Calculate NDVI from Sentinel 2 and reclassify in QGIS
Вставка
- Опубліковано 27 сер 2024
- This video shows how to download a Sentinel 2 image from USGS Earth Explorer (earthexplorer...., use QGIS to stack the bands in a virtual raster, use the Raster calculator to calculate NDVI and reclassify the result.
QGIS 3.16.5 was used, because it's the current LTR version.
This video is part of IHE Delft OpenCourseWare and was included in the Training of Trainers 1 course for the WaterPIP Knowledge Hubs.
More free materials are available at the GIS OpenCourseWare platform: http:/www.gisopencourseware.org
Apologies for the sound quality...
#IHEDelft #QGIS #GISOCW
This helped immensely for my own little vegetation analyses, thank you so much!
Thank you so much! I learned a lot and it is very easy to follow!
Thank you so much for the video! it was very useful and easy to understand, it helped me a lot!
Excuse me. If i want to get NDVI data from each pixel by using sentinel 2 data how can i do? or it can't (I try to use sentinel 2 data to calculate carbon collected from tree in small space) (Sorry if my eng was bad) Thank you!!
There isnt a Sentinel option anymore, any idea why?
Very good video
when you do the ndvi do you need to save it within the folder of the Sentinel data e.g., the one starting with S2B_MSI etc.,? Whenever I do the NDVI calculation it never comes up with -1 to 1 as the limits - always random numbers in between this. Do you know why this might be please?
It is always recommended to save results to a file instead of using temporary outputs. The format of he the temporary outputs might add out of range nodata values. Also you might lose temporary data after closing QGIS.
thanks a lot , very useful and clear
Thanks! Excellent video.
One Q: Should we first calibrate the L1 image to surface reflectance?
Good question. Because NDVI is a ratio it's not needed. Only more sophisticated atmospheric corrections will have impact on the result.
Hi, does this method only work for Sentinel 2 L1C data or does it also work for L2A? I have very differing results (some working, some anomalous) and I am wondering if it is because I used Sentinel L2A?
The method can be applied to Sentinel2 L1C, but the result will be less accurate. L1C is not atmospheric corrected and has top of atmosphere (TOA) reflectences. L2A is atmospheric corrected and recommended to use.
A lot thanks, Regards for you Sir
When calculating the NDVI in the raster calculator the output is never -1 to 1 - it’s always something like -0.7 to 0.6 which really messes up my NDVI map. Is there anything I can do to fix this? Thank you for all your responses to my questions!!
It doesn't have to be necessarily from -1 to 1 if that's not in your data. But you can set the min/max value in the layer styling panel if you need to stretch the colours over the full range.
@@HansvanderKwast when I change it it essentially just makes the map all one colour as it is sensing that the NDVI is all between a certain range and makes it impossible to use. Is it possible the satellite is erroneous? I am doing a time series of a burn scar and the ndvi goes from severely bare soil to dense vegetation back to severely bare soil in the space of 2 months post burn (which is impossible!!). I have no idea what is causing the erroneous output.
@@megan4399 Clouds?
Dear Dr. Hans van der Kwast. Thank you so much for such a wonderful video. I was comparing Landsat 9 (OLI) and Sentinel 2 (Level 2A) BOA images. I used different vegetation indices including NDVI and MNDWI. However, the BOA (L2A) NDVI looks weird. Do you recommend using Level 1C or L2A for vegetation indices? Thanks
How to classify and put the scale range for mangrove vegetation
Hi, thanks for your great video. When I did the raster calculation for NDVI it did not work. Appeared to calculate but did not show a layer. I checked correct projection and correct expression. Any ideas where I have gone wrong. Thanks
If it calculated, you might find the file on your disk.
Can determine the area by color in hectares?
Thank you very much for this tutorial. One question: how can we calculate annual NDVI or monthly NDVI?
This video by my colleague explains how to do that with r.series: ua-cam.com/video/MinaylWMa3Q/v-deo.html
Hi thanks for the video. I just tried uploading a watershed KML file and I got an error telling me that the file is greater than 500 points. Is there a workaround to this limitation? thx
For downloading the satellite image product from Earth Explorer, you only need a simple bounding box polygon. Earth Explorer will not clip the image for you, it will search for tiles that match with the extent of your KML. So you can just upload a simple bounding box in KML or shapefile format or zoom in to your area of interest and use that extent on the web page.
Hi Hans. Thanks for that video. How can I extract a % per band?? I need to clasify a percentage per band to monitor a crop over time?
Not sure what you mean with % per band. NDVI is a single band raster. You can study changes over time by calculating NDVI for images of different dates. If you reclassify the values in the same way, you can calculate % change for each class. Not sure if you want that.
@@HansvanderKwast Yes I want to select a field and break down the NDVI values per pixel into perhaps 5-10 categories. Using this data I should be able to track the progress of the crop? I'm a real beginner to all this. I've just watched your percentage of land use video and I think this is the direction but in a smaller scale.
Thank you for the highly helpful video! I want to do the same procedure but with data for the country of germany. I can't find a fitting dataset on earthexplorer and the website won't let me upload the shapefile of germany since the polygone has more than 30 lines. Do you know where I can get the data? Again thank you for your help!
You just need a rectangle around your study area. You can create one in QGIS. Or you use the map in Earth Explorer instead of a file.
@@HansvanderKwast I tried using the map and drew a large rectangle around germany but earthexplorer gave me nearly 88.000 results with small areas that lie within my rectangle and not a dataset that most precisely resembles my rectangle.
If you want the whole of Germany with Sentinel you need to use Google Earth Engine. Otherwise go for coarser images, like MODIS. The problem is not the data, but what you want to achieve and then what to use for that.
thank you so much for the video. Do you know how to combine sentinel 2 images with black parts in the same raster??
Best to check the tutorials on the Semi-Automatic Classification Plugin. For example this one: fromgistors.blogspot.com/2018/04/cloud-masking-image-mosaic-and-land-change.html
Thank u.
Do we not need to apply corrections?
It depends. If atmospheric conditions are homogeneous you don't need to apply atmospheric correction or convert to surface reflectance. That is because the index takes the ratio of bands.
the link is not working