We found a match
Your institution may have rights to this item. Sign in to continue.
- Title
Random forest based pedotransfer functions and geostatistical methods for mapping soil moisture retention capacity at 100 m resolution.
- Authors
Toth, Brigitta; Szatmári, Gábor; Takács, Katalin; Laborczi, Annamária; Makó, András; Rajkai, Kálmán; Pásztor, László
- Abstract
3D spatial information on soil hydraulic properties can contribute to increase performance ofhydrological, crop, climate and other environmental models and support climate conscious,water preserving planning of soil management practices. Direct measurement of soilhydraulic properties is costly and time consuming, therefore measurement-based catchmentscale mapping of these soil properties is usually not possible.Our aim was to compare performance of soil water retention maps derived with i)pedotransfer functions (PTFs) and ii) geostatistical methods, both using random forestalgorithm. Maps of saturated water content (THS), field capacity (FC) and wilting point (WP)were prepared at 0–30, 30–60 and 60–90 cm depths at 100 m resolution for the Balatoncatchment (5,775 km2).For the construction of the PTFs we related the soil water retention values to basic soilproperties and other environmental parameters, which are available for the catchment. Theanalysis was performed with random forest method on some 10000 samples of the HungarianDetailed Soil Hydrophysical Database (Hungarian acronym: MARTHA) complemented withinformation on topography, climate, parent material, vegetation and land use. As ageostatistical method we used random forest combined with kriging (RFK). For this directmapping data of 359 soil profiles located in the Balaton catchment was used from theMARTHA dataset.The performance of the PTFs based and RFK maps were similar in case of six out of ninelayers. The PTF significantly outperformed the RFK in mapping THS at 30–60 and 60–90 cmsoil depth. In case of WP at 60–90 cm depth RFK was significantly more accurate. Thedifference between predicted soil hydraulic values derived by RFK and applying PTFswere less than 0.025 cm3cm−3 for 65–86 % of the mapped area. Generalized PTFsapplied on soil property maps and spatial auxiliary data was found to be an efficientalternate for mapping soil moisture retention capacity when the database used for theestablishment of the predictions is hydropedologically similar to that of the mappedarea. The research was supported by the Hungarian National Research, Development andInnovation Office (NRDI) under grants KH124765 and the János Bolyai ResearchScholarship of the Hungarian Academy of Sciences.
- Subjects
SOIL mapping; MAGYAR Tudomanyos Akademia; SOIL moisture; SOIL profiles; SOIL management; LAND use; TOPOGRAPHY
- Publication
Geophysical Research Abstracts, 2019, Vol 21, p1
- ISSN
1029-7006
- Publication type
Article