Radar Images and Interpreting the Data – Part 2

October 31, 2019

     Let us quickly summarize what I went through in Part 1: I downloaded the radar polarized-echo products from the UAVSAR JPL website, converted the products to cube files using USGS ISIS3, scaled down the samples and lines by a factor of 2 to reduce the speckle noise effect and produced Stoke Matrix parameters (W11, W14, W44), SC, OC and CPR raster datasets. The next step is to extract scattered polarization state and CPR values to see if we have reduced the speckle noise effect using the reduce command function. To extract quantitative values, I used the Zonal Statistics To Table tool in ArcGIS, which extracts the Min, Max, Mean, Range, Count and STD (standard deviation) of the pixel values into a table. I compared the mean and STD values to see if there were any differences. I noticed that the mean values did not change but the STD values slightly decreased. The HHHH rasters for example report lower STD values from the reduced products (see table 1 and 2).



Table 1 summarizing the HHHH values of the original raster datasets.


Table 2 summarizing the HHHH values of the reduced raster datasets.


     I repeated the steps to extract values from VVVV and HVHV datasets and noticed the same thing. The mean values remained the same and the STD values decreased. Even after the pixels were averaged and fitted into scaled down sample and line dimensions the mean values did not change. It appears to have no effect on the mean but reduces the STD values and errors. 


     I wanted to show you all the UAVSAR datasets I have produced using IDL, ISIS3 and ArcGIS. I have created a slide show gallery below for you to view them. 




     I mentioned in Part 1 of this blog ("Radar Images and Interpreting the Data: Part 1") that I want to learn how to improve my radar interpretation skills and how to plot and analyze quantitative radar polarized-echo and CPR values. I mentioned I reread Campbell (2012) from my literature review reading list. I noted in his work that he used various statistical parameters and values to infer the dominant type of radar backscatter mechanism(s) (from 24 cm wavelength AIRSAR data) of the SP Flow in Arizona. Campbell (2012) looked into using radar parameters to better interpret the types of backscatter mechanisms returning specific CPR values. He used the parameter β (which is a normalized value of Re(ShhSvv*), read Campbell (2002) for details) to report a range value from -1 to 1 to indicate how dominant single bounce and double bounce returns were in SAR data. When β approaches -1 it corresponds to an ideal double bounce return, while 1 corresponds to an ideal single bounce return. He also plotted CPR against the ratio of HH-, VV- and HV-polarized echo data points. By doing this, Campbell (2012) was able to infer that dihedral scattering was the dominant backscatter mechanism at SP Flow and that the highest CPR values were due to double bounce returns. His results also showed that some parts of the SP Flow are predominantly single bounce returns and multiple scattering.


     I have copied the graphs used in Campbell (2012) so you can all get an idea of how he plotted the results.


 Credit: Campbell (2012). Top right histogram of CPR at the 24 cm wavelength extracted from a sample area of SP Flow (see Figure 8 in Campbell (2012)). The top right plot is average CPR of the sample region vs the ratio of HH- and VV-polarized echoes. The plot shows no correlation between the two function, which implies a dihedral backscatter mechanism. The bottom left plot is a positive correlation between CPR and the ratio of HV- and HH-polarized echoes. A positive correlation between these functions is expected for both single bounce and double bounce backscatter returns. The bottom right plot is CPR vs β. A negative correlation between CPR and β, in conjunction with a decline in HH/VV values, implies that the backscatter relates to facet-like echo components.


     I am wanting to reach out and ask whether it is a good idea for me to extract this type of data from the UAVSAR data products. I am not an expert in radar remote sensing, so I am still learning new methods to interpret and process radar data. Any input from experts will be highly appreciated and will help improve my radar interpretation skills. Planetary radar research is a field I want to pursue and I am always trying to become more involved to understand radar instruments and the data products. I will still be reading more papers about radar remote sensing to understand the terminology and mathematics behind the equations used by Campbell (2012) (as well as other workers such as Carter et al. (2004 and 2006) and Neish et al. (2017)).


See you guys next time!



  • Campbell, B. A. (2012). High circular polarization ratios in radar scattering from geologic targets. Journal of Geophysical Research: Planets, 117(E6).

  • Campbell, B. A. (2002). Radar remote sensing of planetary surfaces. Cambridge University Press.

  • Carter, L. M., Campbell, D. B., & Campbell, B. A. (2004). Impact crater related surficial deposits on Venus: Multipolarization radar observations with Arecibo. Journal of Geophysical Research: Planets, 109(E6).

  • Carter, L. M., Campbell, D. B., & Campbell, B. A. (2006). Volcanic deposits in shield fields and highland regions on Venus: Surface properties from radar polarimetry. Journal of Geophysical Research: Planets, 111(E6).

  • Neish, C. D., Hamilton, C. W., Hughes, S. S., Nawotniak, S. K., Garry, W. B., Skok, J. R., ... & Osinski, G. R. (2017). Terrestrial analogues for lunar impact melt flows. Icarus, 281, 73-89.


Share on Facebook
Share on Twitter
Please reload

The Science & 

Mathematics University

© 2023 by Scientist Personal. Proudly created with Wix.com

  • Facebook Social Icon
  • Instagram Social Icon
  • Twitter Social Icon
  • LinkedIn Social Icon
This site was designed with the
website builder. Create your website today.
Start Now