Earth observation satellite data
RADARSAT-2
• Space Agency:
Canadian Space Agency (CSA) within the SOAR-JECAM project no. 5102 “SAR parameters optimization for crop classification”.
• Type:
SAR
• Number of scenes:
4
• Range of dates:
27 and 30 July 2012
• Beam modes/ incidence angles/ spatial resolutions:
- 27 July 2012: FQ8, 26.12-29.40, 5 m
- 30 July 2012: FQ21, 38.59-42.16, 5 m
• Scenes id:
-RS2_OK31850_PK312741_DK278901_FQ8W_20120727_154010_HH_VV_HV_VH_SLC
-RS2_OK31850_PK312742_DK278902_FQ8W_20120727_154014_HH_VV_HV_VH_SLC
-RS2_OK31850_PK312743_DK278903_FQ21W_20120730_155240_HH_VV_HV_VH_SLC
-RS2_OK31850_PK312744_DK278904_FQ21W_20120730_155243_HH_VV_HV_VH_SLC
Radarsat-2, 2012-07-30
• Processing level:
SLC (single look complex)
• Pre-processing:
- Images were filtered using refined Lee filter (7 by 7 pixels) and were converted to Pauli decomposition (Cloude and Pottier, 1996).
- Filtered, ortho-rectified, and converted to Pauli decomposition
• Examples:
- r2.pdf
- r2-no-ground.pdf
EO-1
• Space Agency:
NASA
• Type:
Optical
• Number of scenes:
2
• Range of dates:
- 28 July 2012
- 05 August 2012
EO-1, 2012-08-05
• Look angle:
-26.71, spatial resolution: 30 m
• Scenes id
- EO1A1810252012210110KF
- EO1A1810252012218110KF
• Pre-processing:
Data were converted to TOA reflectance (Chandler et al., 2009)
• Examples:
- eo1.pdf
- eo1-no-ground.pdf
Ground observations to support satellite images were carried out on 4 August 2012 to collect in situ measurements. In total, information on 271 fields was collected. The following crop types were present: maize, soy beans, sunflower, sugar beet, harvested winter and spring crops, and minor crops like buckwheat.
Table 1 – Ground survey statistics in 2012
| Crop | Fields observed, % | Total area (ha), % |
| Maize | 63 (23.2%) | 6264 (29.5 %) |
| Soy beans | 46 (17.0%) | 3553 (16.7 %) |
| Sunflower | 19 (7.0%) | 1722 (8.1 %) |
| Sugar beet | 12 (4.4%) | 1130 (5.3 %) |
| Winter and spring crops (already harvested) | 79 (29.2 %) | 6184 (29.1 %) |
| Non agricultural land and minor crops | 52 (19.2 %) | 2369 (11.2 %) |
| Total | 271 (100 %) | 21222 (100 %) |
Crop classification map:
• Data sources:
EO-1
- 28 July 2012
- 05 August 2012
Radarsat-2
- 27 July 2012
- 30 July 2012
• Classification algorithm
- Neural network: MLP with cross-entropy error function
- Maps was created
• Examples:
- classif.pdf
Classification map of Radarsat-2 and EO-1/ALI images using ensemble of MLPs
Major crop types:
- Maize
- Soybeans
- Sunflower
- Sugar beet
- Winter and spring crops (wheat)
Sugar beet
Sunflower
Maize
Soy beans
