Random forest classification using Sentinel-1 and Sentinel-2 series for vegetation monitoring in the Pays de Brest (France)

Simona Niculescu Sr.; Antoine Billey; Halima Talab-Ou-Ali Jr.

doi:10.1117/12.2325546

9 October 2018 Random forest classification using Sentinel-1 and Sentinel-2 series for vegetation monitoring in the Pays de Brest (France)

Simona Niculescu Sr., Antoine Billey, Halima Talab-Ou-Ali Jr.

Author Affiliations +

Proceedings Volume 10783, Remote Sensing for Agriculture, Ecosystems, and Hydrology XX; 1078305 (2018) https://doi.org/10.1117/12.2325546
Event: SPIE Remote Sensing, 2018, Berlin, Germany

Abstract

Nowadays, optical and radar remote sensing data are increasingly used for land-cover/vegetation mapping and monitoring. Their technical capabilities and tools are improving all the time and provide more accurate results. By the recent arrival of the Sentinel-1 and Sentinel-2 series, available free, processing and methods of analysis must be increased more and more in the field of cartography. This paper aims to present vegetation mapping method in the Pays de Brest area by using a time series stacking of Sentinel-1, Sentinel-2 and SPOT-6 satellites data using the algorithm Random Forest supervised classification. The types of vegetation mapping in first time are those belonging to the major vegetation types, but especially those that can be observed on the processed images that are the Sentinel-1, Sentinel-2 series and SPOT-6. The types of classes considered for this study are: no vegetation, forest and undergrowth, moors and lawns, summer crops, winter crops, grassland and water. Several time series stacking has been made on that series containing 140 images radar representing different dates (2017) and the best combination method is to use both the two polarizations VV and VH to the calculation of the matrix of confusion. On the other hand, combinations of SAR images with different vegetation indices (NDVI, NDWI, S2rep, IRECI) calculated from the Images Sentinel-2 have been made. The series of times series stacking ends with combinations between SPOT-6 and Sentinel-1. The times series stacking Sentinel-1, Sentinel-2 and SPOT-6 are satisfactory, with an overall accuracy that reaches 93%. Such precision is very good for data that are available free.

Conference Presentation

Citation Download Citation

Simona Niculescu Sr., Antoine Billey, and Halima Talab-Ou-Ali Jr. "Random forest classification using Sentinel-1 and Sentinel-2 series for vegetation monitoring in the Pays de Brest (France)", Proc. SPIE 10783, Remote Sensing for Agriculture, Ecosystems, and Hydrology XX, 1078305 (9 October 2018); https://doi.org/10.1117/12.2325546

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available