Use of Multi-temporal Polarimetric SAR Data for Crop Characterisation
Jai Singh Parihar(1), Chakrapani Patnaik(1) and Saroj Maity(1)
(1) Indian Space Research Organisation, Space Applications Centre, Ahmedabad 380015, India
Microwave remote sensing is an invaluable tool especially during cloud cover. Single parameter microwave remote sensing (polarisation, frequency, temporal) cannot give detailed information of the target. Multi-polarisation and multi temporal SAR data can be used to derive information and gives a better understanding of the target. Under the Radarsat – 2 SOAR project, (Project ID 1040) four quad polarisation multi-temporal fine beam data from Radarsat - 2 (FQ16 at 35.5° incidence angle) have been acquired to identify and study the response of maize and cotton primarily. In addition, other ground cover like other crops, forest, urban and water bodies were also studied. The SAR data acquired was in SLC mode.
The study area of 25 km x 25 km is centered around 75° 37’E longitude and 14° 35’ N latitude in Haveri district of Karnataka state in India. Cotton and maize are sown in the second fortnight of June to early July. The dates of data acquisition were July 04, July 28, August 21 and September 14 2008, respectively. The soil varies from black cotton (highest clay) to red soils (vertisols). The area has minor undulations in the terrain with a large open forest. The major crops grown in this area are cotton, maize, garlic and onion followed by sunflower.
An adaptive filter was used to reduce the speckles and the data was calibrated using information obtained from the header file for each date. The phase information has been used to better understand SAR-crop interaction.
Ground truth was collected synchronous to each pass of the satellite and various crop parameters like row spacing, orientation, age, height, biomass, LAI, variety, leaf angle, number of scatterers, etc; and soil parameters like roughness, moisture were collected. Field boundaries are demarcated using IRS LISS-IV data. Preliminary analysis revealed that there was a significant increase in temporal backscatter for the crops and separability in a single date using polarimetric data was better than multitemporal single polarised data for major land cover. Though the co-polarisation ratios were within similar order of difference for maize and cotton, it was seen that the cross polarisation ratios for cotton were significantly more than maize between early and late stages. This could be attributed to the more canopy volume of cotton and the increased number of scatterers as the crop grows compared to maize. The water cloud model using the polarimetric components is attempted and it shows the effect of polarisation on crop biophysical parameters. The paper highlights the salient findings using multi-temporal polarimetric SAR data in crops.