Fig 1.
Research area location.
Fig 2.
Frequency distribution of soil petroleum hydrocarbons at sampling points.
Table 1.
Data distribution.
Fig 3.
Channel attention mechanism.
Fig 4.
Proposed 3DCNN architecture.
Fig 5.
Petroleum hydrocarbon array data flow layer-wise.
Table 2.
Hyperparameters of experiment.
Table 3.
Hyperparameters of experiment.
Fig 6.
Loss of 3DCNN.
Fig 7.
Loss of 3DCNN with CAM.
Fig 8.
Result of four interpolation method.
Table 4.
Result of 4 kinds of different spatial interpolation.
Table 5.
Result of ablation experiment.
Table 6.
Result of features ablation experiment.
Fig 9.
Petroleum hydrocarbon prediction map using 3DCNN interpolation method.