Hii
I need to classify crops in agricultural fields using Remote sensing data(Landsat, Sentinel spatial data). I have tried this using Sentinedl data but facing lot of errors. Can some help me in doing this task ?.
If you tried to supervised classification, you should need more training point. The prior Knowledge is important in classification.Additionally，the machine learning algorithm can improve classified accuracy to reduce misclassification.
Did you already see these excellent tutorials/webinars?
- Crop mapping with Sentinel-2 (Video): https://www.youtube.com/watch?v=jpPoZ6wv9dM
- Crop mapping with Sentinel-2 (PDF): https://rus-copernicus.eu/portal/wp-content/uploads/library/education/training/LAND01_CropMapping_Seville_Tutorial.pdf
- Crop type mapping with Sentinel-1: https://www.youtube.com/watch?v=htme1WfRPh0
Dear Andreas Braun and Xiangyu Ge
Thank you for your reply and valuable information.
Dear Dagani Koteswara Rao,
1. You may follow some norms which might be very much helpful in this regards.
a) Basic Elements of Image Interpretation (EII) such as Shape, Size, Tone, Pattern, Texture & Site will help you to find out fundamentals of crop type.
b) Image Acquisition date will guide you to find out the season of crop from where you will come to know Intra-regional or Inter-regional crop pattern.
c) Calculate NDVI at the time of Peak Vegetative Stage (PVS) and classify as accordingly.
d) NDVI, NDWI, EVI, VHI, VCI etc. varies crop to crop & time to time. Analyze them according to your interest.
e) Knowledge based classification technique may be very useful in this case
f) If the image is more or less homogeneous then OBIA (Object Based Image Analysis) technique will work better to find out crop boundary (using eCognition software).
2. And for crop classification or identification using Optical RS data like Landsat you can follow the following article as well.
Article MAPPING RICE CROPPING SYSTEM IN THE LOWER GANGETIC PLAIN USI...
3. Mr. Andreas Braun shared a very nice talk for Sentinel data.
Thanks in advance.
Regards,
MONIR
You've got to use multitemporal data because some crops differ from others only during specific part of the growing season.
Basically,
you need more sampling. This depends on your knowledge of the types of crops grown there. After successfully completing your classification, endeavor to conduct an accuracy assessment too. Please, see the attached file for the indices that could aid your accurate classifications.
Hello Dagani Koteswara Rao ,
Crop Classification using Sentinel-2 requires basic steps:
a) Basic Processing
b) Reprojection (Highly Recommended)
c) Ground field data of the particular crops
d) Classifier: Random Forest
e) Mask of crops as well as various fileds
Theses are the basic steps required for the crop identification and classification.
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing