In:
Soil Science Society of America Journal, Wiley, Vol. 87, No. 3 ( 2023-05), p. 613-630
Abstract:
Random forest and cubist performed better on spatial prediction of the topsoil pH compared with other machine learning (ML) models. Hybrid models combining ML algorithms with residual kriging outperformed their standalone model counterparts. There is no significant difference between the prediction results of different hybrid models except for artificial neural network kriging. Boruta algorithm revealed that topsoil pH of cropland is mainly affected by altitude and climatic factors related to soil water availability.
Type of Medium:
Online Resource
ISSN:
0361-5995
,
1435-0661
Language:
English
Publisher:
Wiley
Publication Date:
2023
detail.hit.zdb_id:
241415-6
detail.hit.zdb_id:
2239747-4
detail.hit.zdb_id:
196788-5
detail.hit.zdb_id:
1481691-X
SSG:
13
SSG:
21
Permalink