I am doing analysis of MODIS data in order to build yield model from the Vegetation Indices.
Recently, I used R since I have found it's been used widely by many research regarding time series Remote Sensing.
I found several packages, but it was still confusing me, "what should I do first with my Raster data downloaded from USGS?" before I use the packages to analysis my stack of data.
If you have some best practice tutorial, I will be glad if you can share it to me!
Many thanks!
Regards.
You need to know the seasonal transition parameters in order to extract phenology data out of remote sensing time-series data. I recommend you yo check TIMESAT (http://web.nateko.lu.se/timesat/timesat.asp) for the extraction of these parameters which are mainly: beginning of season, end of season,length of season, time of middle of season, maximum value, amplitude etc. You can model the same approach or try different fittings for your time series data with R but the main idea is explained briefly in the software manual.
After extracting phenological data you can build your own model between yield and vegetation indices.
You can also check our study, performed with Landsat data, which is not totally aiming yield prediction but stating that the use of phenological data for agricultural applications helps to improve accuracy: Conference Paper Phenology based NDVI time-series compensation for yield esti...
You can use TIMESAT package (http://web.nateko.lu.se/timesat/timesat.asp)
Ayda Aktaş Thanks for your information and great answer!
I have used TIMESAT to smooth my time series-raster stack data from MODIS and It works very well!
But, I found other issues regarding extracting seasonal information. What my understanding is that TIMESAT can only extract maximum 2 season within 1 year (or certain time period), while I am willing to gain all of the peak season information within one year.
Do you have any idea how to extract all of the peak season information?
I will be so glad if you can share your experiences.
Thank you.
Ayda Aktaş I have also read your conference paper. It is really interesting! Instead of only using VI values, you used also other meteorological information to build your model.
I look forward to see your other research!
Hello,
I want to extract seasonal metrics of precipitation, evapotranspiration and vegetation indices in TIMESAT, for a long series of 18 years. However, I'm doing a one-year test to see if it runs. I converted the 23 images into the ENVI (BSQ) format, and created the textlist file to input the dataset into TIMESAT.
In the window that generates the curves, it is giving the following error:
Warning - too many missing data points or constant data Pixel will be excluded - check data range and amplitude.
How can you help me, please?
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R