Global Hydrological and Land Surface Models (GHM/LSMs) embody numerous interacting predictors and equations, complicating the diagnosis of primary hydrological relationships. We propose a model diagnostic approach based on Random Forest feature importance to detect the input variables that most influence simulated hydrological processes. We analyzed the JULES, ORCHIDEE, HTESSEL, SURFEX and PCR-GLOBWB models for the relative importance of precipitation, climate, soil, land cover and topographic slope as predictors of simulated average evaporation, runoff, and surface and subsurface runoffs. The machine learning model could reproduce GHM/LSMs outputs with a coefficient of determination over 0.85 in all cases and often considerably better. The GHM/LSMs agreed precipitation, climate and land cover share equal importance for evaporation prediction, and mean precipitation is the most important predictor of runoff. However, the GHM/LSMs disagreed on which features determine surface and subsurface runoff processes, especially with regards to the relative importance of soil texture and topographic slope.

The assessment of soil health has evolved from focusing primary on agricultural productivity to an integrated evaluation of soil biota and biotic processes that impact soil properties. Consequently, soil health assessment has shifted from a predominantly physico-chemical approach to incorporating ecological, biological and molecular microbiology methods. These methods enable a comprehensive exploration of soil microbial community properties and their responses to environmental changes arising from climate change and anthropogenic disturbances. Despite the increasing availability of soil health indicators (physical, chemical, and biological), a holistic mechanistic linkage between indicators and soil functions across multiple spatiotemporal scales has not yet been fully established. This article reviews the state-of-the-art of soil health monitoring, focusing on understanding how soil-microbiome-plant processes contribute to feedback mechanisms and causes of changes in soil properties, as well as the impact these changes have on soil functions. Furthermore, we survey the opportunities afforded by the soil-plant digital twin approach, an integrative framework that amalgamates process-based models, Earth Observation data, data assimilation, and physics-informed machine learning, to achieve a nuanced comprehension of soil health. This review delineates the prospective trajectory for monitoring soil health by embracing a digital twin approach to systematically observe and model the soil-plant system. We further identify gaps and opportunities, and provide perspectives for future research for an enhanced understanding of the intricate interplay between soil properties, soil hydrological processes, soil-plant hydraulics, soil microbiomes, and landscape genomics.

Modelling of the land surface water-, energy-, and carbon balance provides insight into the behaviour of the Earth System, under current and future conditions. Currently, there exists a substantial variability between model outputs, for a range of model types, whereby differences between model input parameters could be an important reason. For large-scale land surface, hydrological, and crop models, soil hydraulic properties (SHP) are required as inputs, which are estimated from pedotransfer functions (PTFs). To analyse the functional sensitivity of widely used PTFs, the water fluxes for different scenarios using HYDRUS-1D was simulated and predictions compared. The results showed that using different PTFs causes substantial variability in predicted fluxes. In addition, an in-depth analysis of the soil SHPs and derived soil characteristics was performed to analyse why the SHPs estimated from the different PTFs cause the model to behave differently. The results obtained provide guidelines for the selection of PTFs in large scale models. The model performance in terms of numerical stability, time-integrated behaviour of cumulative fluxes, as well as instantaneous fluxes was evaluated, in order to compare the suitability of the PTFs. Based on this, the Rosetta, Wösten, and Tóth PTF seem to be the most robust PTFs for the Mualem van Genuchten SHPs and the PTF of Cosby et al. (1984) for the Brooks Corey functions. Based on our findings, we strongly recommend to harmonize the PTFs used in model inter-comparison studies to avoid artefacts originating from the choice of PTF rather from different model structures.

Soil moisture is a critical component in many meteorological, hydrological, and agricultural applications, and understanding its spatial and temporal dynamics is vital for the understanding of these processes. Satellite-based remote sensing offers the ability to synoptically capture this spatiotemporal information over large areas, compared to more site-based in-situ field measurements. In this study, we use Sentinel-1 SAR imagery of the River Thames catchment, United Kingdom, over the period 2015 - 2020. A backscatter normalisation process is applied to account for the use of multiple satellite viewing geometries. A change-detection algorithm utilising backscatter power is then applied to the timeseries, to estimate relative surface soil moisture (rSSM) across the study area. To determine information across the large river watershed, smaller sub-catchments, and intra-field scales, the rSSM time series is replicated at multiple spatial scales (1 km, 500m, 250m, and 100m). Although positive biases are present during the growing season of arable farmland, comparison with rainfall data and in-situ soil moisture probes shows there is good agreement with the temporal cycle of soil moisture. These data are being used to evaluate natural flood management by land use and management across a wide area to better understand relationships between surface wetness and water storage in relation to land cover and underlying geology for the Landwise project (Landwise-NFM.org).

Many remote sensing-based evapotranspiration (RSBET) algorithms have been proposed in the past decades and evaluated using flux tower data, mainly over North America and Europe. Model evaluation across South America has been done locally or using only a single algorithm at a time. Here, we provide the first evaluation of multiple RSBET models, at a daily scale, across a wide variety of biomes, climate zones, and land uses in South America. We used meteorological data from 25 flux towers to force four remote sensing based ET models: Priestley & Taylor Jet Propulsion Laboratory (PT-JPL), Global Land Evaporation Amsterdam Model (GLEAM), Penman-Monteith Mu model (PM-MOD), and Penman-Monteith Nagler model (PM-VI). ET was predicted satisfactorily by all four models, with correlations consistently higher (R²>0.6) for GLEAM and PT-JPL, and PM-MOD and PM-VI presenting overall better responses in terms of PBIAS (-10