Soil moisture mapping using Sentinel-1-Data and SNAP?

hi sir! i opened the above link but i cannot see any extinction by which i can directly estimates soil moisture.please help me

@Naz, i think you should see the Arset webinar training regarding soil moisture.

Thanks @gomalhunzai

1 Like

There is no stb-xj in the ftp repository. Just wandering if you can share the download.


The stb-xj is not there anymore, the ftp folder is empty. Just wondering if you can share the downloaded installer/plug-in.


Hi ABraun How can I get Soil moisture map for my study area after correlating backscattering coefficient and volumetric soil moisture content which is collected from the site

there are many ways to do it, but I am no expert in either of them, sorry.
My advice is to look at some studies and try to find the approach which fits your needs best.

Did you get a linear relationship between backscatter and moisture?
The easiest way then would be to apply the regression on an image and mask out all invalid areas (vegetation, urban, water).

The approach described by johngan here is also quite straightforward and promising, but it is based on images from the rainy and dry season: Soil moisture identification -Sentinel
But you could still use your measurements for refinement or validation in this approach.

I think the most important thing for you is to decide for one strategy and see what is needed to accomplish it over the entire workflow, instead of proceeding by single steps unknowing what comes next.

Hope this helps you a bit.

Some interesting approaches:

you are right and at this time i want to adopt the empirical method using onsite SMC data and i want know how i can make soil moisture content map using my data and backscratching coefficients


To derive soil moisture using SAR there are quite a few different techniques. The three most popular ones used extensively in remove sensing are the following:

  • Soil moisture using Neural Network (NN)
    Reference: Sahebi, M.R.; Bonn, F.; Bénié, G.B. Neural networks for the inversion of soil surface parameters from synthetic
    aperture radar satellite data

  • Soil moisture using Water Cloud Model (WCM)
    reference: Estimation of surface soil moisture and roughness from multi-angular
    ASAR imagery in the Watershed Allied Telemetry Experimental Research

  • Soil moisture using change detection
    reference: Thoma, D.P.; Moran, M.S.; Bryant, R.; Rahman, M.; Holifield-Collins, C.D.; Skirvin, S.; Sano, E.E.; Slocum, K.
    Comparison of four models to determine surface soil moisture from C-band radar imagery in a sparsely vegetated semiarid landscape

If you do not know what methodology to use, i would recommend to use the change detection one. Not only there is a lot of literature for this technique and used by many scientists, it is the most straightforward as well.

Using this technique, it requires two images. One image captured under dry weather conditions and the other one captured under wet weather conditions. We assume that the vegetation and surface roughness have not changed between the images acquisitions. So, it is advised to remove areas of high and dense vegetation (use NDVI to mask-out vegations)

The collection of your field measurements can be used in two different ways:

  1. Validate the soil moisture results derived by Sentinel-1 (using any of the above technique mentioned). This can be done by fitting a linear model between the soil moisture of Sentinel-1 and your field measurements. Hopefully, there is going to be a strong correlation between the two measurements

  2. Your fields measurements can be used to derive volumetric soil moisture from Sentinel-1 (assuming that you have collected volumetric data on the filed). You need a linear relationship between Sentinel-1 soil moisture values and the volumetric data collected on the field. Have a look on this paper, especially in figure 5 where authors derive volumetric soil moisture based on regression analysis.


Thank you. and I want somebody to help me I used volumetric soil moisture content data taken from the study area and corresponding sentinel 1 image for linear regression analysis and I was expecting good correlation but I gat very low correlation additionally my backscattering coefficient seems not right any suggestion, please


Firstly, The units of volumetric data should be in m^3. In your case, the numbers in x-axis range from 1 to 20. I do not think the units of these numbers are expressed in m^3 (they are far too large). Are these the correct in-situ measurements ?

Secondly, the SAR backscatter should be in dB units when you perform the regression analysis with your field data. As i can see in your graph, the SAR image is not in dB units

Also, do not expect perfect correlation between the filed data and the SAR backscatter coefficient. Rough terrain or vegetation can produce stronger backscatter compared to flat surfaces. So, you might have very high backscatter values for some areas when compared to your filed measurements.

In addition to johngan’s comments, I think it makes sense to declare the VSM as the dependent variable (y axis) which is predicted by the SAR backscatter (x axis), so you can later apply the regression formula on the image.

I want t say thank you both of you for your constrictive comments.The soil moisture data is exactly taken from site but it is in centimeter cube.Does it is must be in meter cube only?secondly I used by right clicking on the band of back scattering coefficients and linear to from dB but still not changed.

the scale of the dB raster should now be a different one and the values should roughly range between -40 and +5

Hi @johngan

What will be the reason for the very low correlation between backscatter coefficient in dB units and field moisture content? Is it because of any mistakes in preprocessing steps?


Firstly, Let’s assume that the soil moisture equipment is reliable and the person that took the measurements knew what he was doing. Hence, we know that in-situ measurements are correct.

Usually, we expect to have a good correlation between backscatter intensity and filed measurements (we can reach R=80%, not in all cases though).

What could affect the bacscatter coefficient can be attribute to two factors:

The presence of vegetation. If the SAR image is acquired at the point where crops are significantly grown, then only small proportion of the energy reaches the actual surface, while most of the energy interacts with the crops.

Surface roughness. If the area of interest is uneven (with lots of slopes) then the terrain contributes significantly to the energy that comes back to the satellite. So, in that case, the backscatter intensity is a combination of soil moisture and surface roughness.

The contributions from vegetation and surface roughness can be removed if you have in-siture measurements. Water Cloud Model is a popular approach for removing the vegetation contribution

1 Like

But sir, when I reviewed journals, I saw that along with moisture data, surface roughness is also another input for water cloud model. I don’t have surface roughness data.

When trying to estimate soil moisture from SAR, there are limitations. Those limitations are well known and highlighted in all papers (vegetation and surface roughness). We either accept those limitations and produce a soil moisture product which contains an amount of error in it, or we try to find a method to minimize that error.

WCM is one of the methods that is used to minimize the effect of roughness and vegetation. Unfortunately, it requires filed data. If you are a student where you need to produce a scientific product, then you should make sure you understand the topic very well and have all the necessary data collected before moving to developing the algorithm. If, for any reason you did no manage to collect all the data necessary (e.g surface roughness) then there is not much you can do. As a workaround on that, you can use a DEM and derive the surface roughness. Of course, the DEM should be of high resolution and be acquired at the same date with the SAR dataset.

I do not know how accurate the WCM model can be if you ignore the surface roughness factor and incorporate only the vegetation.


Sir, I derived ruggedness index from jaxa DEM. As the ruggedness index quantifies topographic heterogeneity, is it OK to use in water cloud model ?

I think this should work.
You just need to try and see how well it works