Model

All models are wrong, but some are useful.——George E. P. Box

Attachments

Supplementary Information includes Supplementary text S1-2, and Table S1.

Text S1

Source and sink factors of soil microplastics

  1. Source factors: clay, silt, sand, total organic carbon (TOC), electrical conductance (EC), pH, bulk density (BD);
  2. Sink factors: population density (PD), point of interest (POI), farmland area (FA), road area (RA), construction land area (CA).

Text S2

POI databases

POI data is the core data of location-based services. In GIS, POI data records the spatial and attribute information of geographical entities. In addition, POI data can identify different pollutant emission point sources, which can easily and accurately correlate pollutants and establish relevant and quantitative models. Microplastics are new pollutants produced by human activities, among which plastic waste and domestic sewage are important sources of microplastics. Therefore, we used Ospider 3.0 to capture 9 types of POI data related to plastics, including.Recycle-POI: Recycling station or waste or garbage.

  1. Shopping-POI: Supermarket, Mall.
  2. Electronic-POI: Electronic factory.
  3. Auto-tyre-POI: Automobile service, Tyre production.
  4. Plastic-POI: Plastic production, Package production, manufacturing, sales.
  5. Express-POI: Express delivery or logistics express.
  6. Lanudry-POI: Dry cleaning, laundry.
  7. Clothing-POI: Clothing, shopping services.
  8. Accommodation-POI: Hotel, Residence.

Table S1

Comparison of simulation goodness and prediction accuracy of different machine learning models in predicting soil microplastics abundance

Comparison of simulation goodness and prediction accuracy of different machine learning models in predicting soil microplastics abundance
  BP-GAO RF LSTM XGBoost RBF SVR-RBF
Train_$\mathrm{R^2}$ 0.53891 0.920112 0.468024 0.999997 0.999728 0.935819
Train_RMSE 0.140926 0.059283 0.138087 0.000374 0.003821 0.062681
Train_MAE 0.120594 0.054773 0.09757 0.00026 0.002627 0.045125
Test_$\mathrm{R^2}$ 0.482759 0.96512 0.963398 0.828356 0.410678 0.984548
Test_RMSE 0.164101 0.038982 0.182309 0.085429 0.147808 0.015366
Test_MAE 0.137078 0.030853 0.134099 0.072607 0.129603 0.011207
Go back to Model-Machine Learning