Repeat-pass interferometric coherence as a SAR tool for crop monitoring
- Autores: Arturo Villarroya Carpio
- Directores de la Tesis: Juan Manuel López Sánchez (dir. tes.)
- Lectura: En la Universitat d'Alacant / Universidad de Alicante ( España ) en 2025
- Idioma: inglés
- Número de páginas: 154
- Tribunal Calificador de la Tesis: Irena Hajnsek (presid.), Jesús Selva Vera (secret.), José Luis Álvarez Pérez (voc.)
- Programa de doctorado: Programa de Doctorado en Informática por la Universidad de Alicante
- Materias:
- Enlaces
- Tesis en acceso abierto en: RUA
- Resumen
The use of remote sensing data from Earth Observation satellites offers valuable tools for agricultural monitoring, as a periodic, automatic, global and non-destructive way to obtain information about vegetated areas. It allows for data acquisition over wide areas with a high temporal resolution (to detect changes in the phenological stage of vegetation) and, when necessary, over a long time series (to capture the variability of climatic conditions).
The use of active radar sensors offers unique advantages due to its improved spatial resolution over wide areas compared to passive microwave remote sensing and its generally better temporal coverage than optical imagery, on account of not being affected by the presence of clouds and the day/night cycle (Steele-Dunne et al., 2017, Mandal et al., 2021).
Besides the observables derived from measured backscattered intensity (backscattering coefficients and polarimetric descriptors), radar satellites also provide access to interferometric data. In InSAR pairs of images are combined to produce phase measurements related to the scene vertical dimension and other scene properties (Bamler and Hartl, 1998). A key interferometric observable is the interferometric coherence, which is employed as a direct measure on the quality of the associated interferometric phase, and hence of the derived products. Diverse factors influence the interferometric coherence (Zebker and Villasenor, 1992) including the changes in the surface along time. Therefore, coherence is by itself a valuable measurement with sensitivity to the properties and evolution: growth, structural changes, wind-induced movements of branches and leaves and changes in vegetation water content of the crop in the scene. Coherence is also sensitive to soil moisture and vegetation water stress, which are difficult to measure on a great scale without the use of satellite data.
This body of work has focused on evaluating the use of repeat-pass synthetic aperture radar interferometric coherence as a tool for crop monitoring. The studies employed C-band data from Sentinel-1 and a ground-based experiment, as well as X-band data from TerraSAR-X, TanDEM-X, and PAZ, comparing them with optical vegetation indices and hydrometeorological variables.
The initial studies demonstrated that the repeat-pass coherence amplitude of Sentinel-1 with a 6-day interval can be used as a vegetation index for crop monitoring. VV coherence was particularly useful for describing crop evolution, while VH coherence performed better for less dense crops. Furthermore, coherence and backscatter coefficients provide complementary information, suggesting their combined use.
On the other hand, bias removal in coherence, as well as separating different sources of decorrelation, are not necessary steps, as they were found to offer no significant improvements when working with VV and VH individually. Therefore, measured coherence can be exploited directly without additional processing, provided that the ENL (Equivalent Number of Looks) is sufficiently high.
By combining time series from TerraSAR-X, TanDEM-X, and PAZ, X-band coherence time series with short temporal baselines (as brief as 4 and 7 days) were analysed. Although X-band was expected to have limitations for repeat-pass interferometry in agricultural areas, the results showed high sensitivity during the early growth stages of short crops (e.g., broccoli, onion, and sugar beet). In these cases, the increasing fraction of vegetation cover gradually acted as a source of decorrelation. For taller crops, coherence saturated more rapidly, with C-band exhibiting better sensitivity.
For the first time, a new radar vegetation index for co-polarised data, called coRVI, was used. When compared to NDVI, it produced results comparable to those obtained with coherence, suggesting that these two radar-derived products could be used complementarily when available.
Finally, a set of ground-measured, sub-daily SAR data was analysed. The acquisition of SAR data throughout the day was valuable for observing SAR interferometry's sensitivity to short-term changes in soil moisture, temperature, wind, and other factors related to daily vegetation cycles. The study of the decorrelation rate revealed a periodic behaviour in coherence, possibly linked to daily cycles of soil moisture and vegetation water dynamics. This indicates that coherence is sensitive to the water cycle, although separating individual contributions was not possible.
Additionally, the impact of changes in incidence angle and acquisition timing on coherence time series was evaluated. The findings concluded that acquisitions during the night or early morning are ideal for preserving high coherence. These observations may be relevant for a potential future mission based on a SAR system with sub-daily acquisition configurations.
Bamler, R., Hartl, P. Synthetic aperture radar interferometry. Inverse Problems 1998;14:R1 54. DOI: 10.1088/0266-5611/14/4/001.
Mandal, D., Bhattacharya, A., Rao, Y.S.. Radar remote sensing for crop biophysical parameter estimation. Springer, 2021. DOI: 10.1007/978-981-16-4424-5.
Steele-Dunne, S.C., McNairn, H., Monsivais-Huertero, A., Judge, J., Liu, P.W., Papathanassiou, K. Radar remote sensing of agricultural canopies: A review. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 2017;10(5):2249-2273. DOI: 10.1109/JSTARS.2016.2639043 Zebker, H.A., Villasenor, J.D. Decorrelation in interferometric radar echoes. IEEE Transactions on Geoscience and Remote Sensing 1992;30(5):950-959. DOI: 10.1109/36.175330.
