Analysis of the radar temporal coherence at C band over an olive orchard in semi-arid region

In recent decades, climate change has led to a sharp increase in water demand. Particularly in agriculture, this has put a great strain on already scarce water resources, increased the need for irrigation water, and led to overuse of groundwater. Therefore, sustainable management of water resources while maintaining good agricultural yield by monitoring crop water status is necessary for sustainable and rational management of these resources, especially in arid and semi-arid regions. For this purpose, a detailed knowledge of the different processes describing the diurnal water cycle of plants in a large area is essential. However, micrometeorological or physiological experimental measurements and their partitioning are laborious to perform and not very representative of large areas.

In this regard, remote sensing is a particularly suitable tool for monitoring agricultural areas because of its global and repeated observation. Several studies have highlighted the sensitivity of radar data to vegetation water content especially over the rainforest with spatial scatterometers that observe differences between morning and evening acquisitions. On the other hand, in situ radar experiments with high temporal frequency have made it possible to analyze radar responses over tropical and boreal forests.

This study relates to a similar experiment conducted on an olive orchard located in the semi-arid Mediterranean region of Chichaoua in central Morocco. It allows the acquisition of in situ C-band radar measurements in crop fields, which are acquired continuously, from a tower-based radar system, with a time step of 15 minutes.

The temporal evolution of the interferometric coherence r is analyzed on different baselines Dt, ranging from 15 minutes to 30 days, for the main physiological stages of the olive tree. Four different two-month periods, from December 2020 to November 2022, are chosen as the main physiological stages based on field observations.

The obtained results of r, especially at 15-min min-steps, show a global behavior similar to that observed in tropical and boreal forests: high values (r ≈1) are observed during the night (weak wind, vegetation resting), then a decrease/increase during the day mainly anti-symetric to the wind cycle. As over boreal and tropical forest, a decrease in r is observed before the wind picks up, with is time coincident with sap flows and ETR variations, traducing its sensitivity to water plant content.

Results show that over olive orchard, the r diurnal cycle is less marked than over boreal and tropical forests, due to lower ETR rates and certainly due to a significant soil contribution over this less dense vegetation layer. Furthermore, r values decrease when temporal baselines increase, but values are still meaningful for Dt = 6 days (r = 0.3 compared to 0.6 for Dt = 15 min. for the summer period), available with Sentinel-1 missions.

The present study provides particularly interesting results confirming the sensitivity of C-band coherence to vegetation water status, especially in the early morning. Further work needs to be pursued to verify if we are able to detect the water stress of these plants in semi-arid areas such as Chichaoua through coherence.