Combining physical-based models and satellite images for the spatio-temporal assessment of soil infiltration capacity

Abstract

The performance of managed artificial recharge (MAR) facilities by means of surface ponds (SP) is controlled by the temporal evolution of the global infiltration capacity $I_{c}$ of topsoils. Cost-effective maintenance operations that aim to maintain controlled infiltration values during the activity of the SP require the full knowledge of the spatio-temporal variability of $I_{c}$. This task is deemed uncertain. The natural reduction in time of $I_{c}$ depends on complex physical, biological and chemical reactions that clog the soil pores and has been observed to decay exponentially to an asymptotic non-zero value. Moreover, the relative influence of single clogging processes depend on some initial parameters of the soil, such as the initial infiltration capacity ($I_{c,0}$). This property is also uncertain, as aquifers are typically heterogeneous and scarcely characterized in practical situations. We suggest a method to obtain maps of $I_{c}$ using a geostatistical approach, which is suitable to be extended to engineering risk assessment concerning management of SP facilities. We propose to combine geostatistical inference and a temporally-lumped physical model to reproduce non-uniform clogging in topsoils of a SP, using field campaigns of local and large scale tests and additionally by means of satellite images as secondary information. We then postulate a power-law relationship between the parameter of the exponential law, $\lambda$, and $I_{c,0}$. It is found that calibrating the two parameters of the power law model it is possible to fit the temporal evolution of total infiltration rate at the pond scale in a MAR test facility. The results can be used to design appropriate measures to selectively limit clogging during operation, extending the life of the infiltration pond.

Full Document