|Cendrero Mateo, M.p.|
|De La Cruz, F.|
Submitted to: IEEE Transactions on Geoscience and Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/1/2011
Publication Date: 4/1/2012
Citation: Moran, M.S., Alonso, L., Moreno, J., Cendrero Mateo, M., De La Cruz, F., Montoro, A. 2012. A radarsat-2 quad-polarized time series for monitoring crop and soil conditions in Barrax, Spain. IEEE Transactions on Geoscience and Remote Sensing. 50(4): 1057-1070. DOI: 10.1109/TGRS.2011.2166080. Interpretive Summary: New satellite sensors to measure radar backscatter over broad agricultural regions have been recently launched and more are planned for launch in the next decade. There is great interest in using radar images for monitoring crop and soil condition, including monitoring crop biomass, leaf area, crop residue, plant water content, crop growth stage, soil tillage and soil water content. In this study, we analyzed a time-series of 57 radar images for large fields of wheat, barley, oat, corn, onion and alfalfa in an agricultural area near Madrid Spain. We found that radar images provided information that could be used to map the distribution of crops over large areas. Further, images also provided valuable information about crop growth stage and soil moisture condition that could be used for crop management and yield prediction. The results from this study provided recommendations for planning future satellite systems to optimize the collection of meaningful information about crop and soil conditions.
Technical Abstract: The European Space Agency (ESA) along with multiple university and agency investigators joined to conduct the AgriSAR Campaign in 2009. The main objective was to analyze a dense time series of RADARSAT-2 quad-pol data to define and quantify the performance of Sentinel-1 and other future ESA C-Band SAR missions for classifying and monitoring agricultural crops within Global Monitoring for Environment and Security (GMES) services. In the Barrax region in La Mancha, Spain, 57 RADARSAT-2 C-band quad-pol SAR images and 5 RapidEye 4-band optical images were acquired from April to September 2009, covering multiple large fields of irrigated and non-irrigated crops. Using ascending and descending RADARSAT-2 orbits and beam incidence angles ranging from 23° to 41°, SAR images were acquired on average every 3 days. On the ground, records were kept of meteorological conditions, precipitation, and for select fields, crop type, phenology, irrigation and yield. An analysis of the sensitivity of SAR backscatter (so) to crop and soil conditions was conducted using the entire multi-view, quad-pol image time-series for large fields of wheat, barley, oat, corn, onion and alfalfa in Barrax. Preliminary results showed that the cross-polarized soHV was particularly useful for monitoring both crop and soil conditions and was found to be the least sensitive to differences in beam incidence angle; the greatest separability of barley, corn and onion occurred in the Spring after the barley had been harvested or in the narrow time window associated with grain crop heading when corn and onion were still immature; the time series of so offered reliable information about crop growth stage, such as jointing and heading in grain crops and leaf growth and reproduction in corn and onion; there was a positive correlation between RADARSAT-2 so and RapidEye Normalized Difference Vegetation Index (NDVI), where the increase in so was of the same order as the increase in NDVI; and the impact of view direction and incidence angle on the time-series was minimal compared to the signal response to crop and soil conditions. Related to planning for Sentinel-1, we found that quad-polarization with image acquisition frequency from 3-6 days was best suited for distinguishing crop types and monitoring crop phenology; dual-polarization with an acquisition frequency of 3-6 days was sufficient for mapping crop green biomass; and single- or dual-polarization with daily image acquisition was necessary to capture rapid changes in soil moisture condition. For all mission applications, a dense time-series (acquisitions every 3-6 days) was required to discriminate variations in surface crop and soil conditions from differences induced by changes in satellite/sensor configuration.