Using dense time-series of C-Band SAR imagery for classification of diverse, worldwide agricultural systems

Authors: DINGLE ROBERTS, LAURA - Agriculture And Agri-Food Canada; DAVIDSON, ANDREW - Agriculture And Agri-Food Canada; MCNAIRN, HEATHER - Agriculture And Agri-Food Canada; HOSSEINI, MEHDI - Carleton University - Canada; MITCHELL, SCOTT - Carleton University - Canada; DE ABELLEYRA, DIEGO - Instituto Nacional De Tecnologia Agropecuaria; VERON, SANTIAGO - Instituto Nacional De Tecnologia Agropecuaria; DEFOURNY, PIERRE - Universite Catholique De Lille; LE MAIRE, GUERRIC - Cirad-La Recherche Agronomique Pour Le Developpe; PLANELLS, MILENA - Center For The Study Of The Biosphère From Space(CESBIO); VALERO, SILVIA - Center For The Study Of The Biosphère From Space(CESBIO); AHMADIAN, NIMA - Julius Kuhn Institute; Coffin, Alisa; Bosch, David; Cosh, Michael; SIQUEIRA, PAUL - University Of Massachusetts, Amherst; BASSO, BRUNO - Michigan State University; Saliendra, Nicanor

Submitted to: International Geoscience and Remote Sensing Symposium Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 5/27/2019
Publication Date: 9/30/2019
Citation: Dingle Robertson, L., Davidson, A., Mcnairn, H., Hosseini, M., Mitchell, S., De Abelleyra, D., Veron, S., Defourny, P., Le Maire, G., Planells, M., Valero, S., Ahmadian, N., Coffin, A.W., Bosch, D.D., Cosh, M.H., Siqueira, P., Basso, B., Saliendra, N.Z. 2019. Using dense time-series of C-Band SAR imagery for classification of diverse, worldwide agricultural systems. In: Proceedings of the 2019 IEEE International Geoscience and Remote Sensing Symposium, Yokohama, Japan, 2019, pp. 6231-6234. https://doi.org/101109/IGARSS.2019.8898364.
DOI: https://doi.org/10.1109/IGARSS.2019.8898364

Interpretive Summary: Cloudy conditions impede and reduce the usefulness of optical satellite imagery to assess earth surface conditions. However, Synthetic Aperture Radar (SAR) sensors are not constrained by clouds. Therefore, they are useful tools for being able to assess crop characteristic on the earth's surface, especially in cloudy areas. With the launch of Sentinel-1A and B satellites by the European Space Agency, the ongoing availability of the Canadian Space Agency's RADARSAT-2 imagery, and the expected launch of the RADARSAT Constellation Mission (RCM), dense time series of SAR data will now be readily available. But, for crop classification and mapping, SAR imagery has yet to be used to its full potential and has generally been combined with optical imagery. The JECAM SAR Inter-Comparison Experiment is a multi-year, multi-partner project that aims to compare global methods for crop monitoring and inventory using SAR data. To accomplish this, sets of dense time-series SAR imagery, which include RADARSAT-2 and Sentinel-1 data, were prepared for this experiment. Two classification methods developed by Agriculture and Agri-Food Canada were applied to SAR only data-stacks, and also to traditional data-stacks of optical/SAR combinations. This paper outlines the results of these dense time-series classifications and how these results were affected by changing numbers of agriculture classes, numbers of available SAR imagery and numbers of training and validation data points for individual crop types. In general, for the dense time-series SAR stacks, overall crop classification accuracies of greater than 85%, a typical operational goal, were obtained for 6 of 12 sites. These results have important implications for particularly cloudy regions where the availability of optical imagery is limited.

Technical Abstract: Cloudy conditions impede and reduce the utility of optical imagery. With the launch of Sentinel-1A and B, the ongoing availability of RADARSAT-2 imagery, and the expected launch of the RADARSAT Constellation Mission (RCM), dense time series of C-band Synthetic Aperture Radar (SAR) data will now be readily available. For crop classification and mapping, SAR imagery has yet to be used to its full potential and has generally been combined with optical imagery. The JECAM SAR Inter-Comparison Experiment is a multi-year, multi-partner project that aims to compare global methods for SAR-based crop monitoring and inventory. Sets of dense time-series SAR imagery which include RADARSAT-2 and Sentinel-1 data were prepared for this experiment. AAFC’s operational Decision Tree (DT) and newly implemented Random Forest (RF) classification methodologies were applied to these SAR only data-stacks, and to optimized, traditional data-stacks of optical/SAR combinations. This paper outlines the results of these dense time-series classifications and how these results were affected by changing numbers of agriculture classes, numbers of available SAR imagery and numbers of training and validation data points for individual crop types. In general, for the dense time-series SAR stacks, overall accuracies of greater than 85%, a typical operational goal, were obtained for 6 of 12 sites. These results have important operational implications for particularly cloudy regions where the availability of optical imagery is limited.