Estimation Of Water Use Efficiency Of Winter Wheat Based On AquaCrop Model And Multi-source Remote Sensing Data In Northern

Water resource is an important limiting factor in crop irrigation management. Since the demand for domestic and industrial water has increased with the growing population, there is increasing competition in agriculture water resource usage. Water resource for agriculture is becoming increasingly scarce in irrigation agriculture regions. Water productivity (WP) or water use efficiency (WUE, which is advocated as an indicator to analyze water use) was proposed to aid in water management. The current problems of serious water shortages and higher demand for crop yield can be solved by improving WUE in this region. The reasonable field irrigation practices can increase crop WUE and crop yield, thereby solving the problem of water shortages in agricultural irrigation regions. However, the current traditional methods of WUE determination require substantial amounts of manpower and time for scientists to obtain WUE results. With the rapid development of remote sensing technology in agricultural water management, it is possible to estimate WUE using remote sensing image data at the regional scale.In this study, field ASD hyperspectral data, satellite images (optical and radar images data), meteorological data, leaf area index (LAI), canopy cover, dry biomass and yield were obtained at the key growth stages of winter wheat. The WUE of winter wheat was estimated based on AquaCrop model and multi-source remote sensing data. Five aspects were mainly analyzed:1) Global sensitivity analysis of AquaCrop model applied to wheat yield and the related dynamic output variables; 2) assessment of the AquaCrop model for use in simulation of irrigated winter wheat canopy cover, biomass, and grain yield in the North China Plain; 3) assessing WUE in winter wheat combining the AquaCrop model and field ASD hyperspectral data; 4) estimating of LAI and biomass of winter wheat with a new optical and radar combination vegetation index using HJ and RADARSAT-2 images data; 5) estimation of WUE in winter wheat in Yangling district based on AquaCrop model, optical and radar images data using particle swarm optimization algorithm. Some important conclusions can be drawn as follows:1) Crop parameter sensitivity changed in response to crop parameter variation range. The ranking of the most important crop parameters based on the first order sensitivity index (FOSI) of winter wheat maximum dry biomass from highest to lowest was wp, cc, stbio, and mcc and that based on the FOSI of spring wheat maximum dry biomass was stbio, cc, wp, and mcc when the parameters’ variation ranges were set to ±10% fluctuations of the nominal values. Similar to the results above, the importance and selection of the crop parameters based on the FOSI of winter wheat and spring wheat maximum dry biomass were similar although some differences existed when the crop parameters’ variation ranges were set to ±30% and ±50% fluctuations of the nominal values, respectively. Some differences existed in selection and importance of the crop parameters based on the total sensitivity index (TSI) of winter wheat maximum dry biomass (wp, cc, stbio, rootdep, polmn, mcc, and psto) and spring wheat maximum dry biomass (stbio, cc, wp, and mcc) when the parameters’ variation ranges were set to ±10% fluctuations of the nominal values. However, the selection and importance of the crop parameters based on the TSI of winter wheat maximum dry biomass (wp, cc, stbio, and mcc) and spring wheat maximum dry biomass (wp, cc, stbio, and mcc) were similar when the parameters’ variation ranges were set to ±30% fluctuations of the nominal values. Larger differences existed in the order of the most important crop parameters based on the TSI of winter wheat maximum dry biomass (wp, cc, stbio, mcc) and spring wheat maximum dry biomass (stbio, cc, wp, dcc, rmexup, pstoshp, hilen, anaer, hi, mcc, remd, eme, and psto) when the parameters’variation ranges were set to ±50% fluctuations of the nominal values. The results from the time-series analysis revealed that the FOSI of winter or spring wheat canopy cover and dry biomass were the most sensitive to ssc, stbio, plomn, wp, cc, mcc, and num based on the times-series analysis. The TSI of crop parameters and the TSI of time series were more sensitive than the FOSI of crop parameters and the FOSI of time series. The FOSI and the FOSI time series showed good consistency between winter wheat in China and spring wheat in Canada. However, the TSI and the TSI time series revealed some differences between winter wheat and spring wheat.2) This paper demonstrated that the AquaCrop model adequately simulated the CC, BY, and GY of winter wheat under different planting dates and irrigation strategies. The simulated CC agreed well with the measured CC across all 4 years. The R2, RMSE, E of CC winter wheat ranged from 0.89 to 0.98,3.18% to 7.19% and 0.90 to 0.96, respectively. The measured and simulated BY were also closely related. The AquaCrop model calibrated the BY with the prediction error statistics of 0.92<R2<0.98,1.12<RMSE<1.84 ton/ha and 0.92<E<0.96. The simulated GY was also consistent with the measured GY with the R2, RMSE and E values of 0.93,0.52 ton/ha and 0.92, respectively. The results demonstrated that frequent irrigation obviously improved BY, GY, biomass WUE and grain WUE for winter wheat in 2010/2011. These results suggest that the AquaCrop model could be used to predict CC, BY and GY of winter wheat with a high degree of reliability under various planting dates and irrigation strategies situations in the North China Plain (NCP). Therefore, we concluded that AquaCrop is a useful decision-making tool for use in efforts to optimize wheat winter planting dates, and irrigation strategies.3) In summary, the most important conclusions that can be drawn from this study are as follows:(1) there is a good relationship between NDMI and biomass, with R2 and RMSE values of 0.84 and 1.43 ton/ha, respectively; (2) all spectral indices considered were highly related to water use efficiency (WUE); our results showed that TBWI is the best regression model for WUE estimation, with R2 and RMSE values of 0.73 and 0.15 kg/m3, respectively; (3) the R2 and RMSE values showed good performance of the simulated values for biomass and yield from the AquaCrop model in winter wheat; and (4) the results showed that the data assimilation methods (R2=0.79 and RMSE= 0.12 kg/m3) are better than the empirical statistics spectral indices methods for estimation accuracy of WUE.4) The highly significant correlation between LAI, biomass and optical spectral vegetation indices (OSVIs; Enhanced Vegetation Index, EVI; Modified triangular vegetation index 2, MTVI2.), radar polarization vegetation indices (RPVIs; radar vegetation index, RVI; double-bounce eigenvalue relative difference, DERD.). The ORVIs [MTVT2×DERD (R2=0.67) and MTVI2×RVI (R2= 0.68)] were highly related with LAI, but the ORVIs [Optimized soil adjusted vegetation index (OSAVI)xDERD (R2=0.79) and EVIxRVI (R2= 0.80)], the estimation accuracy of LAI and biomass were better using ORVIs than using OSVIs and RPVIs lonely. This study may provide a guideline for improving the estimations of LAI and biomass of winter wheat using multisource remote sensing data.5) Winter wheat GY and WUE were estimated based on the data assimilation methods which combined AquaCrop model, optical and radar images data using article swarm optimization (PSO) optimization algorithm in Yangling district. The method one:CC as the dynamic variable was inputted into the processing of assimilation method, a good consistency was found between the predicted GY and the measured GY (R2= 0.31 and RMSE=0.94 ton/ha) and the predicted WUE and measured WUE (R2= 0.34 and RMSE=0.29 kg/m3). The method two:BY as the dynamic variable was inputted into apply the processing of assimilation method, the predict GY and WUE were very in agreement with the measured GY (R2= 0.42 and RMSE=0.81 ton/ha) and WUE (R2= 0.34 and RMSE=0.25 kg/m3), respectively. The results presented that the BY as the dynamic variable was better than the CC as the dynamic variable for yield and WUE estimation.