Abstract: This study synthetically considered the spatial position accuracy and data precision for crop planting structure in order to ensure that the hydrological and nutrient loss processes were more veritably simulated and the simulation accuracy was further improved. The classification and extraction of field crops were conducted and the land use map was modified using Normalized Difference Vegetation Index (NDVI) threshold method and Support Vector Machine (SVM) method based on GF-1 16 m WFV4 medium resolution remote sensing images in Hetao Irrigation District. The effect of the corrected spatial position and the improved accuracy of crop planting structure on the simulation accuracy of SWAT (Soil and Water Assessment Tool) model were evaluated using the modified land use map. The results showed that the classification of crops based on GF-1 16 m WFV4 remote sensing images agreed with the actual spatial distribution of crops in Hetao Irrigation District, with an overall accuracy of 89.61%, a mapping accuracy of over 88%, a user accuracy of over 88% and a Kappa coefficient of 0.86. The parameters with high level of sensitivity to the simulation of runoff and nitrate nitrogen were quite stable in the irrigation district. The simulation accuracy in terms of runoff was significantly affected by groundwater delay coefficient (GW_DELAY), groundwater evaporation coefficient (GW_REVAP), base flow alpha factor (ALPHA_BF), and soil evaporation compensation factor (ESCO). In addition, the simulation accuracy of nitrate nitrogen was markedly affected by nitrogen concentration in rainfall (RCN), the nitrate percolation coefficient (NPERCO), and the denitrification exponential rate coefficient (CDN). The corrected spatial position accuracy and data precision of crop planting structure effectively improved the accuracy of simulated values for runoff and nitrate nitrogen. In the calibration period (2009-2014), the R2 for simulated runoff and nitrate nitrogen were improved to 0.76 and 0.70 from 0.63 and 0.62, respectively by correcting crop pattern locations. The efficiency coefficients were improved to 0.69 and 0.55 from 0.53 and 0.50, respectively, while the relative errors were decreased by 6.00% and 4.94%, respectively. In the validation period (2015-2016), R2 was improved to 0.82 and 0.63 from 0.71 and 0.58, respectively. The efficiency coefficients were improved to 0.82 and 0.63 from 0.71 and 0.58, respectively, while the relative errors were decreased. Additionally, the R2 of simulated runoff and nitrate nitrogen were improved to 0.76 and 0.70 from 0.68 and 0.66 by improving the accuracy of crop pattern data, respectively, in the calibration period. The efficiency coefficients were improved to 0.69 and 0.55 from 0.60 and 0.53, respectively, while the relative errors were decreased. However, in the validation period, R2 was improved to 0.82 and 0.63 from 0.77 and 0.60, respectively, and efficiency coefficients were improved to 0.79 and 0.53 from 0.76 and 0.50, respectively. The simulation results of runoff and nitrate nitrogen based on SWAT model was easily affected by corrected spatial of crop pattern compared with data accuracy. Comprehensively correcting the spatial position and improving the data accuracy of crop planting structure effectively improved the simulation accuracy of the SWAT model in the irrigation district.