The depth distribution of carbon inputs to soil has been unquantified globally, hindering our understanding of belowground carbon dynamics. We synthesize global observational data to infer the allocation of carbon inputs to soil depths down to 2 m, and map depth-specific carbon inputs globally at 1 km resolution. Global average carbon input to the 0–20 cm soil layer is 1.1 Mg C ha–1 yr–1, accounting for >50% of total soil carbon inputs. Across the globe, the depth distribution of carbon inputs shows large variability, and there are relatively more carbon inputs to deeper layers in hotter and drier regions. Edaphic, climatic and topographic properties (in the order of importance) explain >80% of such variability in soil depths; and the direction and magnitude of the influence of individual properties are soil depth- and biome-dependent. Our results provide global benchmarks for prediction of whole-soil carbon profiles across global biomes.

Understanding the partitioning of runoff into baseflow and quickflow is crucial for informed decision-making in water resources management, guiding the implementation of flood mitigation strategies, and supporting the development of drought resilience measures. Methods that combine the physically-based Budyko framework with machine learning (ML) have shown promise in estimating global runoff. However, such ‘hybrid’ approaches have not been used for baseflow estimation. Here, we develop a Budyko-constrained ML approach for baseflow estimation by incorporating the Budyko-based baseflow coefficient (BFC) curve as a physical constraint. We estimate the parameters of the original Budyko curve and the newly developed BFC curve based on 13 climatic and physiographic characteristics using boosted regression trees (BRT). BRT models are trained and tested in 1226 catchments worldwide and subsequently applied to the entire global land surface at a 0.25° grid scale. The catchment-trained models exhibit strong performance during the testing phase, with R2 values of 0.96 and 0.88 for runoff and baseflow, respectively. Results reveal that, on average, 30.3% (spatial standard deviation std=26.5%) of the continental precipitation is partitioned into runoff, of which 20.6% (std=22.1%) is baseflow and 9.7% (std=10.3%) is quickflow. Among the 13 climatic and physiographic characteristics, topography and soil-related characteristics generally emerge as the most important drivers, although significant regional variability is observed. Comparisons with previous datasets suggest that global runoff partitioning is still highly uncertain and warrants further research.

Partitioning evapotranspiration (ET) into evaporation (E) and transpiration (T) is essential for understanding the global hydrological cycle and improving water resource management. However, ET partitioning in various ecosystems is challenging as some assumptions are restricted to certain areas or plant types. Here, we developed a novel ET partitioning method coupling definitions of leaf and ecosystem water use efficiencies (WUEleaf and WUEeco, respectively). We used 25 eddy covariance flux sites for 196 site-years to evaluate T:ET characteristics of seven plant functional types (PFTs) at different spatiotemporal scales. The results indicated the spatiotemporal characteristics of WUEleaf and WUEeco were not consistent, resulting in T:ET variation in the seven PFTs. Deciduous broadleaf forests had the highest mean annual T:ET (0.67), followed by evergreen broadleaf forests (0.63), grasslands (0.52), evergreen needleleaf forests (0.46), and woody savanna (0.41), and C3 croplands had higher T:ET (0.65) than C4 croplands (0.48). The annual mean leaf area index (LAI) explained about 26% of the variation in T:ET, with the trend in T:ET consistent with the known effects of LAI. The overall trends and magnitude of T:ET in this study were similar to different results of ET partitioning methods globally. Importantly, this method improved T:ET estimation accuracy in vegetation-sparse and water-limited areas. Our novel ET partitioning method is suitable for estimating T:ET at various spatiotemporal scales and provides insight into the conversion of WUE at different scales.

It is widely recognized that multi-year drought can induce changes in catchment hydrological behaviors. However, at present, our understanding about multi-year, drought-induced changes in catchment hydrological behaviors and its driving factors at the process level is still very limited. This study proposed a new approach using a data assimilation technique with a process-based hydrological model to detect multi-year drought-induced changes in catchment hydrological behaviors and to identify driving factors for the changes in an unimpaired Australian catchment (Wee Jasper) which experienced prolonged drought from 1997 to 2009. Modelling experiments demonstrated that the multi-year drought caused a significant change in the catchment rainfall-runoff relationship, indicated by significant step changes in the estimated time-variant hydrological parameters SC (indicating catchment active water storage capacity) and C (reflecting catchment evapotranspiration dynamics), whose average values increased 23.4% and 10.2%, respectively, due to drought. The change in the rainfall-runoff relationship identified by the data assimilation method is consistent with that arrived at by a statistical examination. The proposed method provides insights about the drivers of the changes in the rainfall-runoff relationship at the processes level. Declining groundwater and deep soil moisture depleted by persistent evapotranspiration of deep-rooted woody vegetation during drought are the main driving factors for the catchment behaviors change in the Wee Jasper catchment. The new method proposed in this study was found to be an effective technique for detecting both the change of hydrological behaviors induced by prolonged drought and its driving factors at the process level.

The Budyko framework, which can describe a simple but useful partitioning of precipitation under supply and energy limits, is used widely to estimate evapotranspiration (ET). Previous studies have investigated time-variant Budyko functions on annual or interannual scales but seldom on the intra-annual scale. This study used a monthly two-parameter ( κ and y0) Budyko function and three schemes that considered single observations (ET or streamflow (Q)) and dual observations (ET and Q) to assimilate the time-varying parameters using the ensemble Kalman filter method. The study considered the contiguous USA (CONUS) using the Model Parameter Estimation Experiment dataset. The time-varying parameters were explained on the basis of time series analysis and correlation with meteorological data. Three conclusions were as follows. (1) The identified time-varying parameters ( κ and y0) of the Budyko function could effectively simulate ET. (2) The assimilation using only ET observations could identify a plausible set for parameter κ but was inadequate for y0 . (3) Most time-varying parameters exhibited a 12-month period, and the trend and change points detected for Midwest CONUS were related to anthropogenic influences such as extraction and use of groundwater. The findings show that changing environment can be detected by using the proposed time-varying parameters of the Budyko function.