Estimation of soil water content and evapotranspiration from irrigated cropland on the North China Plain

Abstract

For nearly 30 y, cropland on the North China Plain (NCP) has been irrigated primarily by pumping groundwater with no sustainable management strategy. This has caused a continuous decline of the water table. A sustainable groundwater management and irrigation strategy must be established in order to prevent further decline of the water table; to do this, one must quantify soil water content and daily rates of deep percolation and locate evapotranspiration from irrigated cropland. For that purpose, we developed a three-layer soil-water balance (SWB) model based on an approach described by Kendy et al. (2003). In this model, the unsaturated soil zone is divided into three layers: a surface active layer, a middle active soil layer, and a lowest passive soil layer. The middle and the lowest layers dynamically change with the development of crop rooting depth. A simple "tipping bucket" routine and an exponential equation are used to redistribute soil water in the three soil layers. The actual evapotranspiration estimated is partitioned into soil evaporation and crop transpiration using a dual crop coefficient reference approach. At first, the model was calibrated using data obtained from five deficiently irrigated field plots located at an experimental site in the NCP between 1998 and 2003. Then, the model was validated by comparing estimated soil water contents with measured ones at three other plots with nondeficient irrigation. The estimates of actual evapotranspiration were compared with those measured with a large-scale weighing lysimeter (3 m 2 ). The index of agreement (IA) for soil water contents varied between 0.62 and 0.80; the concordance correlation coefficient (CCC) and the root mean square error obtained from the same comparison were 0.34-0.65 and 0.043-0.074 cm 3 cm -3 , respectively. The rates of 10 d mean evapotranspiration estimated by the model show a good fit to those measured by the large-scale lysimeter; this is indicated by IA = 0.94 and CCC = 0.88. Our results indicate that at the irrigated cropland on the plain, deep soil water-percolation rates are usually <200 mm y -1 under nondeficient-irrigation conditions.

Simulation of evapotranspiration and carbon dioxide flux in the wheat‐maize rotation croplands of the North China Plain using the Simple Biosphere Model

“…and are not available for model calibration and validation. Alternatively, the current models on this topic are usually based on the calibrations at the point‐scale (Jiang et al , 2008; Sun et al , 2009), making the point‐scale observations critical. However, measurements of the actual ET with different approaches still have large uncertainties.…”

“…the total of latent and sensible heat fluxes were under‐measured). On the other hand, the measurements of the large weighing lysimeter could be over‐measured because of the ‘oasis effect’ (Jiang et al , 2008). However, all these approaches are commonly used to measure ET, and which approach is better is difficult to justify if they are correctly handled.…”

Comparison of real evapotranspiration measured by weighing lysimeters with simulations based on the Penman formula and a crop growth model

“…In comparison, the application of a physiological-based grass cover growth model increased the simulation quality (IA was 0.83-0.92 and RMSE 0.6 and 0.9 mm d -1 , Table 3). In a similar study (Herbst et al, 2005) in Jiang et al (2008). In a previous investigation using the present data set of lysimeters 3-4 for the application of the soil water balance model SIMWASER (Stenitzer et al, 2007), NS for ETr was 0.18.…”

Optimizing the photosynthetic parameter Vcmax by assimilating MODIS-fPAR and MODIS-NDVI with a process-based ecosystem model

“…Except for the error in model parameters, there are also errors in observation. Measurements from the large weighing lysimeter could be inflated due to the "oasis effect" (Jiang et al, 2008), while the eddy covariance technique does not always achieve energy budget closure, and this may introduce the underestimation of latent and sensible heat flux (Liu et al, 2009). Therefore, the simulated ET was slightly lower than the corresponding observed lysimeter data at the three sites (the slop of the regression is less than 1 in Fig.…”