A new approach for generating optimal GLDAS hydrological products and uncertainties

Area of Science:

• Hydrology and Earth System Science
• Geophysics and Environmental Monitoring
• Data Assimilation and Modeling

Background:

• Estimating optimal surface state information and uncertainties from multiple land surface models (LSMs) is crucial for accurate hydrological assessments.
• Existing Global Land Data Assimilation System (GLDAS) products require improved methods for uncertainty quantification and optimal value generation.
• Hydrological products like soil moisture, snow water equivalent, and canopy water are key indicators of land surface conditions.

Purpose of the Study:

• To develop and validate a novel approach for generating optimal surface state information and associated uncertainties from GLDAS estimates.
• To assess the performance of the proposed method using key hydrological products (soil moisture, snow water equivalent, canopy water) over the Canadian Prairies.
• To compare the results with established hydrological models and satellite-based observations.

Main Methods:

• Simultaneous application of Förstner and best quadratic unbiased variance component estimators with the least-squares method.
• Focus on soil moisture (SM), snow water equivalent (SWE), and canopy water (CAN) from GLDAS.
• Validation against WaterGAP Global Hydrological Model (WGHM) and Gravity Recovery and Climate Experiment (GRACE) terrestrial water storage anomalies.

Main Results:

• The proposed estimators yielded optimal SM and SWE values with minimal differences from mean values (e.g., 26 mm for SM, 9 mm for SWE).
• Estimated uncertainties for SM, SWE, and CAN varied annually, with total water storage (TWS) uncertainties closely mirroring SM.
• High correlations were achieved between the proposed method's outputs and WGHM (SWE: r=0.97, TWS: r=0.91) and GRACE (TWS: r=0.71, corrected TWS: r=0.81).

Conclusions:

• The novel approach effectively generates optimal surface state information and quantifies uncertainties from GLDAS products.
• The method demonstrates superior performance compared to individual LSMs or their average, with improved RMSE and mean absolute error.
• The results show strong agreement with independent hydrological models and satellite data, highlighting its potential for enhanced land surface monitoring.