The use of Sentinel-1 and -2 data for monitoring maize production in Rwanda

Jason Polly; Meghan Hegarty-Craver; James Rineer; Margaret O'Neil; Daniel Lapidus; Robert Beach; Dorota S. Temple

doi:10.1117/12.2533221

21 October 2019 The use of Sentinel-1 and -2 data for monitoring maize production in Rwanda

Jason Polly, Meghan Hegarty-Craver, James Rineer, Margaret O'Neil, Daniel Lapidus, Robert Beach, Dorota S. Temple

Proceedings Volume 11149, Remote Sensing for Agriculture, Ecosystems, and Hydrology XXI; 111491Y (2019) https://doi.org/10.1117/12.2533221
Event: SPIE Remote Sensing, 2019, Strasbourg, France

Abstract

Although Rwanda has accomplished significant improvements in food production in recent years, one fifth of its population remains food insecure. Agricultural information is currently collected through seasonal agricultural surveys, but more frequent and timely data collection is needed to adequately inform public and private decision-makers about the status of crops during the growing season. Sentinel-1 and -2 data are freely available with new images provided every 4-5 days. While analysis of these multispectral images has been used for agricultural applications, there are few applications to smallholder agriculture. Major challenges for satellite image analysis in the context of Rwanda include heavily clouded scenes and small plot sizes that are often intercropped. Sentinel-2 scenes corresponding to mid-season were analyzed, and spectral signatures of maize could be distinguished from those of other crops. Seasonal mean filtering was applied to Sentinel-1 scenes, and there was significant overlap in the spectral signatures across different types of vegetation. Random Forest models for classification of Sentinel scenes were developed using a training dataset that was constructed from high-resolution multispectral images acquired by unmanned aerial vehicles (UAVs) in several different locations in Rwanda and labeled as to the crop type by trained observers. The models were applied to satellite images of the whole country of Rwanda and validated using a test dataset from the UAV images. The Sentinel-2 model had the user’s accuracy for maize classification of 75%, while the Sentinel-1 model overestimated the maize area resulting in a user’s accuracy of <50%.

Citation Download Citation

Jason Polly, Meghan Hegarty-Craver, James Rineer, Margaret O'Neil, Daniel Lapidus, Robert Beach, and Dorota S. Temple "The use of Sentinel-1 and -2 data for monitoring maize production in Rwanda", Proc. SPIE 11149, Remote Sensing for Agriculture, Ecosystems, and Hydrology XXI, 111491Y (21 October 2019); https://doi.org/10.1117/12.2533221

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