Unveiling spatially explicit soil nitrogen mineralization potential in Northeast China: A meta-analysis coupled by experimental validation

• Regional-scale analysis reveals mixed kinetics govern soil Nm in over 80 % of cases, with a mean Nmp of 113 mg N kg −1 . • Explainable machine learning identified soil texture and biogeochemical pools as dominant drivers of regional Nmp variation. • Spatial Nmp maps show a ‘north–south high, central low’ soil N pattern in Northeast China, aiding precision N management Paddy rice systems in Northeast China rely on comparatively modest fertilizer nitrogen (N) inputs yet exhibit high N use efficiency, making the soil intrinsic N supply capacity a critical factor for rice production sustainability in this region. N mineralization potential (Nmp) underpins this capacity, but its regional behavior and controls remain poorly constrained. Here we compiled data from 15 peer-reviewed studies, augmented by an incubation experiment on 24 representative paddy soils, to create the first regional dataset of Nmp. Dynamic-kinetic fitting showed that mixed kinetics models best describe over 80 % of incubation curves, and meta -analysis yielded a region-wide mean Nmp of 113 mg N kg −1 . Using deep learning regression, permutation importance and SHapley additive explanations, we identified sand content, CEC, total N, total organic carbon, and growing degree day (GDD) as the dominant predictors of Nmp. The final model, trained jointly on literature and experimental data, explained 59 % of the variance in Nmp (RMSE = 36.4). Upscaling Nmp to a 250 m grid, using the refined deep-learning model and high-resolution soil properties and GDD layers, resolved a pronounced “north–south high, central low” pattern across the Northeast China rice cultivated region. Our integrated workflow—combining meta -analysis, process-based kinetics and interpretable machine learning—provided a transparent, spatially explicit assessment of soil N supply capacity in one of the Chinese most important rice-producing regions. This spatially explicit foundation offers a practical way for site-specific N fertilizer management considering the soil mineralizable-N reserves, upon which context-specific decisions that consider economic and social dimensions can be built.