Spatial interpolation methods to predict airborne pesticide drift deposits on soils using knapsack sprayers

Area of Science:

• Environmental Science
• Geospatial Analysis
• Agricultural Science

Background:

• Accurate spatial prediction of pesticide drift is crucial for environmental risk assessment and regulatory compliance.
• Understanding the uncertainties in mass balance calculations is essential for effective pesticide management.

Purpose of the Study:

• To compare the performance of eight different spatial interpolation methods for predicting pesticide drift deposits on soil.
• To evaluate the impact of the number of sampling locations on prediction accuracy and the modeling of extreme data behavior.

Main Methods:

• Eight spatial interpolation methods were employed: Thiessen, kriging, spatial vine copulas, Karhunen-Loève expansion, and Integrated Nested Laplace Approximation (INLA).
• Predictions were generated using two datasets with 39 and 47 locations, respectively.
• Leave-one-out cross-validation was used to assess prediction accuracy.

Main Results:

• Increasing the number of locations improved prediction accuracy and the modeling of extreme data behavior.
• The Thiessen method exhibited the highest prediction errors, while linear interpolation methods and spatial copulas showed improved accuracy.
• Spatial copula methods demonstrated a notable increase in prediction accuracy, outperforming other methods in modeling extreme values.

Conclusions:

• Spatial interpolation methods vary significantly in their ability to predict pesticide drift, with spatial copulas offering superior performance.
• Increasing data density enhances the reliability of spatial predictions for pesticide drift.
• This study provides a foundation for modeling mass balance uncertainties in pesticide application studies.