Influence of water use efficiency parameterizations on flux variance similarity-based partitioning of evapotranspiration

Despite the high sensitivity of water use efficiency (WUE) estimates to intracellular carbon dioxide concentrations (c(i)) in the Flux Variance Similarity (FVS)-based partitioning method, a systematic analysis of the sensitivity of WUE to c(i) parameterizations has largely been lacking. Using high-frequency (10 Hz) eddy covariance data for two crop sites: wheat (Triticum aestivum L.) and canola (Brassica napus L.), we performed a sensitivity analysis of four ci parameterizations (constant c(i) value, constant c(i)/c(a) ratio, and c(i)/c(a) as square root and linear functions of vapor pressure deficit) and compared them with the optimized WUE approach with no adjustable parameter. The results illustrated the role of c(i) parameterizations on the evapotranspiration (ET) partitioning results (i.e., transpiration (T) to ET ratios). Notably, constant c(i) value and constant c(i)/c(a) ratio parameterizations for the largest considered c(i) values (commonly used default values in most previous studies) showed comparable T:ET with the optimized WUE approach. Additionally, all these three models produced reduced T:ET in wet periods and increased T:ET in dry periods. In contrast, square root and linear models were unable to accurately capture expected trends of T:ET for wet and dry periods, and also showed large discrepancies when compared with the optimal WUE approach. The results suggest that optimal parameterizations of c(i) should be derived in constant c(i) value and constant c(i)/c(a) ratio methods to accurately capture temporal variations of WUE and T:ET. The results also indicate the potential of the optimum model for inter-model comparison, especially in sensitivity analysis, for FVS partitioning in C-3 species. This study provides novel insights into the implications of the choice of parameterization on the WUE estimations and partitioning outcomes.