecasting, planning, allocating and managing surface and groundwater resources and for hydrologic modelling. The objectives of this research were to (i) modify Jarvis‐type model for estimating rL for three riparian plant species [Common reed (Phragmites australis), Peach‐leaf willow (Salix amygdaloides) and Cottonwood (Populus sect. Aigeiros)], (ii) investigate the rL response to microclimate/environmental variables, and (iii) evaluate the photosynthetic photon flux density (PPFD) versus rL response curves for estimating hourly rL. A Jarvis‐type model was modified by re‐parametrization using porometer‐measured data collected via extensive field campaigns in 2009 and 2010. Flooding in 2010 impacted the phenology and physiology of all species with increased rL and impedance in growth due to oxygen stress. While R2 values were generally low for both years for all species, the response of rL to the same climate variables under the same environmental conditions exhibited substantial variations between the species. Overall, the response of rL to climatic variables was species‐specific, complex and varied with surface and atmospheric conditions. For all species, stronger relationships between rL versus air temperature and net shortwave radiation were observed. The Jarvis‐type model underestimated rL for all species and hence required independent re‐parameterization for each specie independently. The model performance after re‐parameterization improved substantially during both calibration and validation. Validation of the new modified Jarvis‐type model (NMJ‐model) resulted in good estimates of rL with root‐mean‐squared difference (RMSD) and R2 values of 31 s m−1 and 0.51 for Common reed; 56 s m−1 and 0.81 for Peach‐leaf willow; and 237 s m−1 and 0.51 for Cottonwood, respectively. Strong agreements were found between measured and re‐parameterized NMJ‐model‐estimated rL values for all three species in both years, but the relationships improved significantly in 2009 with no flooding. The NMJ‐model‐estimated rL values were within 10, 7 and 11% of the measured values for Common reed, Peach‐leaf willow and Cottonwood, respectively. rL values estimated using PPFD versus rL response curves developed for individual species were used to estimate hourly rL and the response curves were able to effectively track the trends and magnitudes of rL. With re‐parameterization, the NMJ‐models developed and tested for each riparian vegetation specie provide robust estimates of rL that can be used to quantify evaporative losses of these riparian plant communities for more complete local and regional water balance analyses by accounting for riparian vegetation water use.