Retrieving crop parameters based on tandem ERS 1/2 interferometric coherence images Xavier Blaes, Pierre Defourny * Department of Environmental Sciences and Land Use Planning, Universite ´ Catholique de Louvain, Croix du Sud 2/16, B-1348 Louvain-la-Neuve, Belgium Received 12 March 2003; received in revised form 1 August 2003; accepted 18 August 2003 Abstract One-day interval coherence images derived from pairs of ERS SAR tandem acquisitions are suitable for crop monitoring. Coherence images were analyzed and compared to field measurements of four crops, i.e., winter wheat, sugar beet, potato and maize, taken during the satellite overpass. First, the sensitivity of the coherence to the plant height and the canopy cover was statistically investigated. Regression analyses were computed and the coefficients of determination (R 2 ) ranged from 0.64 to 0.92. The shape of these relationships varied according to the geometric factors which are crop-type dependent. A prediction model of the wheat height was calculated and was able to estimate the plant height with a mean absolute error of approximately 7 cm. While this high performance obtained on the average matched the observed range of the field height for a given date, it was not sufficient for monitoring at the field level. However, such a performance level may meet the information requirements for an operational crop monitoring system at the regional level, which includes a much larger diversity of growing conditions. Moreover, the soil roughness change associated with the sowing practices that occurred between the two tandem acquisitions strongly decreased the coherence signal. This dataset also indicated that variation in the soil moisture influenced the backscattering coefficient more than it influenced the coherence signal. This result enhanced the InSAR coherence potentialities to estimate the crop parameters during the growing season. D 2003 Elsevier Inc. All rights reserved. Keywords: SAR interferometry; Interferometric coherence; Crop monitoring; Biophysical parameters retrieval 1. Introduction The use of radar remote sensing for crop mapping and retrieving biophysical parameters has been extensively stud- ied in the last few decades. SAR sensors have an inherent advantage over optical sensors due to their ability to penetrate cloud cover leading to nearly all-weather acquisi- tion capabilities. Furthermore, the ERS-1/2 satellite repeat orbits provided SAR imagery for interferometric analysis. Radar interferometry is a technique that extracts three- dimensional information of the observed surface by using the phase content derived from a couple of images. The interferometric phase measures the path length difference between the target and the two sensor locations from which a three-dimensional position of the images resolution ele- ments (i.e., height maps) is derived. The interferometric correlation, or coherence, measures the variance of the interferometric phase estimate. It decreases with increasing system noise, volume scattering and temporal changes and therefore contains thematic information. The ERS 1/2 tandem acquisition period starting in April 1995 was a unique opportunity for the application of space- borne SAR interferometry. This tandem opportunity provid- ed 1-day interval SAR acquisition capabilities from two identical C-band instruments. ERS SAR interferometry had already been validated using ERS-1 data with a 3- or 35-day repeat cycle. The 1-day offset offered by the tandem mission increased the probability of having a high coherence be- tween the acquired data. Different types of temporal change can be distinguished with repeat-pass SAR interferometry (Wegmu ¨ller & Werner, 1995). The first is a coherent backscatter intensity change, i.e., high correlation with significant backscatter intensity change, resulting from a stable scatterers geometry in com- bination with a permittivity change. This is observed as a re- sult of changing soil moisture, freezing and thawing. Smaller changes occur due to variations of soil moisture and plant water content. Secondly, scatterer geometry change causes a loss of coherence. Mechanical cultivation, e.g., ploughing, 0034-4257/$ - see front matter D 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.rse.2003.08.008 * Corresponding author. Fax: +32-10-47-88-98. E-mail address: defourny@enge.ucl.ac.be (P. Defourny). www.elsevier.com/locate/rse Remote Sensing of Environment 88 (2003) 374 – 385