Digital soil mapping of available water capacity for metropolitan France
Résumé
Soil hydraulic characteristics are often required for modeling hydrological processes, crop pro-
ductivity, biogeochemical cycles, and movement of pollutants. The objective of this study was
producing a digital soil map of available water capacity (AWC) for metropolitan France following
GlobalSoilMap (GSM) specifications. The digital soil mapping approach consisted in predicting
sand and clay for the six GSM depth intervals and then applying pedotransfer functions (PTF)
for estimating AWC (cm3 cm-3) by layer. We used the French SOLHYDRO database to develop
continuous PTF for soil moisture at field capacity (pF<2.0) and permanent wilting point (pF<4.2) using sand and clay as predictor variables. The calibration dataset comprised profiles from the French Soil Survey program. The independent dataset consisted of profile data from the RMQS, which is distributed on a 16 km x16 km grid. Particle size measurements were processed with equal-area spline functions, and later transformed with the additive log ratio transformation (alr).
Regression models were fitted for sandalr and clayalr, followed by cokriging of the residuals using
a linear model or coregionalization. We considered the uncertainty of the input data by sampling
from the conjoint distribution of sandalr and clayalr in a sequential Gaussian simulation approach,
for producing uncertainty estimates for AWC by depth interval. Total AWC for the profile was
calculated by aggregation to the predicted soil depth (Lacoste et al., 2016) to a maximum of 2
m. The lack of georeferenced observations for AWC did not allow us to validate both the spatial
models and the PTF estimates together, and therefore we validated separately the performance
of the PTF with cross-validation (R2<0.67 – 0.84, RMSE <0.027 – 0.043 cm3 cm-3) and the
spatial predictions of sand and clay.
Lacoste, M., Mulder, V. L., Martin, M. P., Arrouays, D., 2016. Evaluating large-extent spatial
modeling approaches: a case study for total soil depth for France. Geoderma Regional, 7, 137 –
152.
Origine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...