.png){kind=icon}
.png){kind=icon}
Return
Two-dimensional data examination by exploratory functional data analysis to improve detection of scattered soil contamination by Cu-bearing pesticides
.png){kind=icon}
PRE
AI
T
Š
M
K
I
S
J
DOI：10.1016/j.gexplo.2026.108054.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The paper addresses a persistent challenge of distinguishing natural and anthropogenic causes of agricultural soil contamination, in particular differentiation of Cu-pesticide impacts from fluvial contamination and influence of Cu-rich bedrock. It proposes new ways to identify scattered or diffuse contamination of soils, whose definitions are actually missing and cannot be based on the presence of high-value outliers. Major part of results was taken from the database of the Czech state monitoring of agricultural soils using 2 M HNO3 extraction. That monitoring has ca. 1 observation per km2 of the top horizon of agricultural soils, of which 0.7% is represented by vineyards and hop gardens affected by Cu-bearing pesticides. The used methods included exploratory functional data analysis (EFDA), in particular probability density functions (PDFs) of concentration data series. The entire national monitoring dataset was separated into 77 administrative districts. Those subsets were subjected to EFDA and contamination was revealed by comparing Cu PDFs in districts with and without contamination based on the sharpness of the main concentration mode and heavy tails in high concentrations. Zinc was tested as an auxiliary element correcting for a part of Cu concentration variability, that was verified by soil sampling. EFDA provides unprecedented way to identify scattered contamination without high-value outliers only by shape analysis of the shape of main concentration mode and heavy tails; it successfully identified districts in which at least 3% agricultural soils are used for grapevine or hop production with very limited or no interference from other contamination sources and from bedrock geology, yet not achieved. Heatmaps of probability density functions and their clustering can thus be recommended for exploration of soil monitoring datasets and possibly other subsets of big datasets, where shape of data distribution functions is relevant.
Keywords:
Soil
Contamination
Fungicides
Copper
Functional analysis
Data exploration
Journal
IF:
3.3
Papers:
3.8K
Citations:
9.5K
Organization
Cited Papers