Gupta et al. (2026) SPAR-TC: A framework for accounting spatial representativeness in triple collocation

Identification

• Journal: Environmental Modelling & Software
• Year: 2026
• Date: 2026-01-13
• Authors: Diksha Gupta, C.T. Dhanya
• DOI: 10.1016/j.envsoft.2026.106874

Research Groups

• Department of Civil Engineering, Indian Institute of Technology, Delhi, India

Short Summary

This paper introduces the Spatially Representative Triple Collocation (SPAR-TC) framework to explicitly account for spatial representativeness errors in geophysical measurements, demonstrating that it yields more reliable error variance estimates, especially in heterogeneous regions, compared to traditional Triple Collocation.

Objective

• To develop and evaluate a new Triple Collocation framework (SPAR-TC) that explicitly accounts for spatial representativeness errors arising from different spatial measurement systems, thereby improving the quantification of "true" error variance in geophysical datasets.

Study Configuration

• Spatial Scale: Spatially heterogeneous regions; varying spatial scales of measurement systems; application to remotely sensed precipitation data.
• Temporal Scale: Not explicitly specified in the provided text, but implied for geophysical monitoring applications.

Methodology and Data

• Models used:
  • Spatially Representative Triple Collocation (SPAR-TC) framework
  • Traditional Triple Collocation (TC) framework (for comparison)
• Data sources:
  • Synthetic soil moisture experiment data
  • Real-world remotely sensed precipitation data
  • Ground-based observations (for consistency checks of error variance estimates)

Main Results

• SPAR-TC provides more reliable estimates of "true" error variance compared to traditional TC.
• The performance improvement of SPAR-TC in estimating error variance is particularly pronounced in spatially heterogeneous regions.
• Both SPAR-TC and traditional TC produce comparable rankings of datasets based on their performance.
• SPAR-TC's error variance estimates show greater consistency with ground-based observations than those derived from traditional TC.

Contributions

• Introduction of SPAR-TC, a novel framework that explicitly addresses and accounts for spatial representativeness errors within the Triple Collocation methodology.
• Improved quantification of "true" error variance for geophysical datasets, particularly those with differing spatial support and in spatially heterogeneous environments.
• Enhanced robustness in error characterization for remotely sensed data by integrating the spatial variability of the "ground truth."

Funding

• Not specified in the provided text.

Citation

@article{Gupta2026SPARTC,
  author = {Gupta, Diksha and Dhanya, C.T.},
  title = {SPAR-TC: A framework for accounting spatial representativeness in triple collocation},
  journal = {Environmental Modelling & Software},
  year = {2026},
  doi = {10.1016/j.envsoft.2026.106874},
  url = {https://doi.org/10.1016/j.envsoft.2026.106874}
}