Quantification of Uncertainty in Surface Energy Fluxes Toward Assessment of Energy Imbalance Using Global Products

Abstract

Accurate assessment of surface energy balance (SEB) closure is critical for understanding global warming and regional climate variability, yet substantial disagreement persists among reanalysis and land data assimilation products across regional scale. This study systematically intercompares SEB closure across the Indian mainland during 1995–2024 using eight gridded data sets: ERA5‐Land, MERRA‐2, NCEP‐NCAR 1, JRA‐3Q, FLDAS and three GLDAS variants (NOAH, VIC and CLSM). Spatially averaged monthly energy imbalance varied substantially across products, with NCEP‐NCAR 1 showing the highest mean value (7.61 W m ⁻² ) and FLDAS and NOAH the lowest (−0.61 and −0.41 W m ⁻² , respectively). Trend analysis indicated significant increases in FLDAS (Sen = +0.039 W m ⁻² /dec, p  = 0.031) and MERRA‐2 (+0.023 W m ⁻² /dec, p  = 0.002) and a significant decline in NOAH (−0.066 W m ⁻² /dec, p  < 0.001), while ERA5‐Land, JRA‐3Q, CLSM and VIC showed no significant trends. Regional analysis revealed that closure quality is strongly influenced by terrain and moisture pattern, with LDAS products and MERRA‐2 performing most consistently with ERA5‐Land showing slightly greater variability and NCEP‐NCAR 1 performing worst in mountainous and humid regions. Validation against IMD observations further points toward ERA5‐Land ( R ²  = 0.6681, r  = 0.8174, RMSE = 30.1119) as the best‐performing product for incoming shortwave radiation, followed by VIC, CLSM, JRA‐3Q, and NOAH while NCEP NCAR 1 ( R ²  = 0.5388, r  = 0.734, RMSE = 64.0844) showed the weakest performance, interpreting that inter‐product agreement does not guarantee observational accuracy.