.png){kind=icon}
Return
A multiobjective optimization spatial sampling method for regional yield potential estimates using multisource environmental data
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eja.2026.128221.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In terms of regional applications of crop growth models, selecting typical agricultural stations via spatial sampling can effectively reduce regional data acquisition costs and ensure the spatial representativeness of simulation results. With the major winter wheat-producing regions in China as the research area and on the basis of data from 129 agricultural stations, a multiobjective optimization spatial sampling method that combines conditioned Latin hypercube sampling (cLHS) and multipath spatial simulated annealing (MP-SSA) was proposed, and its performance was compared with stratified spatial sampling (SSS). Results indicated that the SHapley Additive exPlanations (SHAP) method was capable of identifying key environmental variables: including digital elevation model (DEM), March sunshine duration (SSD3), and the March leaf area index (LAI3). The multiobjective optimization function (MOOF) designed based on the feature space (FS) and the geographical space (GS), achieved higher sampling accuracy than those utilizing FS or GS alone. When the sample size was less than 30, FGS produced smaller P-values than FS, GS and SSS sampling methods, indicating high significance, and the errors of regional yield potential were all within 300 kg/ha. By setting independent annealing paths for each objective, the MP-SSA method avoids the issue whereby traditional algorithms tend to become trapped at local optima and achieves a balance between attribute representativeness and spatial representativeness. When the sample size reaches 50, the sampling results of different optimization functions tend to stabilize. The method proposed in this study can be used to determine typical agricultural stations with significant representativeness of the natural environment, ensuring the optimal estimation of regional yield potential simulations.
Keywords:
Yield potential
Spatial sampling
WheatGrow model
Environmental variables
Journal
IF:
5.5
Papers:
755
Citations:
1.3W
Organization
View more organizations
Cited Papers
No cited papers available