Abstract
In this paper, we present a new approach to the fusion of Sentinel 1 (S1) and Sentinel 2 (S2) data for land cover mapping. The proposed solution aims at improving methods based on Sentinel 2 data, that are unusable in case of cloud cover. This goal is achieved by using S1 data to generate S2-like segmentation maps to be used to integrate S2 acquisitions forbidden by cloud cover. In particular, we propose for the first time in remote sensing a multi-temporal W-Net approach for the segmentation of Interferometric Wide swath mode (IW) Sentinel-1 data collected along ascending/descending orbit to discriminate rice, water, and bare soil. The quantitative assessment of segmentation accuracy shows an improvement of 0.18 and 0.25 in terms of accuracy and F1-score by applying the proposed multi-temporal procedure with respect to the previous single-date approach. Advantages and disadvantages of the proposed W-Net based solution have been tested in the National Park of Albufera, Valencia, and we show a performance gain in terms of the classical metrics used in segmentation tasks and the computational time.
Highlights
IntroductionRemote sensing images can be continuously utilized to monitor land cover/land use changes around the world with extremely accurate precision
Experimental results indicate that the proposed architecture outperforms the SoA deep learning networks, and in particular we underline the improvement in multi-date configuration
The use of multi-temporal input stack gives better results considering some well-known segmentation metrics: accuracy, precision, recall, and F1-score. In this multi-temporal configuration we can assume that the convolutional layers of the considered networks mitigate the effect of the multiplicative speckle noise, and it is helpful in time management, because of lower training time required for deeper input stacks
Summary
Remote sensing images can be continuously utilized to monitor land cover/land use changes around the world with extremely accurate precision. Difference Water Index (NDWI) [10,11,12], and others have been developed to provide quantitative estimates of selected surface covers [6]. Multispectral sensors are unusable in presence of cloud cover and this prevents the possibility to guarantee a continuous monitoring. In order to overcome this drawback, recent works [13,14,15,16] proposed the use of Synthetic Aperture Radar (SAR) imaging sensors, able to acquire even in presence of clouds.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
