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.
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.
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.