Google Earth Engine - Rice/Paddy Crop Classification using Sentinel-1 SAR data

In this video complete process of Rice crop classification is explained using Sentinel-1 SAR data in Google Earth engine.
Google Earth Engine combines a multi-petabyte catalog of satellite imagery and geospatial datasets with planetary-scale analysis capabilities.
Code Shown in the video:
West Bengal - Birbhum district
code.earthengine.google.com/e...
Haryana - Karnal District
code.earthengine.google.com/d...

[@OpenGeoLab]

Just remove one bracket ( ) from both sides in .or filter of ascending and descending...in the code given in the description of video

[@prof.biplabbiswas5576]

Excellent tutorial -as always. Many Thanks for giving us so interesting topic for research.

[@DrZubairulIslam]

Thanks Buddy. Great video.

[@pritammeshram3633]

Very useful 👍

[@sewunetshiferaw1117]

I found your lab very useful. Can you please put the code how to use all the classes in different color, so that i will have all the pixels classified in the area. thank you in advance

[@Agritechnika]

Sir, is there any dataset regarding rainfall at particular date in a district, so that we can avoid the errors that may occur by rain water deposition in the field by not taking the image acquired at that day or nearby. Also thank you sir for the wonderful video. This was extremely helpful.

[@user-sx3ho8oi8m]

Hi how can i export the classified rice area to vector and raster?

[@user-lh6uv4co7r]

Thank u so much for making this useful tutorial.I have a question of "Unable to perform this geometry operation. Please specify a non-zero error margin." someone said decrease the size of training data will help.It works but i don't know why,does gee limit the size of training data?Thanks a lot!

[@pritamdas8582]

Hi, I appreciate your efforts for making this type of hands-on exercise and descriptive videos. The same code and methodology have been deployed on the same district 'Birbhum' for year 2020 for Aman/summer season i.e. having july-Aug transplanting date. I ran the code and it gave me 301,220 Ha of rice has been planted, but the govt economics and statistics "Crop Production Statistics" says that this district had only 77286 Ha of rice in 2020. And in your video too shows that 180457 Ha in 2021! That value is larger than the govt statistics. What I'm missing? How we can validate our findings with govt statistics? Kindly send reply asap. Thank you in advance.

[@VinodKumar-rd4ek]

Sir, can you share the methodology of crop classification as NRSC 3 level classification only show rabi and kharif crops, but not wheat and rice. Thanks in advance

[@IAKhan-km4ph]

Thank you . I got the error: Dictionary (Error) Image.multiply: If one image has no bands, the other must also have no bands. Got 0 and 1.

[@kaustubhmujumdar3062]

Nice and thought provoking videio!! We have passed stacked bands like ['VH2', ;'VH3',' 'VH9] through RGB channels, considering 'VH3' as a rice transplanting date, and then picked the magenta colour. So, VH9 i.e.oct21,1st is exactly which time period considering rice crop phenological cycle? If we consider VH1/VH2 rice i.e. rice having transplanting dates in june, then whether VH9 corresponds to maturity or fully ripen stage of that rice. If this is the consideration, then for rice fields having late transplanting dates in Aug, / sept. (in Tamilnadu) should we modify VH9 to VH10, VH11 etc., in order to effectively catch magenta colour

[@user-sx3ho8oi8m]

how can i export it to raster or vector

[@jibanray6713]

Sir..how to differentiate Weed and Rice crop, please help me out.

[@VinodKumar-rd4ek]

Sir, can u please tell any methodology given by any institute to classify according to crops like wheat and rabi and rice in kharif. Because NRSC given only only three level classification till rabi and kharif crop. Please guide.

[@OpenGeoLab]

NRSc creates Land use land cover with major classes I.e agriculture urban water bodies etc.. If you want to classify crops I.e wheat and rice...you need some ground truth data regarding crop I.e latitude and longitude of the crop fields and after overlaying those on the satellite image...you can create training dataset and this training data can be used to train machine learning algorithm I.e random forest SVM neural network and and crop classification acn be performed on the satellite image. Both single date imagenor multi temporal images both can be used for classification This whole.process can be carried out in Google earth engine for free..using sentinel data which is available.innearth engine..algorithms I.e random forest and Suppprt vector machine is available in the earth engine...you can searchbon Google image classification in earth engine and you will find many examples...

[@shahabadeli2517]

thank you for this tutorial... i have this error when I run code: Dictionary (Error) Image.multiply: If one image has no bands, the other must also have no bands. Got 0 and 1. I remove one bracket also.

[@mudasaramjad6684]

Sir thanks a lot for such a useful tutorial. I have a question that if we want to do more crop classification such as wheat , cotton rice, sugarcan and also we don’t have field base data. In such case how can we create our training data set ? Can we also find Ammonia emission from the crops using Google earth engine ?? Please answer my this question. Thanks

[@OpenGeoLab]

You can use multi temporal NDVI data over the crop growing season and based on temporal NDVI profile...different crops can be classified..but still some ground information is needed to understand the phenology of the crop... Check this video kzbin.info/www/bejne/h6XMZYKEodmFa9k

[@satishs-xx6ik]

how to get the data set

[@lucasamorim6788]

Thank you very much! I tried it in southern Brazil, but the stacked image shows rice fields in yellow, not in cyan. Do you know why this happens? Thanks a lot!

[@lucasamorim6788]

I used different dates (from january 1st to may 31st) and now the stacked image shows the rice fields in cyan.

[@OpenGeoLab]

Yellow color means ....transplanting of Rice is in the third date of the 3 date of the band combination...we are passing three SAR images to Red Green and Blue filter...in your case..transplanting is in the last date which is passed through Blue band..so SAR backscatter is low in Blue band and high is Red and green...and Red and green color combination gives yellow..it is all about band combination.. If you move the last date to Red band and add two subsequent dates images to green and blue..band...then rice area will look cyan..because now...low backscatter image is passed through Red band and backscatter is high in dates passed through green and blue band filter...so green + blue color is cyan. If transplantation date image passed to Green band...then rice area look like magenta..because Red + blue color is magenta... All three clours also possible in the image if there is transplanting of rice is done in all the three dates passing through RGB filter.

[@OpenGeoLab]

Yellow color means ....transplanting of Rice is in the third date of the 3 date of the band combination...we are passing three SAR images to Red Green and Blue filter...in your case..transplanting is in the last date which is passed through Blue band..so SAR backscatter is low in Blue band and high is Red and green...and Red and green color combination gives yellow..it is all about band combination.. If you move the last date to Red band and add two subsequent dates images to green and blue..band...then rice area will look cyan..because now...low backscatter image is passed through Red band and backscatter is high in dates passed through green and blue band filter...so green + blue color is cyan. If transplantation date image passed to Green band...then rice area look like magenta..because Red + blue color is magenta... All three clours also possible in the image if there is transplanting of rice is done in all the three dates passing through RGB filter.

[@agyaat33]

Nice video Sir please tell me how much work can we do using google earth search engine? can we monitor crop diseases, thermal maps etc. please reply

[@OpenGeoLab]

In Google earth engine whole lot of satellite data is available.....like MODIS, Sentinel, landsat having thermal as well as VNIR data..one can do analysis for pest attacks i.e white fly, wheat rust etc..using temporal NDVi data

[@jibanray6713]

How can i contact you sir...i.need your help to Identify weed in Agricultural Field (Rice) ising satellite data (Sentinel-2). Please send me your contact or make a full video sir

[@simmirajpal]

code is considering only descending images , whereas on Google developers they have considered mean composite of both ascending & descending .. plz specify

[@OpenGeoLab]

Just remove one bracket ( ) from both sides in or filter of ascending and descending...it may work...

[@pikineestest7362]

Thanks You Sir. But when i try to print stacked variable in my polygon (municipality in Senegal) it return me 0 bands. Thanks for helping.

[@OpenGeoLab]

Can you share the link for your code....or area of interest..i will try to run at my end...

[@pikineestest7362]

Hello Sir I found this error, I just delete a Descending option and keep Ascending option. For another people who get this error. Thanks so much.

[@OpenGeoLab]

Just you have to remove one bracket both sides() in or filter ascending and descending ..

[@sewunetshiferaw1117]

oh nice commenting the mask works

[@kutuberoy]

Hello, How can we get this data and the notebooks. I click on the link but the data does not appear

[@OpenGeoLab]

You have to sign up for Google earth engine using your Gmail account...then u will be able to see the code ..after that click on run button to execute the code..

[@kutuberoy]

@@OpenGeoLab I am already signed up and everything . I can see other things, but not this data. Can you quickly help me set up ? I can call you.

[@OpenGeoLab]

After executing the code...check the layers panel on the top right of the map...u can switch on and switch off the layers...bottom layer is SAR data..

[@nilkanthherbha6657]

Hello, I appreciate your efforts for making knowledgeable videos. How can I contact you? Please share LinkedIn if possible

[@OpenGeoLab]

Email is [e-mail]