Dastjerdi et al. (2026) Comparing novel backward hydrological models for watershed-scale precipitation estimation: an evaluation of inverted PDM and Kirchner-hybrid structures

Identification

• Journal: Scientific Reports
• Year: 2026
• Date: 2026-03-20
• Authors: Pouria Asgari Dastjerdi, Mohsen Nasseri
• DOI: 10.1038/s41598-026-42647-0

Research Groups

• School of Civil Engineering, College of Engineering, University of Tehran, Tehran, Iran

Short Summary

This study developed and evaluated two novel backward hydrological models, an inverted Probability Distributed Model (PDM) and a hybrid Soil Moisture to Rain (SM2RAIN)-Kirchner model, for daily watershed-scale precipitation estimation. The locally calibrated backward models significantly outperformed established Global Gridded Precipitation Products (GGPPs), with the Kirchner model achieving the highest performance (KGE = 0.62) and the inverted PDM proving robust (KGE = 0.55).

Objective

• To develop and evaluate two novel "bottom-up" hydrological models (inverted PDM and hybrid SM2RAIN-Kirchner) for daily watershed-scale precipitation estimation.
• To systematically compare the performance of these new models against multiple SM2RAIN configurations and benchmark Global Gridded Precipitation Products (GGPPs) in data-scarce regions.

Study Configuration

• Spatial Scale: Watershed-scale (Walnut Gulch Experimental Watershed, Arizona, USA).
• Temporal Scale: Daily.

Methodology and Data

• Models used: Inverted Probability Distributed Model (PDM), hybrid Soil Moisture to Rain (SM2RAIN)-Kirchner model, various SM2RAIN configurations, and benchmark Global Gridded Precipitation Products (GGPPs).
• Data sources: Soil Moisture Merged via Modified Collocation (SMMC) product, in-situ soil moisture observations (SMOBS).

Main Results

• Locally calibrated backward models significantly outperformed established Global Gridded Precipitation Products (GGPPs).
• The Kirchner model driven by Soil Moisture Merged via Modified Collocation (SMMC) achieved the highest performance (KGE = 0.62).
• The inverted PDM proved to be a robust new approach, achieving a Kling-Gupta Efficiency (KGE) of 0.55.
• Spatially integrated SMMC product led to more robust and generalizable models compared to those driven by in-situ observations, which caused overfitting in some structures.
• While backward models excelled in quantitative accuracy, they demonstrated less skill in event detection compared to GGPPs.

Contributions

• Introduces viable new model structures (inverted PDM and hybrid SM2RAIN-Kirchner) for backward hydrological modeling and precipitation estimation.
• Confirms the achievability of high-accuracy precipitation estimates in data-scarce regions using merged, globally available datasets.
• Provides a systematic evaluation and comparison of novel backward models against existing SM2RAIN configurations and benchmark GGPPs.
• Highlights a critical trade-off between quantitative accuracy and event detection skill in backward hydrological models.
• Demonstrates the superior robustness and generalizability of models driven by spatially integrated merged soil moisture products over in-situ observations.

Funding

Not specified in the provided text.

Citation

@article{Dastjerdi2026Comparing,
  author = {Dastjerdi, Pouria Asgari and Nasseri, Mohsen},
  title = {Comparing novel backward hydrological models for watershed-scale precipitation estimation: an evaluation of inverted PDM and Kirchner-hybrid structures},
  journal = {Scientific Reports},
  year = {2026},
  doi = {10.1038/s41598-026-42647-0},
  url = {https://doi.org/10.1038/s41598-026-42647-0}
}