Continuous Monitoring of Albufera Wetland (Spain) by SAR and multispectral satellite data

Freshwater environments have undergone important changes in recent years; the various pressures on land use, the effects of climate change and the over-exploitation of water resources are significantly affecting water resource availability and biodiversity in these fragile ecosystems. Constant monitoring of freshwater environments is crucial for their management and protection. This can be obtained by satellite remote sensing that is a powerful, cost-efficient and still under-exploited monitoring tool. The main idea of the research work is to investigate how different kind of satellite data can be exploited to achieve a better description of freshwater environments at adequate space scales and with high temporal resolution. The study-case is the Albufera wetland in Spain, one the most important protected areas in Europe for the presence of many migratory birds species. The Albufera Natural Park includes a lake surrounded by rice fields irrigated by periodic flooding that offer in some periods of the year suitable habitats for many species of birds and others water-related organisms, such as macroinvertebrates and fishes.

The continuous monitoring of the flooded area extension is a prerequisite to understand the link between the water presence and habitat availability. The study combines observation from multiple optical and synthetic aperture radar (SAR) sensors with spatial resolution between 3 and 30 m. Acquisitions from Landsat-8, Sentinel-2 satellites were used in the optical and infrared bands. The revisit time ranges between 5 and 10 days even if, in case of cloud cover, the revisit time increases consistently. An unsupervised classification method, based on the application of a threshold, was used, in particular multispectral indexes such as MNDWI, NDWI and NDVI were calculated. The NDWI and MNDWI indexes allowed to identify the presence of limpid and turbid water in the October-May period, while in the May-September period the NDVI was used to identify rice plants and therefore indirectly estimate the possible presence of water below the canopy.  In order to increase the time resolution, also in periods with frequent cloud presence, Sentinel-1A and 1B and COSMO-SkyMed SAR images were also used. The Sentinel-1 constellation operates in C band with time resolution of about 5 days; while COSMO-SkyMed operates in X bands with time resolution of about 10 days. The images were processed with both unsupervised and supervised classification methods. The information obtained from images processing were compared with very high-resolution (0.30 m and 0.50 m) satellite images and field measurements in order to validate and calibrate the classification method. The classification obtained with multispectral and SAR data were also cross-validated, providing very satisfactory results. Combination of different satellite data allowed for a significant increase of the temporal resolution of the observations, also in presence of cloud cover. The result of the study showed the dynamic of flooding-drying of the wetland and the flooding duration in different areas of the Albufera Park. This dataset is extremely useful for the optimization of wetland management and for further investigation on the link between flooding duration and habitat availability.