2

in short: My TIMESAT image-output of the seasonal parameter “Large Seasonal Integral” is wrong. TIMESAT continues to write a file with negative values (as 16 byte Integer) although Parameters should be positive and are positive in txt output files (TSM_printseasons). I do have the feeling that this is due to a scaling of the image or wrong dimensions in the header.

The “Small Seasonal Integral” image output is actually wrong as well, but if I define the image as UNSIGNED LONG INTEGER instead of a signed 16 Byte Integer as it supposed to be, the values of the output image are correct. This does not work with “Large Seasonal Integral”. Here , values are positive, but do not correlate to the txt output of TSM_printseasons.

This is the GUI view of my settings file within TIMESAT (looking quite good and neat ) :

enter image description here

This is the txt output of my time series which makes absolute sense. Please note the bigger Large Seasonal Integrals and the smaller Small Seasonal Integral (highlighted row is the 7th season Pixel 999.999 or 998,998 depending on which programm is used):

enter image description here

So if I open all Seasonal Parameters in ENVI, everything fits perfectly to the text output except the Integrals:

enter image description here

I smelled that something is wrong with the scaling, so I changed the header info to unsigned Long integer and then, the Small Integral was right as well. But not the Large Scale, although it changed to positive!

enter image description here

Now if I go back to TIMESAT and display (TSM_imageview) the binary image (meaning no header involved or whatsoever), I have to realize, that the binary image itself contains negative values! Although the text output of TIMESAT gives positive values!

enter image description here

So my colleagues and me are really puzzled. What did I do wrong? Why is it, that only one value is wrong? because the numbers are so large? How should the header look like? Might the problem sit there? But why is the binary image displayed in TIMESAT wrong itself?

Let me also add a screenshot of the header although I find it strange that the results WITHIN TIMESAT are different!

enter image description here

1 Answer 1

2

I just received a mail by one of the authors of the program. I was right in assuming that the problem lies in the output format I specified, since the .txt values are right.

After reproducing the results for a xy time in 32-bit format .... they are now showing good results. I have checked this in TIMESAT (TSM_imageview) and assume that I have left old header files or so in the folder I was working in. Hence the data displayed in ENVI or other programms was wrong.

My advice is to work in a clean environment and concentrate while importing of the files, which is specified in this pdf: http://web.nateko.lu.se/timesat/docs/Getting%20image%20data%20into%20TIMESAT.pdf

It has taken me some attempts to find out, that I also forgot to specify the offset explicitly with 0 instead of leaving it blank ...

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.