Global Prediction of Soil Saturated Hydraulic Conductivity Using Random Forest in a Covariate‐Based GeoTransfer Function (CoGTF) Framework

Gupta, Surya; Lehmann, Peter; Bonetti, Sara; Papritz, Andreas; Or, Dani

doi:10.1029/2020ms002242

Cited by 48 publications

(26 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The first model (Rosetta 3; Zhang & Schaap, 2017) is a neural network trained with data mainly from the US and Europe, predicting

K_{sat}

as a function of bulk density, sand, clay, and silt content. The second model (CoGTF; Gupta, Lehmann, et al., 2021) is a random forest model trained with data from all continents including tropical regions (Gupta, Hengl, et al., 2021). The results shown in Figure 4 depict significantly higher conductivity values using CoGTF (median

K_{s a t}

value for Brazil of 1.0 m/day compared to 0.2 m/day for Rosetta 3) because (a) it was trained with values from tropical regions and (b) its machine learning algorithm identified topographic parameters as important environmental factors determining

K_{sat}

(Gupta, Hengl, et al., 2021) that were not included in Rosetta 3.…”

Section: Resultsmentioning

confidence: 99%

“…Commonly used PTFs are often trained using soil samples from arable lands in temperate regions (Or, 2019) that underrepresent kaolinite‐dominated regions that cover 16% of the land surface (∼20 million km 2 ; see the map of dominant clay mineral type in Figure ). In addition, these PTFs relate the SHMPs to basic soil properties only and do not incorporate the wealth of information contained in local environmental covariates that reflect soil formation processes (Gupta, Lehmann, et al., 2021). The correction of SHMP using clay mineral‐specific PTFs and spatial distribution of environmental covariates paints a significantly different picture for tropical regions parameterized with standard PTFs, often neglecting soil strength (critical for soil erosion rates, carbon transport to oceans, and natural hazard prediction) and underestimating saturated hydraulic conductivities

K_{sat}

in tropical soils.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Clays Are Not Created Equal: How Clay Mineral Type Affects Soil Parameterization

Lehmann

Leshchinsky

Gupta

et al. 2021

Geophysical Research Letters

Self Cite

View full text Add to dashboard Cite

Soil clay minerals dominate the colloidal fraction and specific surface area ( E SSA ) of soil. Clay minerals exhibit a wide range of microstructures and hydration responses that affect soil macroscopic hydraulic, chemical, and mechanical properties. We focus on kaolinite and smectite as end members of the clay mineral family due to their contrasting surface areas, differences in activity (shrink-swell behavior), high abundance in natural soils, and their general separation among climatic regions (e.g., kaolinite dominates in tropical soils that receive high values of mean annual precipitation, see Figure 1). Due to differences in their basic building blocks and properties, kaolinite comprises tightly bound clay platelets that form large tactoids (aggregated stacks of platelets) and is considered inactive, whereas smectite is considered a highly active clay mineral (Jefferson & Smalley, 1997;Skempton, 1953).In many land surface and Earth system models (the latter term includes simulations of aspects of the Earth system affecting the occurrence of natural hazards), information on 'clay' is often used in an undifferentiated manner as soil 'clay content' (defined as the mass fraction of soil particles smaller than 2 μm in diameter) to derive spatially distributed soil hydraulic and mechanical properties using pedotransfer functions (Gutmann & Small, 2007;Van Looy et al., 2017). The large differences in microstructure and hydration response between kaolinite and smectite, along with their remarkable spatial segregation in climatic regions, have a large effect on the regional clay mineral-dependent soil hydraulic and mechanical properties. We seek to

show abstract

“…The first model (Rosetta 3; Zhang & Schaap, 2017) is a neural network trained with data mainly from the US and Europe, predicting

K_{sat}

K_{s a t}

K_{sat}

(Gupta, Hengl, et al., 2021) that were not included in Rosetta 3.…”

Section: Resultsmentioning

confidence: 99%

K_{sat}

in tropical soils.…”

Section: Discussionmentioning

confidence: 99%

Clays Are Not Created Equal: How Clay Mineral Type Affects Soil Parameterization

Lehmann

Leshchinsky

Gupta

et al. 2021

Geophysical Research Letters

Self Cite

View full text Add to dashboard Cite

show abstract

“…The soil physical parameters are determined using a global pedotransfer model, using only texture as input. New, advanced pedotransfer functions have emerged in recent years, using not only texture, but also climatology and land use as predictors (Gupta et al, 2021). As soil moisture is at the basis of many processes in LSM, incorporating these PTF seems the logical new step forward in LSM (Fatichi et al, 2020).…”

Section: Soil Moisturementioning

confidence: 99%

Local scale evaluation of the simulated interactions between energy, water and vegetation in land surface models

Pue

Barrios

Liu

et al. 2022

Preprint

View full text Add to dashboard Cite

Abstract. The processes involved in the exchange of water, energy and carbon in terrestrial ecosystems are strongly intertwined. To accurately represent the terrestrial biosphere in land surface models (LSM), the intrinsic coupling between these processes is required. Soil moisture and leaf area index are two key variables at the nexus of water, energy and vegetation. Here, we evaluated three LSM (ISBA, ORCHIDEE and a diagnostic model, based on the LSA SAF algorithms) in their ability to simulate the latent heat flux (LE) and gross primary production (GPP) coherently, and their interactions through leaf area index (LAI) and soil moisture. The models were validated using in situ eddy covariance observations, soil moisture measurements and remote sensed LAI. It was found that the diagnostic model performed consistently well, regardless land cover, whereas important shortcomings of the prognostic models were revealed for in herbaceous/dry sites. Despite their different architecture and parametrization, ISBA and ORCHIDEE shared some key weaknesses. In both models, LE and GPP were found to be oversensitive to drought stress. Though the simulated soil water dynamics could be improved, this was not the main cause of errors in the surface fluxes. Instead, these errors were strongly correlated to errors in LAI. The simulated phenological cycle in ISBA and ORCHIDEE was delayed compared to observations, and failed to capture the observed seasonal variability. The feedback mechanism between GPP and LAI (i.e. the biomass allocation scheme) was identified as a key element to improve the intricate coupling between energy, water and vegetation in LSM.

show abstract

“…Field and laboratory tests are costly and require considerable times 15 . For his reason, indirect estimation methods have been developed, such as theoretical equations 3 , 16 , pedotransfer functions 15 , 17 , 18 , and machine learning methods 19 – 21 . Machine learning methods, which can have either a regression or classification character, can be an essential tool for assessing the sensitivity of variables influewnced by complex relationships, as is the case for k sat .…”

Section: Introductionmentioning

confidence: 99%

Statistical evaluation of testing conditions on the saturated hydraulic conductivity of Brazilian lateritic soils using artificial intelligence approaches

Souza

Pereira

Mendes

et al. 2022

Sci Rep

View full text Add to dashboard Cite

The saturated hydraulic conductivity, ksat, is a crucial variable to describe the hydromechanical behavior of soils. The value of ksat of lateritic soils that are typically found in tropical regions is highly affected by the soil’s structure, void ratio, and fine particle aggregation. As a result, the determination of ksat in the field or in the laboratory is complex and involves greater variability, depending on the type of test and on the spatial location of sampling. This paper presents a study of ksat values of lateritic soils, analyzing them using Statistic, Multilayer Perceptron Artificial Neural Networks (ANN) and Decision Trees (CHAID). This study aims to support decision-making regarding the type of test and depth chosen for sampling in laterite soils and understanding the factors influencing the permeability of such soils. An extensive literature review on the ksat values of lateritic soils was performed, providing data for the establishment of a database comprise of 722 registries. According to agronomic and geotechnical soil classifications, the Brazilian lateritic soils presents a “moderate” hydraulic conductivity. A significant variation of permeability values along the depth was identified, particularly for depths between 0.1 and 0.2 m. Regarding the importance of testing variables, the ANN indicated a high dependency on the type of test. The decision tree divided field test and laboratory test automatically, inferring the relevance of the type of test to the determination of ksat.

show abstract

Global Prediction of Soil Saturated Hydraulic Conductivity Using Random Forest in a Covariate‐Based GeoTransfer Function (CoGTF) Framework

Cited by 48 publications

References 53 publications

Clays Are Not Created Equal: How Clay Mineral Type Affects Soil Parameterization

Clays Are Not Created Equal: How Clay Mineral Type Affects Soil Parameterization

Local scale evaluation of the simulated interactions between energy, water and vegetation in land surface models

Statistical evaluation of testing conditions on the saturated hydraulic conductivity of Brazilian lateritic soils using artificial intelligence approaches

Contact Info

Product

Resources

About