TIME SERIES ANALYSIS OF UAVSAR L BAND POLSAR DATA FOR AGRICULTURAL LANDS MONITORING

Monitoring of agricultural activities plays an important role in the management of economic and environmental resources. In particular, monitoring of agricultural lands is helpful for yield forecasting, identifying the risk of diseases, application of chemicals, irrigation, water resources management and other applications. Remote sensing has high potential in the agriculture sector for various benefits, such as efficient spatial, spectral, and radiometric resolution of Earth observations, as well as large area and appropriate temporal coverage. Synthetic aperture radar (SAR) signals are almost unaffected by atmospheric conditions and illumination. As a result, SAR observations are very beneficial for crops monitoring. In this paper, we used multitemporal polarimetric SAR (PolSAR) data collected by JPL NASA s L-band Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR) in 2012, over an agriculture area in Winnipeg, Manitoba, Canada. Time series of polarimetric features of various annual crops were extracted and compared to in-situ dry biomass. These features were Linear Intensities, Radar Vegetation Index (RVI) and also Cloude-Pottier, Freeman-Durden and Yamaguchi decompositions. Our experimental tests consisted of statistical and regression analyses among these remotely sensed features and in-situ measurements. The results showed a relatively high correlation between certain radar parameters and crops’ biomass in general. However, winter wheat had a weak correlation (R2=0.65). For example, canola (R2=0.73), corn (R2=0.86) and soybeans (R2=0.88) were quite sensitive to the radar signal.