Soil Conservation: Soil Tillage and Crop Residue

Soil conservation is a major concern in agriculture. Tillage practices have direct bearing on the potential for wind and water erosion, and soil quality, especially as it relates to the maintenance of soil organic matter. The ability to monitor the type of tillage, and amount of residue is important for soil conservation. Tillage affects surface roughness (as a function of implements used and number of passes) and the amount of residue Brisco et al, McNairn et al 1996, McNairn et al 2002a. Due to the sensitivity of radar backscatter to field characteristics including surface roughness, polarimetric data may prove very useful for monitoring soil tillage and crop residue.

There are a number of polarimetric parameters found to be useful for discriminating soil tillage/residue. Highlighted here are the use of Co-polarization Signatures and Co-polarization Phase Difference.

9.1.2.1 Polarimetric Signatures

Polarization signatures are a graphical method of visualizing the backscatter response of a target as a function of incident and backscattered polarizations. Polarization signatures can be used to give a graphical presentation of the backscattering characteristics and can therefore be useful in differentiating target characteristics. In this example, this is shown in relation to variations in tillage and residue cover.

Co-polarization plots were obtained from SIR-C imagery for fields with varying tillage and amounts and types of residue in a study by McNairn et al. McNairn et al.,2002a

A) Tilled Field with Little or No Residue

The following co-polarization plots were obtained from C-band (Figure 9-1) and L-band (Figure 9-2) imagery for tilled fields with varying amounts and types of residue. Incidence angles were between approximately 42 and 50 degrees McNairn et al.,2002a.

Figure 9-1a
A) Pea Residue
(25% residue cover)
Figure 9-1b
B) Lentil Residue
(25% residue cover)
Figure 9-1c
C) Canola Residue
(40% residue cover)
Figure 9-1d
D) Wheat/Barley Residue
(20% residue cover)
Figure 9-1e
E) Sunflower Residue
(40% residue cover)
Figure 9-1. C - Band co-polarization signatures, tilled field with little or no residue (from McNairn et al 2002a).

The smoothest fields, those with the finest residue have a maximum response at VV polarization (Orientation Angle of 90°). Backscatter was approximately equal at all linear polarizations for the canola, wheat/barley and sunflower residue fields, suggesting that these fields appear rougher.

Figure 9-2a
A) Pea Residue
(25% residue cover)
Figure 9-2b
B) Lentil Residue
(25% residue cover)
Figure 9-2c
C) Canola Residue
(40% residue cover)
Figure 9-2d
D) Wheat/Barley Residue
(20% residue cover)
Figure 9-2e
E) Sunflower Residue
(40% residue cover)
Figure 9-2. L-Band co-polarization signatures, tilled field with little or no residue.(from McNairn et al 2002a).

In the L-Band example, fields that had been tilled or had very fine residue cover exhibited responses typical of surface scattering. For these targets, maximum response is at an Orientation Angle of 90° and the target appears flat relative to the wavelength. For these fields, the low pedestal heights(0.18-0.24) indicate minimal depolarisation, hence confirm that surface scattering is dominant. These surfaces are not rough enough and do not have enough vegetative material to cause significant multiple or volume scattering. The difference between VV and HH backscatter is much more pronounced at L-Band than at C-Band, since at this longer wavelength, these surfaces appear much smoother.

B. No-Till Residue Fields with Multiple Scattering

The following co-polarization plots were obtained from C-band (Figure 9-3) and L-band (Figure 9-4) imagery for no-till fields with varying amounts and types of residue. (from McNairn et al 2002a)

Plots for C-Band (Figure 9-3) with the exception of (a) exhibit a saddle type shape typical of double bounce scattering. The pedestal height in these co-polarization plots was larger and thus indicates that greater depolarization was occurring on the no-till fields as compared to the tilled fields with low residue cover. The pea residue field (Figure 9-3a) shows a response similar to that for the field with pea residue that had been tilled (Figure 9-1a). Comparison of these two figures suggests that very fine residue has little effect on radar response as this target appears "smooth" at C-Band.

Figure 9-3a
A) Pea Residue
Figure 9-3b
B) Lentil Residue
Figure 9-3c
C) Canola Residue
Figure 9-3d
D) Wheat/Barley Residue
Figure 9-3e
E) Sunflower Residue
Figure 9-3. C-Band co-polarization signatures, no till fields with multiple scattering.(from McNairn et al 2002a).

The co-polarization plots for L-Band imagery are significantly different when compared to those at C-Band. Relative to C-band the double bounce scattering is substantially reduced with a VV peak often present, indicating significant surface scattering for wheat/barley, and sunflower residues.

Figure 9.4a
A) Pea Residue
Figure 9.4b
B) Lentil Residue
Figure 9.4c
C) Canola Residue
Figure 9.4d
D) Wheat/Barley Residue
Figure 9.4e
E) Sunflower Residue
Figure 9.4. L-Band co-polarization signatures, no till fields with multiple scattering (from McNairn et al 2002a).

A full discussion of polarimetric plots and other polarimetric parameters related to differentiating tillage and field residue is given in McNairn et al

9.1.2.2 Co-Pol Phase Difference (PPD)

The Co-polarization Phase Difference is a polarimetric parameter that may be useful for characterizing backscattering mechanisms. For instance, a single bounce (or odd bounce) scatterer will have a relative phase difference between HH and VV of 0° in the FSA For a double bounce (or even bounce) scatterer, the phase difference will be 180°. If the basis convention is changed to the BSA there is a further sign change, which adds 180° to the phase difference. As an example, bare soils are surface scatterers, and generally have a mean Co-polarized Phase Difference equal to 0° with a small standard deviation.

Ulaby et al. Ulaby et al 1987 suggested that much of the information content provided by the Co-polarized Phase Differences lies in the distribution of these differences (as expressed by the standard deviation) e.g., a rough plowed field has a phase distribution that is broad when compared to that for a smoother disked field.

As shown in the study by McNairn et al. McNairn et al 2002a, the Mean Co-polarized Phase Difference was found to be close to zero for most fields with residue and so provided little useful information . Standing senesced crops did show phase differences significantly greater than 0° with the mean phase differences varying between the fields where multiple scattering is believed to occur. In terms of tillage practices, mean phase differences for fields could not be used to distinguish between the presence or absence of tillage and residue types, although differences were found between standing senesced crops and harvested fields. For example, the standing senesced corn and sunflower fields tended to have much higher mean phase differences varying between -30° and -130° for C-Band and -30° and -90° for L-Band. The field based phase distribution data can be used to differentiate tilled low residue fields from no-till (high residue) fields. The standard deviation of the phase differences for fields with low residue cover was less than 30o; the standard deviation for no-till fields exceeded 45°. These results are consistent with those reported by Ulaby et al. Ulaby et al 1987. An example figure showing the distribution of fields with various Mean Phase Differences is given as Figure 9.5 (modified from McNairn et al 2002a).

Figure 9-5
Figure 9-5. C-Band field based co-polarized phase distribution does vary as a function of residue characteristics (modified from McNairn et al. McNairn et al 2002a).

Whiz quiz

Question: How would the polarization phase difference (PPD) compare for a bare field and a standing corn field?

The answer is...

Whiz quiz - answer

Answer: The PPD would have a mean close to zero with a small standard deviation for the bare field. The corn field would have a mean PPD different from zero with a larger standard deviation.