Jose et al. (2025) Improvement of soil moisture estimates over the indian domain: an anomaly bias correction approach

Identification

• Journal: Theoretical and Applied Climatology
• Year: 2025
• Date: 2025-09-09
• Authors: Vibin Jose, M. L. Riba, Anantharaman Chandrasekar
• DOI: 10.1007/s00704-025-05761-z

Research Groups

• Indian Institute of Space Science and Technology, Thiruvananthapuram, Kerala, India

Short Summary

This study introduces and evaluates an anomaly-based bias correction method for assimilating Soil Moisture Active Passive (SMAP) satellite retrievals into the Noah Land Surface Model (LSM) over the Indian domain. It demonstrates that this novel approach significantly improves soil moisture (SM) estimates and better captures irrigation signals, particularly during dry seasons, outperforming the traditional cumulative distribution function (CDF) matching method.

Objective

• To improve soil moisture (SM) estimates over the Indian domain by testing an anomaly-based bias correction method for assimilating Soil Moisture Active Passive (SMAP) satellite retrievals into the Noah Land Surface Model (LSM).
• To assess the potential of this anomaly correction method to identify irrigation signals and enhance model accuracy, especially compared to the cumulative distribution function (CDF) matching method.

Study Configuration

• Spatial Scale: Indian domain (6.6°N to 38°N latitude, 66°E to 99°E longitude) at a spatial resolution of 0.09° x 0.09°.
• Temporal Scale: Simulations conducted from 1 January 2016 to 31 December 2018, initialized with a spin-up period from 1 January 2014 to 1 January 2016.

Methodology and Data

• Models used:
  • Noah Land Surface Model (LSM) within NASA Land Information System.
  • Ensemble Kalman Filter (EnKF) for data assimilation.
  • Anomaly correction method for bias correction.
  • Cumulative distribution function (CDF) matching technique for comparative bias correction.
• Data sources:
  • Assimilation Data: Soil Moisture Active Passive (SMAP) 9 km daily global Level-3 soil moisture product.
  • Atmospheric Forcing Data (excluding rainfall): Global Data Assimilation System (GDAS).
  • Rainfall Forcing Data:
    • Integrated Multi-satellitE Retrievals for GPM (IMERG-GPM) (0.1° x 0.1°, half-hourly).
    • Global Data Assimilation System (GDAS) precipitation data (0.25° x 0.25°, 3-hourly).
    • Tropical Rainfall Measuring Mission (TRMM) 3B42 V7 rainfall data (0.25° x 0.25°, 3-hourly).
  • Land Cover Data: Moderate Resolution Imaging Spectroradiometer-International Geosphere-Biosphere Programme (MODIS-IGBP) (1 km resolution).
  • Elevation Data: Shuttle Radar Topography Mission (SRTM).
  • Soil Texture Data: State Soil Geographic-Food and Agriculture Organisation (STATSGO-FAO).
  • Other LSM Parameters: National Centres for Environmental Prediction (NCEP) (slope type, maximum albedo, greenness fraction), International Satellite Land Surface Climatology Project 1 (ISLSCP 1) (bottom temperature).
  • Validation Data: Global Land Evaporation Amsterdam Model (GLEAM) version 3.6a soil moisture data (1980-2021).
  • Irrigation Data: Global Map of Irrigation Areas (GMIA) (5 arc minutes resolution).

Main Results

• Data assimilation (DA) with the anomaly correction (DA-AC) method consistently outperformed DA with CDF matching (DA-CDF) and open-loop (OL) Noah LSM estimates, particularly during the winter and pre-monsoon seasons.
• Root Mean Square Error (RMSE) values for DA-AC soil moisture estimates were lower during pre-monsoon and winter seasons across all three precipitation forcings (TRMM, GDAS, IMERG-GPM).
• The improvements observed with DA-AC are attributed to its effectiveness in capturing irrigation signals, which are significant during India's dry pre-monsoon and winter seasons.
• Comparison with the Global Map of Irrigation Area (GMIA) revealed that improvements from GDAS-forced DA-AC were most significant over highly irrigated regions (e.g., West Bengal, Bihar, Uttar Pradesh, Haryana, Punjab, Gujarat).
• The difference between DA-AC soil moisture and OL soil moisture estimates was substantially larger than that between DA-CDF and OL estimates, especially in irrigated areas.
• GDAS-forced soil moisture estimates generally showed lower RMSE values compared to estimates forced by TRMM and IMERG-GPM precipitation data.
• While all GDAS-forced estimates (OL, DA-CDF, DA-AC) underestimated soil moisture compared to GLEAM, DA-AC showed improved results across all five homogeneous regions of India, with more pronounced improvements in irrigated areas.

Contributions

• First application and evaluation of an anomaly-based bias correction method for assimilating satellite soil moisture data into a land surface model over the Indian domain.
• Demonstrates the superior performance of the anomaly correction method over the widely used CDF matching technique in improving soil moisture estimates and explicitly capturing irrigation signals.
• Provides quantitative evidence of enhanced model accuracy, particularly during high-irrigation seasons (winter and pre-monsoon), by reducing RMSE values.
• Highlights the potential of SMAP soil moisture data, when assimilated with an appropriate bias correction, to reveal unmodeled processes like irrigation in land surface models.

Funding

• Not explicitly stated in the provided text, beyond acknowledging the Indian Institute of Space Science and Technology for High Performance Computing.

Citation

@article{Jose2025Improvement,
  author = {Jose, Vibin and Riba, M. L. and Chandrasekar, Anantharaman},
  title = {Improvement of soil moisture estimates over the indian domain: an anomaly bias correction approach},
  journal = {Theoretical and Applied Climatology},
  year = {2025},
  doi = {10.1007/s00704-025-05761-z},
  url = {https://doi.org/10.1007/s00704-025-05761-z}
}