Assessment of interferometric coherence contribution to corn and soybean mapping with Sentinel-1 data time series

Figures & data

Figure 1. Study workflow.

Figure 2. On the right side, in yellow, the study area location in the north of Buenos Aires province. On the left side, study area extension represented by the yellow rectangle (background: ‘color-infrared’ composite from Landsat-9 images acquired on March 27, 2022).

Table 1. Polygons of the classes used in the present study for both classifications training and assessment.

Sentinel-1 satellite/product ID	Acquisition date [day month year]	Sentinel-1 satellite/product ID	Acquisition date [day month year]
S1B/0C1E	14 October 2019	S1A/4011	05 April 2020
S1B/958F	26 October 2019	S1B/6043	11 April 2020
S1A/417C	01 November 2019	S1A/F2EF	17 April 2020
S1B/712E	07 November 2019	S1B/4A2E	23 April 2020
S1A/837B	13 November 2019	S1A/8AA7	29 April 2020
S1B/FFB4	19 November 2019	S1B/B763	05 May 2020
S1A/CE7D	25 November 2019	S1B/BC53	17 May 2020
S1B/8254	01 December 2019	S1A/8D4A	23 May 2020
S1A/5096	07 December 2019	S1B/DC25	29 May 2020
S1B/B4FE	13 December 2019	S1A/57FC	04 June 2020
S1A/4249	19 December 2019	S1B/38B2	10 June 2020
S1B/D1D0	25 December 2019	S1A/9DCE	16 June 2020
S1A/22A4	31 December 2019	S1B/AC28	22 June 2020
S1B/3241	06 January 2020	S1A/AD0F	28 June 2020
S1A/D087	12 January 2020	S1B/C46D	04 July 2020
S1B/2F95	18 January 2020	S1A/CD28	10 July 2020
S1A/A4AA	24 January 2020	S1B/69B3	16 July 2020
S1B/3B2A	30 January 2020	S1B/E823	28 July 2020
S1A/DD17	05 February 2020	S1A/6243	03 August 2020
S1B/88C0	11 February 2020	S1B/2B8C	09 August 2020
S1A/22ED	17 February 2020	S1A/A5EA	15 August 2020
S1B/7184	23 February 2020	S1B/3B4E	21 August 2020
S1A/0E2E	29 February 2020	S1B/E823	28 July 2020
S1A/C38B	12 March 2020	S1A/6243	03 August 2020
S1B/0805	18 March 2020	S1B/2B8C	09 August 2020
S1A/A84E	24 March 2020	S1A/A5EA	15 August 2020
S1B/CB39	30 March 2020	S1B/3B4E	21 August 2020

Table 2. Polygons corresponding to the whole dataset of evaluated classes.

Class	Number of polygons	Area (Ha)
Soybean	114	2794.5
WC-Soybean	123	3708.7
Early Corn	120	2793.8
Late Corn	111	2227.5
WC-Corn	32	772.6
Rangelands	105	2873.3

Figure 3. The five upper plots correspond to the time series of backscatter coefficient of the classes at VV (magenta) and VH (green) polarization. Points represent the average values, while vertical lines represent standard deviation for all polygons of each class. For crop classes, horizontal line represents the seasonal crop calendar with the seeding period (light blue), vegetative stage (green), reproductive stage (yellow) and harvest period (brown). The lower plot corresponds to the daily precipitation data registered by the Pergamino (red) and Arrecifes (blue) stations.

Figure 4. The five upper plots correspond to the time series of coherence of the classes at VV polarization. Points represent the average values, while vertical lines represent standard deviation for all polygons of each class. For crop classes, horizontal line represents the seasonal crop calendar with the seeding period (light blue), vegetative stage (green), reproductive stage (yellow) and harvest period (brown). The lower plot corresponds to the daily precipitation data registered by the Pergamino (red) and Arrecifes (blue) stations.

Table 3. Randomly selected polygons from the whole dataset according to the classifier training strategy (points 1 and 2 of Section 2.3.3).

Class	Number of polygons	Area (Ha)
Soybean	32	775.0
WC-Soybean	28	805.4
Early Corn	32	737.4
Late Corn	37	741.4
WC-Corn	32	772.6
Rangelands	28	791.8

Figure 5. Example of the proposed crop-type map. It corresponds to the obtained using the ${\mathrm{\sigma }}_{\text{VV}}^0$ + ${\mathrm{\sigma }}_{\text{VH}}^0$ + ${\mathrm{\gamma }}_{\mathrm{VV}}$ SAR features for Range 4 to train the classifier. Urban areas were masked out with the Global Man-made Impervious Surface (GMIS) Dataset from LANDSAT (Brown de Colstoun et al. 2017).

Table 4. Overall accuracy and kappa index scores corresponding to each classification test for the analysed time ranges.

Classifier training input	Range 1		Range 2		Range 3		Range 4
	Overall accuracy (%)	Kappa index	Overall accuracy (%)	Kappa index	Overall accuracy (%)	Kappa index	Overall accuracy (%)	Kappa index
${\mathrm{\sigma }}_{\text{VV}}^0$	95.41	0.94	84.55	0.81	91.68	0.90	95.14	0.94
${\mathrm{\sigma }}_{\text{VH}}^0$	96.48	0.96	87.35	0.85	93.50	0.92	96.14	0.95
${\mathrm{\gamma }}_{\mathrm{VV}}$	88.17	0.85	76.89	0.72	77.68	0.73	86.77	0.84
${\mathrm{\sigma }}_{\text{VV}}^0$ + ${\mathrm{\gamma }}_{\mathrm{VV}}$	97.06	0.96	88.10	0.85	93.89	0.92	96.70	0.96
${\mathrm{\sigma }}_{\text{VH}}^0$ + ${\mathrm{\gamma }}_{\mathrm{VV}}$	96.95	0.96	89.34	0.87	94.28	0.93	96.82	0.96
${\mathrm{\sigma }}_{\text{VV}}^0$ + ${\mathrm{\sigma }}_{\text{VH}}^0$	97.15	0.96	89.49	0.87	94.45	0.93	96.97	0.96
${\mathrm{\sigma }}_{\text{VV}}^0$ + ${\mathrm{\sigma }}_{\text{VH}}^0$ + ${\mathrm{\gamma }}_{\mathrm{VV}}$	97.48	0.97	90.16	0.88	94.76	0.94	97.29	0.97

Table 5. User’s and producer’s accuracy for all classification test corresponding to each evaluated classification set.

Classifier training input	User’s accuracy (%)						Producer’s accuracy (%)
	S	WC-S	EC	LC	WC-C	R	S	WC-S	EC	LC	WC-C	R
	Range 1
${\mathrm{\sigma }}_{\text{VV}}^0$	96.8	98.5	96.8	96.2	59.1	97.1	98.3	96.0	97.2	86.7	88.7	97.7
${\mathrm{\sigma }}_{\text{VH}}^0$	98.5	98.3	98.8	95.2	64.8	98.0	97.2	98.0	98.2	87.6	90.8	99.7
${\gamma }_{\mathrm{VV}}$	96.6	89.7	92.3	80.5	38.7	97.1	94.7	83.9	95.9	76.6	85.3	89.2
${\mathrm{\sigma }}_{\text{VV}}^0 + {\gamma }_{\mathrm{VV}}$	97.9	99.0	98.3	97.0	66.6	99.4	98.8	97.5	99.3	91.6	90.7	97.7
${\mathrm{\sigma }}_{\text{VH}}^0 + {\gamma }_{\mathrm{VV}}$	98.6	98.7	99.0	95.6	67.6	98.4	97.7	97.9	99.0	89.8	91.7	99.3
${\mathrm{\sigma }}_{\text{VV}}^0 + {\mathrm{\sigma }}_{\text{VH}}^0$	98.9	98.6	98.7	95.8	69.8	99.1	98.1	98.2	98.6	90.1	91.0	99.7
${\mathrm{\sigma }}_{\text{VV}}^0 + {\mathrm{\sigma }}_{\text{VH}}^0 + {\gamma }_{\mathrm{VV}}$	98.9	98.6	99.0	96.7	69.6	99.6	98.2	98.4	99.4	91.0	91.4	99.5
	Range 2
${\mathrm{\sigma }}_{\text{VV}}^0$	92.9	92.2	91.3	79.8	29.1	87.7	93.9	74.9	93.4	69.2	75.9	92.9
${\mathrm{\sigma }}_{\text{VH}}^0$	96.6	93.0	95.0	81.1	30.4	96.6	93.1	71.1	97.8	83.6	85.6	96.6
${\mathrm{\gamma }}_{\mathrm{VV}}$	93.3	81.2	89.4	63.9	26.1	79.3	90.8	61.7	92.4	61.5	75.6	93.3
${\mathrm{\sigma }}_{\text{VV}}^0 + {\mathrm{\gamma }}_{\mathrm{VV}}$	94.7	93.4	95.3	86.8	34.3	89.8	95.8	77.3	97.8	76.7	77.7	94.7
${\mathrm{\sigma }}_{\text{VH}}^0 + {\mathrm{\gamma }}_{\mathrm{VV}}$	97.0	94.2	95.4	84.8	34.9	96.6	93.7	76.4	98.3	85.4	85.8	97.0
${\mathrm{\sigma }}_{\text{VV}}^0 + {\mathrm{\sigma }}_{\text{VH}}^0$	96.3	94.9	95.3	85.5	33.1	97.8	94.5	76.8	98.6	84.0	85.1	96.3
${\mathrm{\sigma }}_{\text{VV}}^0 + {\mathrm{\sigma }}_{\text{VH}}^0 + {\gamma }_{\mathrm{VV}}$	96.6	95.0	95.5	85.3	36.4	97.8	94.8	78.7	99.0	84.4	84.1	96.6
	Range 3
${\mathrm{\sigma }}_{\text{VV}}^0$	93.7	97.0	94.5	87.1	46.7	95.9	97.8	89.5	96.6	76.4	78.5	97.6
${\mathrm{\sigma }}_{\text{VH}}^0$	97.7	97.4	97.7	85.4	52.6	97.5	96.9	90.3	97.5	83.3	87.1	99.4
${\gamma }_{\mathrm{VV}}$	93.9	84.5	91.2	62.1	15.6	95.8	90.7	55.7	95.4	68.4	62.1	86.9
${\mathrm{\sigma }}_{\text{VV}}^0 + {\gamma }_{\mathrm{VV}}$	96.5	97.6	96.2	88.4	56.8	97.9	97.3	91.6	98.2	84.9	86.0	97.5
${\mathrm{\sigma }}_{\text{VH}}^0 + {\gamma }_{\mathrm{VV}}$	97.7	97.1	97.9	87.3	55.8	98.8	96.8	91.1	99.0	85.2	88.4	99.3
${\mathrm{\sigma }}_{\text{VV}}^0 + {\mathrm{\sigma }}_{\text{VH}}^0$	97.5	97.7	97.2	87.7	56.7	99.2	97.3	92.4	99.0	83.1	89.2	99.5
${\mathrm{\sigma }}_{\text{VV}}^0 + {\mathrm{\sigma }}_{\text{VH}}^0 + {\gamma }_{\mathrm{VV}}$	97.4	97.5	97.4	88.3	59.0	99.6	97.3	92.7	99.4	84.4	89.2	99.4
	Range 4
${\mathrm{\sigma }}_{\text{VV}}^0$	96.4	98.7	95.2	95.9	60.5	97.0	98.4	95.7	97.2	85.7	88.8	97.4
${\mathrm{\sigma }}_{\text{VH}}^0$	98.5	98.4	98.2	94.0	63.5	97.9	97.2	97.3	97.9	87.1	91.1	99.6
${\gamma }_{\mathrm{VV}}$	96.5	91.4	92.8	75.3	32.1	97.4	94.7	77.6	96.1	79.4	80.7	88.9
${\mathrm{\sigma }}_{\text{VV}}^0 + {\gamma }_{\mathrm{VV}}$	97.8	98.9	97.6	96.3	66.7	99.0	98.7	96.7	98.8	91.3	90.7	97.7
${\mathrm{\sigma }}_{\text{VH}}^0 + {\gamma }_{\mathrm{VV}}$	98.6	98.5	98.4	94.2	70.4	98.9	97.3	97.2	98.8	90.8	92.5	99.2
${\mathrm{\sigma }}_{\text{VV}}^0 + {\mathrm{\sigma }}_{\text{VH}}^0$	98.8	98.5	97.9	96.0	68.2	99.3	97.9	98.1	98.9	88.9	91.7	99.6
${\mathrm{\sigma }}_{\text{VV}}^0 + {\mathrm{\sigma }}_{\text{VH}}^0 + {\gamma }_{\mathrm{VV}}$	98.9	98.7	98.0	95.4	73.3	99.5	97.7	97.9	99.3	91.2	92.2	99.5

S: Soybean; WC-S: WC-Soybean; EC: Early Corn; LC: Late Corn; WC-C: WC-Corn; R: Rangelands.

Brown de Colstoun EC, Huang C, Wang P, Tilton JC, Tan B, Phillips J, Niemczura S, Ling P-Y, Wolfe RE. 2017. Global man-made impervious surface (GMIS) dataset from Landsat. Palisades, NY: NASA Socioeconomic Data and Applications Center (SEDAC); [accessed 2021 Oct 19].

 Google Scholar