YangMPS: Multi-Point Statistics for Non-Stationary Spatial Processes

YangMPS is a novel multi-point statistics simulation method designed to address non-stationarity in spatial data. Developed by Junjie Li and published on figshare in May 2026, the method separates spatial variables into regular and irregular components for more accurate modeling. Tests on simulated and real-world datasets showed it achieves significantly better accuracy than existing direct sampling and detrending methods.

Use Cases

• Modeling non-stationary spatial processes based on the described trend-irregular component separation.
• Simulating geographical phenomena using density peak clustering and kernel density estimation of multi-point patterns.
• Benchmarking new spatial simulation methods against direct sampling and detrending techniques.
• Reconstructing irregular spatial components via Monte Carlo simulation using derived cumulative distribution functions.

Strengths

• Method is theoretically proven to obtain optimal unbiased trend estimates via Yang Chizhong filtering.
• Published under the permissive CC-BY-4.0 license for open use and modification.
• Reported to achieve significantly better accuracy than existing methods in tests.

Limitations

• The dataset is very small at 280.4 KB, indicating limited scope or example data.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale applications.

Provenance

Source

figshare

Collection Method

Developed as a research method; likely contains implementation code and test data.

Freshness

Last updated 2026-05-11 07:41:44; freshness should be verified.