Landscape metrics as predictors of water-related ecosystem services: Insights from hydrological modeling and data-based approaches applied on the Arno River Basin, Italy.

The study addresses the challenge of integrating complex landscape-hydrological interactions into predictive models for improved water resource management. The aim is to investigate the effectiveness of landscape metrics-quantitative indices measuring landscape composition and configuration-as predictors of WES in the Arno River Basin, Italy. Utilizing two hydrological models alongside a random forest algorithm, we assessed spatial and temporal variations in water yield, runoff, and groundwater recharge. The findings indicate that landscape metrics derived from high-resolution land use data significantly impact WES outcomes. Specifically, the models demonstrated average landscape metric importances of 16.8 % for spatial and 17.8 % for temporal predictions concerning runoff. For water yield, these averages were 32.9 % spatially and 43.5 % temporally, while groundwater modeling showed importances of 14.09 % spatially and 33.8 % temporally. Key landscape metrics identified include the core area index for broad-leaved forests and the perimeter-to-area ratio for non-irrigated agricultural areas as critical spatial and temporal predictors of water yield and groundwater recharge. Thresholds were observed, indicating landscape configurations that minimize hydrological variability. For instance, runoff variation is minimal when the landscape exhibits high forest fragmentation (over 1000 coniferous patches), low aggregation (aggregation index <75), and reduced connectivity (cohesion index under 80). Similarly, groundwater variation is minimized with decreased boundary length of vegetation patches (perimeter-to-area ratio <0.8), agricultural lands (perimeter-to-area ratio under 1), and the presence of low core agricultural areas (core area index above 8). The identified thresholds could inform land-use policies, such as targeted afforestation or crop diversification strategies, to optimize WES provision.

Large Temporal Variations of Functional Properties of Outdoor Equestrian Arena Surfaces and a New Concept of Evaluating Reactivity With Light Weight Deflectometer Settlement Curves.

Sports physiological properties of ten sand or sand-mineral outdoor arenas, five with vertical drainage systems and five with an ebb and flow like system were assessed over a period of 8 weeks. For each arena, the riding zone was spatially delineated, nine locations at medium to intensely used zones were selected by simple random sampling and used along the whole measurement period. A total of 72 values for the dynamic deflection modulus (Evd), attenuation (s/v), settlement (s) and moisture content (Vol %) were analyzed for each arena. A novel technique to analyze the settlement curves of the light weight deflectometer (LWD) to describe reactivity of the footing surface was introduced. Statistical testing was done by linear mixed models. Three of the five arenas with a vertical watering system were judged to be hard (Evd > 20 MN/m2), whereas all five arenas with an ebb and flow like watering systems were medium hard (Evd = 10-20 MN/m2) over the entire 8 weeks. Significant (P < .01) temporal differences in Evd, s/v and moisture were demonstrated for both watering systems; however, the spatial and temporal variations were much lower with the ebb-flow system. Temporal consistency in the parameters over the test weeks appeared to be a criterion for stability of the arena surface. The analysis of the settlement curves of the LWD showed that the slope symmetry has a large potential to describe the restoration of the energy of an equestrian surface than only the settlement, which requires further validation.

Mapping the geogenic radon potential for Germany by machine learning.

The radioactive gas radon (Rn) is considered as an indoor air pollutant due to its detrimental effects on human health. In fact, exposure to Rn belongs to the most important causes for lung cancer after tobacco smoking. The dominant source of indoor Rn is the ground beneath the house. The geogenic Rn potential (GRP) - a function of soil gas Rn concentration and soil gas permeability - quantifies what "earth delivers in terms of Rn" and represents a hazard indicator for elevated indoor Rn concentration. In this study, we aim at developing an improved spatial continuous GRP map based on 4448 field measurements of GRP distributed across Germany. We fitted three different machine learning algorithms, multivariate adaptive regression splines, random forest and support vector machines utilizing 36 candidate predictors. Predictor selection, hyperparameter tuning and performance assessment were conducted using a spatial cross-validation where the data was iteratively left out by spatial blocks of 40 km*40 km. This procedure counteracts the effect of spatial auto-correlation in predictor and response data and minimizes dependence of training and test data. The spatial cross-validated performance statistics revealed that random forest provided the most accurate predictions. The predictors selected as informative reflect geology, climate (temperature, precipitation and soil moisture), soil hydraulic, soil physical (field capacity, coarse fraction) and soil chemical properties (potassium and nitrogen concentration). Model interpretation techniques such as predictor importance as well as partial and spatial dependence plots confirmed the hypothesized dominant effect of geology on GRP, but also revealed significant contributions of the other predictors. Partial and spatial dependence plots gave further valuable insight into the quantitative predictor-response relationship and its spatial distribution. A comparison with a previous version of the German GRP map using 1359 independent test data indicates a significantly better performance of the random forest based map.

Random forest as a generic framework for predictive modeling of spatial and spatio-temporal variables.

Random forest and similar Machine Learning techniques are already used to generate spatial predictions, but spatial location of points (geography) is often ignored in the modeling process. Spatial auto-correlation, especially if still existent in the cross-validation residuals, indicates that the predictions are maybe biased, and this is suboptimal. This paper presents a random forest for spatial predictions framework (RFsp) where buffer distances from observation points are used as explanatory variables, thus incorporating geographical proximity effects into the prediction process. The RFsp framework is illustrated with examples that use textbook datasets and apply spatial and spatio-temporal prediction to numeric, binary, categorical, multivariate and spatiotemporal variables. Performance of the RFsp framework is compared with the state-of-the-art kriging techniques using fivefold cross-validation with refitting. The results show that RFsp can obtain equally accurate and unbiased predictions as different versions of kriging. Advantages of using RFsp over kriging are that it needs no rigid statistical assumptions about the distribution and stationarity of the target variable, it is more flexible towards incorporating, combining and extending covariates of different types, and it possibly yields more informative maps characterizing the prediction error. RFsp appears to be especially attractive for building multivariate spatial prediction models that can be used as "knowledge engines" in various geoscience fields. Some disadvantages of RFsp are the exponentially growing computational intensity with increase of calibration data and covariates and the high sensitivity of predictions to input data quality. The key to the success of the RFsp framework might be the training data quality-especially quality of spatial sampling (to minimize extrapolation problems and any type of bias in data), and quality of model validation (to ensure that accuracy is not effected by overfitting). For many data sets, especially those with lower number of points and covariates and close-to-linear relationships, model-based geostatistics can still lead to more accurate predictions than RFsp.