Accurate monitoring of farmland environment and soil pollution is of great significance for food security and ecologically sustainable development in recent years. Taking Xiong'an New Area as the research area, this study aims to inverse the heavy metal content in farmland. Spectral response data of heavy metal content was extracted from multi-source remote sensing data Sentinel-2 (L2A), Zhuhai-1 (OHS), and field measurement. Pollution levels of heavy metal were evaluated in farmland soil using the one-factor and Nemero indexes. The heavy metal elements were screened to cause soil pollution in farmland, such as copper (Cu), zinc (Zn) and plumbago (Pb). Partial least squares regression (PLSR) was integrated with principal component and multiple linear regression. The quadratic inversion model was then established for the contents of the three heavy metal elements. The quantitative inversion of the heavy metal contents was realized in a wide range of soil. Sentinel-2 was used to extract seven vegetation indices, while Zhuhai-1 was to extract the original spectral reflectance of the sample points, as well as four transformed spectral reflectance of sample points. Pearson's correlation coefficient was used to analyze the correlation between spectral reflectance and vegetation index with the content of three heavy metal elements. A comparison was also made on the correlation coefficients, sensitive bands and vegetation indices. The spectral response indicators were modeled as the independent variables, while the measured parameters of the soil heavy metal contents were the dependent variables. The models were finally evaluated using the coefficient of determination (R²), mean absolute error (MAE), and root mean squared error (RMSE). The results showed that excellent performance was achieved in the overall inversion accuracy of the three models. The R², RMSE, and MAE values of the Pb content inversion model were 0.490, 4.66, and 1.92 mg/kg, respectively. The inversion model was obtained for Cu with the R² of 0.491, the RMSE of 16.85 mg/kg, and the MAE of 3.69 mg/kg. While, the inversion model of Zn was achieved with the R² of 0.664, the RMSE of 20.63 mg/kg, and the MAE of 9.36 mg/kg. The spatial distribution of soil heavy metal content in farmland was mapped, according to the constructed inversion model of soil heavy metal content. The visualization of the inversion was conducive to the distribution of regional soil heavy metals from a more intuitive perspective. Among them, the Pb element content in soil in most areas was within the risk screening standard; Cu element exceeded the risk screening value of soil pollution in the southwestern and western parts of the test area; The western and southwestern parts of the test area were more seriously contaminated by Zn element, indicating the sporadic distributions in some farmlands in the southeastern part. The Cu and Zn contents in other areas were within the national risk control value of soil pollution. Therefore, the multi-source remote sensing data and spectral response can be expected to synergistically invert the Cu, Zn and Pb content of soil heavy metals. High feasibility and accuracy can also be obtained to investigate soil heavy metal pollution. At the same time, the findings can provide a theoretical basis for monitoring the heavy metal content of soil in large areas. A new idea can also be offered to construct the inverse model of heavy metal content in farmland soil.

Timely and accurate acquisition of winter wheat distribution is beneficial to the estimation of grain yield and optimization of planting structure. However, the previous means of acquisition cannot fully meet the needs of large-scale applications in recent years. Deep learning can be expected to introduce into agricultural remote sensing, in order to obtain the spatial distribution of agricultural resources, and further realize the development of agricultural modernization. Taking the winter wheat as the target object, the high-resolution Gaofen-2 remote sensing images were utilized to extract the winter wheat over a large area. Winter wheat was scattered distribution with the different field shapes in the remote sensing images. At the same time, the complex features were contained in the remote sensing images, leading to the interference in the recognition of winter wheat. In this study, a CAHRNet semantic segmentation network model was proposed to enhance the recognition accuracy and fuzzy details of field edges in the spatial distribution of winter wheat. The improved model first replaced the backbone feature network from ResNet with a high-resolution network HRNet. A unique parallel structure was connected to the multiple layers of sub-networks. The multiple sub-networks were continuously interacted with each other to repeatedly fuse the feature generated by multi-scale sub-networks. The spatial feature was enriched to rapidly locate the target object. The network was then maintained to extract the high-resolution feature. Secondly, the CAHRNet model was used to extract the spatial distribution of the winter wheat. The OCR module was introduced to aggregate the contextual feature of the target object. The extracted feature was classified by the backbone network. The aggregation of all the object regions was then evaluated to divide the regions into different categories. The correlation between the regions and the pixel points was enhanced to clarify the approximate location of the target object. Lastly, coordinate attention was introduced to capture the remote dependencies and retain the precise location of the target object, according to the dual spatial orientations. The correlation between multiple pixel points was enhanced to more accurately locate the exact position of the target object. The coordinate attention enhanced the spatial features of the target object and then reduced the detail loss caused by the model during feature extraction. Much more details were obtained than before in the prediction. The CAHRNet network model enhanced the supervision of the spatial features of the target object. The spatial relationship between the winter wheat and background pixels was established to enhance the region-pixel correlation. At the same time, the contextual feature of individual pixel points was utilized to enhance the pixel-point to pixel-point attention. The edge details of the target object region were preserved to highlight the overall features of winter wheat. The segmentation accuracy was improved to realize the fine extraction of winter wheat spatial distribution. The experimental results show that the optimal evaluation was achieved in each index on the high-resolution remote sensing dataset, with the MIoU and PA reaching 81.72% and 97.08%, respectively. The MIoU performance was improved by 9.90 and 2.44 percentage points, respectively, compared with the typical U-Net and DeepLabv3+, while the PA model was improved by 6.80 and 1.59 percentage points, respectively. The finding can provide technical support for the further acquisition of winter wheat crop distribution.