Downscaling soil hydrological mapping used to predict catchment hydrological response with random forests

Gagkas, Zisis; Lilly, Allan

doi:10.1016/j.geoderma.2019.01.048

Cited by 18 publications

(11 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, the RF model can be recommended as the best model for the prediction of land suitability class for rain-fed wheat and barley in western Iran. This result is comparable to the findings of other researchers who demonstrated the reliable performance of RF [75][76][77][78][79].…”

Section: Comparison Of Different ML Modelssupporting

confidence: 90%

Land Suitability Assessment and Agricultural Production Sustainability Using Machine Learning Models

et al. 2020

View full text Add to dashboard Cite

Land suitability assessment is essential for increasing production and planning a sustainable agricultural system, but such information is commonly scarce in the semi-arid regions of Iran. Therefore, our aim is to assess land suitability for two main crops (i.e., rain-fed wheat and barley) based on the Food and Agriculture Organization (FAO) “land suitability assessment framework” for 65 km2 of agricultural land in Kurdistan province, Iran. Soil samples were collected from genetic layers of 100 soil profiles and the physical-chemical properties of the soil samples were analyzed. Topography and climate data were also recorded. After calculating the land suitability classes for the two crops, they were mapped using machine learning (ML) and traditional approaches. The maps predicted by the two approaches revealed notable differences. For example, in the case of rain-fed wheat, results showed the higher accuracy of ML-based land suitability maps compared to the maps obtained by traditional approach. Furthermore, the findings indicated that the areas with classes of N2 (≈18%↑) and S3 (≈28%↑) were higher and area with the class N1 (≈24%↓) was less predicted in the traditional approach compared to the ML-based approach. The major limitations of the study area were rainfall at the flowering stage, severe slopes, shallow soil depth, high pH, and large gravel content. Therefore, to increase production and create a sustainable agricultural system, land improvement operations are suggested.

show abstract

Section: Comparison Of Different ML Modelssupporting

confidence: 90%

Land Suitability Assessment and Agricultural Production Sustainability Using Machine Learning Models

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Area weighted BFI based on proportions of different HOST soil classes in each subcatchment (Gagkas & Lilly, ).…”

Section: Methodsmentioning

confidence: 99%

Using isotopes to understand the evolution of water ages in disturbed mixed land‐use catchments

Dimitrova‐Petrova

Geris

Wilkinson

et al. 2020

Hydrological Processes

Self Cite

View full text Add to dashboard Cite

Tracer studies have been key to unravelling catchment hydrological processes, yet most insights have been gained in environments with relatively low human impact.We investigated the spatial variability of stream isotopes and water ages to infer dominant flow paths in a~10-km 2 nested catchment in a disturbed, predominantly agricultural environment in Scotland. We collected long-term (>5 years) stable isotope data of precipitation, artificial drainage, and four streams with varying soil and land use types in their catchment areas. Using a gamma model, Mean Transit Times (MTTs) were then estimated in order to understand the spatial variability of controls on water ages. Despite contrasting catchment characteristics, we found that MTTs in the streams were generally very similar and short (<1 year). MTTs of water in artificial drains were even shorter, ranging between 1 to 10 months for a typical field drain and <0.5 to 1 month for a country road drain. At the catchment scale, lack of heterogeneity in the response could be explained by the extensive artificial surface and subsurface drainage, "short-circuiting" younger water to the streams during storms. Under such conditions, additional intense disturbance associated with highway construction during the study period had no major effect on the stream isotope dynamics. Supplementary short-term (~14 months) sampling of mobile soil water in dominant soil-land use units also revealed that agricultural practices (ploughing of poorly draining soils and soil compaction due to grazing on freely draining soils) resulted in subtle MTT variations in soil water in the upper profile. Overall, the isotope dynamics and inferred MTTs suggest that the evolution of stream water ages in such a complex human-influenced environment are largely related to near-surface soil processes and the dominant soil management practices. This has direct implications for understanding and managing flood risk and contaminant transport in such environments. K E Y W O R D Sagricultural catchment, Anthropocene hydrology, artificial drainage, gamma model, land use management, soil water isotopes, transit times, water isotopes

show abstract

“…The drawbacks are that they are complex to interpret and cannot extrapolate outside the training range. Given their advantages, this algorithm is particularly suited for the estimation of spatial variables such as soil properties (Booker and Woods, 2014;Hengl et al, 2018;Gagkas and Lilly, 2019;Stein et al, 2021). In the present work, an RF model is generated to estimate the values of the A parameter of the SMA model, representing the soil water holding capacity, with the properties of the 5 × 5 km grid cells, namely altitude, land cover, mean annual precipitation, temperature and PET, using random forests.…”

Section: Regionalization With Random Forestsmentioning

confidence: 99%

Estimating soil moisture conditions for drought monitoring with random forests and a simple soil moisture accounting scheme

Tramblay

Quintana-Seguí²

2022

Nat. Hazards Earth Syst. Sci.

View full text Add to dashboard Cite

Abstract. Soil moisture is a key variable for drought monitoring, but soil moisture measurements networks are very scarce. Land-surface models can provide a valuable alternative for simulating soil moisture dynamics, but only a few countries have such modelling schemes implemented for monitoring soil moisture at high spatial resolution. In this study, a soil moisture accounting model (SMA) was regionalized over the Iberian Peninsula, taking as a reference the soil moisture simulated by a high-resolution land-surface model. To estimate the soil water holding capacity, the sole parameter required to run the SMA model, two approaches were compared: the direct estimation from European soil maps using pedotransfer functions or an indirect estimation by a machine learning approach, random forests, using as predictors altitude, temperature, precipitation, potential evapotranspiration and land use. Results showed that the random forest model estimates are more robust, especially for estimating low soil moisture levels. Consequently, the proposed approach can provide an efficient way to simulate daily soil moisture and therefore monitor soil moisture droughts, in contexts where high-resolution soil maps are not available, as it relies on a set of covariates that can be reliably estimated from global databases.

show abstract

Downscaling soil hydrological mapping used to predict catchment hydrological response with random forests

Cited by 18 publications

References 41 publications

Land Suitability Assessment and Agricultural Production Sustainability Using Machine Learning Models

Land Suitability Assessment and Agricultural Production Sustainability Using Machine Learning Models

Using isotopes to understand the evolution of water ages in disturbed mixed land‐use catchments

Estimating soil moisture conditions for drought monitoring with random forests and a simple soil moisture accounting scheme

Contact Info

Product

Resources

About