Chapter 4 An Integrated Surface Parameter Inversion Scheme over Agricultural Fields
4.2 Study Areas and Data Collection
Two study areas selected for analysis and validation in this chapter are located in the southwest of Ontario, Canada, which are shown as red and blue points in the upper left corner of Figure 4-1. We named these two different study areas as study area 2013 and study area 2014 for convenience, as the ground truth data collected in these two study areas were in 2013 and 2014 respectively. There are three major crops growing in both study areas: soybean, corn and winter wheat. For study area 2013, only two wheat fields that are shown in the Pauli image in the upper right corner of Figure 4-1 are selected, in
which 13 sample sites were surveyed on April 29th and18 sample sites were surveyed on
May 9th in 2013. For study area 2014, five fields were selected including two corn fields,
two soybean fields, and one wheat field, with their polygons displayed in the lower left of Figure 4-1. Their distributions are shown in the Pauli image on the right side of Figure 4- 1. A total of 37 sample sites were surveyed for all fields, including 17 points from the corn fields, 16 points from the soybean fields, and 4 points from the wheat field. The distribution of the sample points of each field in the polygons is shown in the lower left of Figure 4-1. Each site is labeled as the capital letter of the first letter of the crop name, plus the field number, plus a hyphen, and plus the sample site number. Take C1-08, for example, it represents the eighth sample site in the first corn field. The RADARSAT-2
data was acquired from the beginning of May to the end of June in 2014 and on April 29th
and May 9th in 2013, while the fieldwork was performed simultaneously when
RADARSAT-2 was over passing. The fieldwork schedule and the image acquisition dates are shown in Table 4-1.
Table 4-1. RADARSAT-2 dataset and fieldwork schedule. The blue cell represents the data acquired on that day while grey cell means no fieldwork or dataset acquired on that day. MV: soil moisture. KS: surface roughness. The resolution (unit: m) is the one after geo-correction using the Mapready software.
FIELD WORK
RADARSAT-2 DATASET
Corn Soybean Wheat
Date MV KS MV KS MV KS Mode Orbit Look direction Resolution 2014-05-04 FQ15-35° Ascending Right 5 2014-05-05 FQ19-39° Descending Right 5 2014-05-15 FQ09-29° Descending Right 5 2014-05-18 FQ05-24° Ascending Right 10 2014-06-04 2014-06-11 FQ05-24° Ascending Right 10 2014-06-21 FQ15-35° Ascending Right 5 2013-04-29 FQ09-29° Ascending Right 10 2013-05-09 FQ19-39° Ascending Right 10
The soil moisture measurements were collected on April 29th and May 9th in the study area 2013 in the low and sparse wheat covered fields that are shown in Figure 4-1. For study area 2014, the soil moisture was collected on five days: May 4th, May 5th, May 18th, June 4th and June 21st, whereas the surface roughness was collected on May 5th, May 18th and June 4th in the soybean and corn fields. The soil moisture was not collected in the wheat field on June 21st because the wheat was already very high and dense with high biomass by then; the penetration of the short wavelength C-Band RADARSAT-2 sensor is limited when the agricultural field is under the dense wheat canopy cover due to the strong attenuation effects (Lopez-Sanchez & Ballester-Berman, 2009). For the same reason, the surface roughness was not measured in the wheat field on June 21st either.
Although fieldwork was conducted on June 4th, no RADARSAT-2 data was available on
this date. The associated ground truth photos are shown in Figure 4-2. It can be seen that at the beginning of May 2014, the soybean fields were not cultivated, with many corn residues from the previous year left on the ground; whereas the corn fields were mainly bare soils, although a few crop stalks were present. The wheat field was in the tillering stage with very low height, and there were still a lot of crop residues present, as can be seen in Figure 4-2(c). In mid-May, many crop residues in the soybean had been flattened
due to human activities. At the end of May, both the soybean and corn fields were under seedbed preparation. At the beginning of June, the corn had emerged and at early vegetative growth stage with very low vegetation cover fraction, whereas the wheat is growing taller and denser, as shown in Figure 4-2(c). Till the end of June, the corn continues to grow taller, and the soybean was budding in low height, as shown in Figure 4-2(a) and Figure 4-2(b).
4.2.2 Ground Truth Measurement
For ground truth measurements, soil moisture and surface roughness were measured in these fields during the early growing stages. Soil moisture was measured using the TDR (Time-Domain Reflectometry) Probe for all the sample sites, with an average of 6 points measured at each sample site within a 10 m by 10 m area surrounding the centroid of the sample site. The surface roughness was measured for only half of the sample sites in the corn and soybean fields using a one-meter long profiler with 200 pins and an interval of 0.5 cm. For the corn field, because it had been ploughed before May, there were many large clods in the field, and rain events made the clots smooth without obvious oriented roughness patterns. Hence, the surface roughness measurement was randomly taken 6 times from the relatively smooth and rough surfaces within a 10 m by 10 m square surrounding the centroid of the sample site, and their average is taken as the value of the roughness for that site. For the soybean field, the roughness was measured in the same way with disregard for corn residue.
For the fields covered with standing corn stubbles or vegetation, the height of the stubble or vegetation were also measured simultaneously. The ranges of the soil moisture measurements, the root mean square (RMS) of surface height and the height of vegetation or corn stubble at the early growing stage are listed in Table 4-2. The RADARSAT-2 data was pre-processed with the radiometric correction performed first to covert the data to sigma naught, i.e., the backscattering coefficient. It was then filtered using the Boxcar method with the window size of 5 by 5, and geo-corrected using the MapReady software developed by the Alaska Satellite Facility (ASF) with the resolution after geo-correction shown in Table 4-1.
Table 4-2. Measured ground truth on different dates in different agricultural fields. MV: soil moisture ([vol.] %). RMS: root mean square of the surface height (cm). H: height of vegetation or corn residue (cm). Note: “-“ means no data was collected on that day. The height measured in soybean field before June 4th 2014 is the height of the standing corn stubbles.
Date Corn Field Soybean Field Wheat Field
MV RMS H MV RMS H MV RMS H 2014-05-04 12-48 - - 21-50 - 15-45 37-49 - 8-11 2014-05-05 8-45 1.5-5.3 - 15-50 1.4-4.0 15-45 15-41 - 8-11 2014-05-18 14-44 1.3-4.5 - 28-50 1.4-3.2 15-45 35-50 - 13-18 2014-06-04 11-26 1.1-2.3 5-8 15-40 1.4-2.5 - 23-33 - 20-35 2014-06-11 - 1.1-2.3 7-13 - 1.4-2.5 2-3 - - - 2014-06-21 5-25 - 20-26 5-35 - 5-8 - - - 2013-04-29 - - - 19-50 - 5-8 2013-05-09 - - - 13-34 - 10-15
Figure 4-1. Study area locations, field polygons and Pauli images (right) from the fully polarimetric RADARSAT-2 data in 2013 and 2014. The Pauli image in 2013 is acquired on April 29th 2013 while the one at the bottom is acquired on
May 5th 2014. It should be noted that only wheat fields are measured in 2013
(a)
(b)
(c)
Figure 4-2. Ground truth photos in the corn, soybean and wheat fields on different dates in 2014: (a) corn fields (b) soybean fields (c) wheat fields. From left to right, the photos of the soybeans and corn were taken on May 5th, May 18th and
June 21st, respectively, whereas the photos of the wheat were taken on May
5th, May 18th and June 11th.