ORBITAL EFFECTS ON ERS-1 SAR TEMPORAL BACKSCATTER PROFILES OF AGRICULTURAL CROPS

abstract

Multitemporal radar backscatter characteristics of crops and their underlying soils are analyzed for an agricultural area in southwestern Ontario, Canada using nine dates of ERS-1 SAR imagery acquired during the 1993 growing season. From the calibrated data, SAR temporal backscatter profiles were generated for each crop type. It was found that the radar backscatter coefficients varied considerably from one date to the next. Attempts were made to explain the variations based on changes in environmental conditions or potential variations due to characteristics of the SAR system. The results of the study indicate that small changes in incidence-angle can have strong impacts on radar backscatter. Thus, attention must be given to local incidence-angle effects when using ERS-1 SAR data, especially when comparing backscatter coefficients of the same area from different scenes or different areas within the same scene.

1. Introduction

Spaceborne SAR is potentially an important data source for agricultural applications. It satisfies a basic agricultural requirement for reliable and frequent imaging throughout the crop growing season. Before the launch of the long-duration spaceborne SAR systems, airborne studies had demonstrated that multitemporal SAR data can enhance the ability to distinguish between various crop types (e.g., Brisco et al., 1984; Brown et al., 1984; Fischer and Mussakowski, 1989; Foody et al., 1989; Dobbins et al., 1992). However, comparatively few datasets have been available for study because of the increased cost and logistics of generating multitemporal SAR data with airborne platforms (Brisco et al., 1992). With the launch of ERS-1&2, JERS-1 and RADARSAT, world-wide spaceborne SAR data became routinely available. To be able to extract information from multitemporal spaceborne SAR, it is important to understand the radar backscatter behaviours of individual crops throughout the growing season and to determine what factors are likely to cause variation in backscatter between different dates.

In this study, we analyze the multitemporal radar backscatter characteristics of crops and their underlying soils over the growing season and attempt to relate the variations between dates to agricultural/environmental parameters and SAR system parameters.

2. study area and data description

The study area is situated in an agricultural area in Oxford County, southern Ontario, Canada (Figure 1). Approximately 15 km x 4 km, this area has been selected as one of the few representative agricultural ‘supersites’ across Canada at which the relationships between radar data and agriculture are being studied (Brown et al., 1991).

Figure 1. The study area and the ERS-1 satellite orbits

The ERS-1 C-VV SAR data were acquired on nine dates during two descending passes in the 1993 growing season (Figure 1, Table 1). Extensive ground data were collected by field teams at the times of the satellite overpasses. It included observations and measurements of the crops (e.g., crop type, growth stage, percentage cover, canopy height, row spacing, plant condition) and their soil characteristics (e.g., soil type, surface roughness, moisture content, row direction). The field boundaries were digitized from a SPOT image.

Table 1. ERS-1 Data

3. Methodology

3.1 Preprocessing

Field-Boundary Preparation

Since the development of image segmentation techniques is not the aim of this study, the geocoded (i.e., Universal Transverse Mercator (UTM) coordinates) field-boundary file for the study area was generated using a PAMAP GIS. First, the field-boundary file was converted from a vector format to a raster format; then a 5-pixel buffer was applied to the field boundaries to eliminate the effects of field boundary pixels and minor image registration errors on crop discrimination. The file was then imported into the PCI EASI/PACE image processing system.

SAR Data Radiometric Calibration and Geometric Correction

Quantitative comparisons of the multitemporal SAR data require calibrated images. Thus, the SAR data used in this study were corrected for SAR antenna pattern and compensated for range-spreading loss at ESA D-PAF, Germany. The calibration accuracy is + 0.42 dB (Laur et al., 1993).

The SAR images were then geometrically corrected to field boundaries with a 12.5 m pixel spacing using a second-order polynomial and a nearest-neighbour resampling algorithm.

3.2 Derivation of the Radar Backscatter Coefficient s °

The generation of ERS-1 SAR temporal backscatter profiles of agricultural crops requires relating pixel digital numbers (DN) on SAR images to backscatter coefficients of corresponding distributed targets in the scene. According to Laur (1992), the complete equation to be applied to determine the backscatter coefficient s ° of an area located at incidence angle a is:

where: K = K (a _ref = 23° )

R is the slant range distance at the distributed target location and R _ref is a reference slant range distance; i.e., the mid-swath slant-range distance R _ref = 847.0 km

g²(q ) is the two-way antenna pattern profile and q is the look angle at the distributed target location

Since the ESA SAR PRI products have compensated for range spreading loss and antenna pattern, the equation can be simplified to:

where : 1 i=N

<I> = ---- · å DN_i²

N i=1

DN_i is the digital number of a given pixel i and is proportional to the square-root of the intensity I_i received from the ground resolution cell corresponding to pixel i.

N is a large pixel number (more than 500) to ensure statistical validity to the estimation of the mean intensity.

Expressed in decibels (s ° (dB) = 10 · log₁₀ s ° ), we have:

s ° (dB) = <I>(dB) - K(dB) + b (dB)

where K = 58.63 dB

In PRI images, the range of incidence angles a is typically from 19.5° at the near-range to 26.6° at the far-range. The correction factor b can vary from - 0.7 dB to +0.6 dB with image swath (Laur, 1992).

3.3 Temporal Backscatter Profile Generation

SAR temporal backscatter profiles were generated in EXCEL for major crop types (i.e., corn, soybeans, winter wheat, barley/oats, alfalfa and pasture). First, the temporal backscatter profile for each individual field was generated; then the general temporal backscatter profile for each crop was generated by averaging the s ° of all fields for that crop type on each date.

4. Results and Discussion

A wide range of parameters affects the backscatter of microwaves from vegetation and soil. The important system parameters, however, are frequency, polarization and incidence angle. The crucial features of the target in determining the proportion of radiation returning to the sensor are plant canopy (e.g., plant type, height, density, biomass, water content and growth stage) and soil parameters (e.g., soil moisture content, roughness and tillage direction). The SAR temporal backscatter profiles for each crop show the complexity of the relationship between microwave and agricultural parameters over the growing season.

Using multitemporal ERS-1 SAR data during the 1993 growing seasons, the radar backscatter characteristics of crops and their underlying soils were analyzed. The SAR temporal backscatter profiles were generated for each crop type (Figure 2). In general, radar backscatter was primarily influenced by soil in the early season when fields were bare or exhibited a limited crop canopy. With crop development, radar backscatter decreased due to attenuation and absorption by vegetation canopies. The decreasing trend continued until crops were at the seed development stage. Then the backscatter started to increase as the crops reached the senescent stage (Figures 2 & 3).

Figure 2. ERS-1 SAR temporal backscatter profiles for major crops during the 1993 growing season

Figure 3. ERS-1 SAR temporal backscatter profile for corn

Although Figures 2 and 3 match the general trend described earlier, the radar backscatter coefficients varied considerably from one date to the next. Given the absolute calibration accuracy of +/- 0.42 dB, these variations could not be related to SAR calibration. Attempts to explain the variations based on changes in environmental conditions, such as local meteorological conditions and crop development, were unsuccessful.

Attention was then turned to potential variations due to characteristics of the SAR system. For individual crop types, SAR temporal backscatter profiles were generated separately for each orbit. It was found that the profiles were relatively smooth when the two orbits were separated (Figure 4).

Figure 4. ERS-1 SAR temporal backscatter profiles for corn derived from two orbits: error buffers included

ERS-1 orbital (incidence angle) effects were observed on all crops (Figures 5 - 8). For the 4° difference of incidence angle between the two orbits in the study area (about 21.5° for orbit 2 and 25.5° for orbit 1), the average difference of radar backscatter was approximately

3 dB. This finding is comparable to the results of Ulaby et al. (1986, Figure 9). Using C-HH SAR, Ulaby et al. (1986) measured incidence-angle effects on a corn canopy. The estimated change of backscatter is about 3-4 dB from incidence-angle 20° to 25°.

Figure 5. ERS-1 SAR temporal backscatter profiles for wheat derived from two orbits: error buffers included

Figure 6. ERS-1 SAR temporal backscatter profiles for barley/oats derived from two orbits: error buffers included

Figure 7. ERS-1 SAR temporal backscatter profiles for soybeans derived from two orbits: error buffers included

Figure 8. ERS-1 SAR temporal backscatter profiles for alfalfa/hay derived from two orbits: error buffers included

Figure 9. Measured d ° of a fully mature corn canopy in four consecutive stages of defoliation; all the measurements were made on the same day (Ulaby et al., 1986)

5. Conclusions

Multitemporal radar backscatter characteristics of crops and their underlying soils were analyzed for an agricultural area in southwestern Ontario, Canada. Nine dates of ERS-1 SAR imagery were acquired for two descending passes during the 1993 growing season. These data were corrected for SAR antenna pattern and compensated for range-spreading loss at ESA D-PAF. For major crop types, SAR temporal backscatter profiles were generated. It was found that the radar backscatter coefficients varied considerably from one date to the next. Given the absolute calibration accuracy of +/- 0.42 dB, these variations could not be related to SAR calibration. Attempts to explain the variations based on changes in environmental conditions, such as local meteorological conditions and crop development, were also unsuccessful.

Attention was then turned to potential variations due to characteristics of the SAR system. For individual crop types, SAR temporal backscatter profiles were generated separately for each orbit. It was found that the profiles were relatively smooth when the two orbits were separated. For the 4° difference of incidence angle between the two orbits in the study area, the average differences in backscatter coefficients were approximately 3 dB. These results indicate that small changes in incidence-angle can have strong impacts on radar backscatter. Thus, attention must be given to local incidence-angle effects when using ERS-1 SAR data, especially when comparing backscatter coefficients of the same area from different scenes or different areas within the same scene.

6. ACKNOWLEDGEMENTS

Financial support for this project was provided by a Centre of Excellence Grant from the Government of Ontario to the Institute for Space and Terrestrial Science and through an NSERC Research Grant awarded to P. J. Howarth. Special thanks to ESA/ESRIN ERS Help Desk and ESA D-PAF for their assistance with image calibration.

7. references

Brisco, B, F.T. Ulaby, and R. Protz. 1984. Improving crop classification through attention to the timing of airborne radar acquisitions. Photogrammetric Engineering and Remote Sensing, Vol. 50, No. 6, pp. 739-745.

Brisco, B., R. J. Brown, J. G. Gairns and B. Snider. 1992. Temporal ground-based scatterometer observations of crops in western Canada. Canadian Journal of Remote Sensing. Vol. 18, No. 1, pp. 14-21.

Brown, R. J., B. Guindon, P.M. Teillet and D. G. Goodenough. 1984. Crop type determination from multitemporal SAR imagery. Proceedings, 9th Canadian Symposium on Remote Sensing, St. John's, Newfoundland, Canada, pp. 683-691.

Brown, R.J., R. Leconte, B.G. Brisco, C.A. Hutton, D. Mullins, J.G. Gairns, Q.H.J. Gwyn, R. Protz, J. Fischer, P.J. Howarth, P.M. Treitz, J.B. Boisvert, and K.P.B. Thomson. 1991. Oxford County Soil Moisture Experiment (OXSOME) Overview. Proceedings, 14th Canadian Symposium on Remote Sensing, Calgary, Alberta, Canada, pp. 512-518.

Dobbins, R., K. Korporal, P. Nixon, B. Brisco and R. Brown. 1992. A comparison between multi-date C-HH and C-VV SAR digital imagery for potato crop monitoring. Proceedings, 15th Canadian Symposium on Remote Sensing. June 1-4, Toronto, Ontario, Canada, pp. 245-250.

Fischer, J.A., and R.S. Mussakowski. 1989. Preliminary evaluation of multi-date SAR data for the identification of agricultural crops in Southern Ontario. Proceedings, IGARSS‘89 / 12th Canadian Symposium on Remote Sensing, Vancouver, British Columbia, Canada, Vol. 2, pp. 430-433.

Foody, G.M., P.J. Curran, G.B. Groom, and D.C. Munro. 1989. Multi-temporal synthetic aperture radar data for crop classification. Geocarto International, Vol. 4, No. 3, pp. 19-29.

Laur, H. 1992. ERS-1 SAR Calibration: Derivation of Backscattering Coefficient s ° in ERS-1.SAR.PRI Product, ESRIN/ESA Document, Issue 1, Rev. 0, 17pp.

Laur, H., P. Meadows, J.I. Sanchez, and E. Dwyer. 1993. ERS-1 SAR radiometric calibration. Proceedings of SAR Calibration Workshop: CEOS Calibration Working Group SAR Calibration Sub-Group, pp. 257-281, September 20-24, ESTEC/ESA, Noordwijk, The Netherlands.

Ulaby, F.T., R.K. Moore, and A.K. Fung. 1986. From Theory to Applications. Microwave Remote Sensing: Active and Passive, Vol. III, Addison-Wesley, Reading, Massachusetts.

Keywords: ESA European Space Agency - Agence spatiale europeenne, observation de la terre, earth observation, satellite remote sensing, teledetection, geophysique, altimetrie, radar, chimique atmospherique, geophysics, altimetry, radar, atmospheric chemistry