 |
|
|
|
|
|
|
|
FAQ
|
- When do you release version 3.1?
The beta version was released on Feb. 17, 2012. If you have not registered please do so to get access to the new version. If you have registered but need a new password please contact Lars Eklundh.
- What platforms and operating systems are supported?
TIMESAT 3.1 was developed on Matlab R2010b under Windows XP (32-bit) and Linux CentOS (64-bit). Fortran programmes were compiled with Intel Fortran compiler version 10.0.27.
- When do you release version 3.0?
It was released on Feb. 9, 2010. If you have not registered please do so to get access to the new version. If you have previously registered you can go directly to the login page (use the old password) to download the new version. Contact Lars Eklundh for a password if you have forgotten the old one.
- What are the major changes between version 3.0 and version 2.3?
The core routines are essentially the same, but we have improved the functionality, added several new functions, and improved the user interface. It is now easier to automate tasks since user settings are stored in separe files. We have also added the possibility to process data under a land cover mask that will enable different settings for different land cover classes. For more details please see the text under "Software versions".
- How can I convert my images to a format that suits TIMESAT?
This is a common question, but we do not have the capacity to maintain information about format conversion from different image processing or GIS packages. Instead we have selected to work with the most generic type of formats, flat binary files, which is supported by most software packages. Please see the text under "File formats" for some more information.
- Some data are missing from my time series - can I still run TIMESAT?
TIMESAT requires an equal number of data files each year (or in the case of data in ASCII file equal number of data values). If an image is missing for a certain date it is necessary to add an image for this date containing a code for missing data (e.g. -999) for each pixel. When running TIMESAT the data range is then specified so that it excludes this missing value. When running from an ASCII file a missing data value can just be added to the list of data values at the correct position. TIMESAT uses least-squares fits that can cope with a certain number of missing data values in a time-series. However, the processing will skip to the next pixel if too many missing data values are encountered.
- I have two years of data but I only get one season in the output from TIMESAT?
TIMESAT will always generate one less season than the number of years of data. This is because one calendar year may not always contain one full growing season (e.g. if the season begins in December and ends in October). TIMESAT will start to process the first full season encountered, continuing until the end of the data set, and generate n-1 seasons. If a user wants to process one particular year it is necessary to surround this year with data on both sides. Thus, a minimum of three years of data is needed to make sure that a comprehensive picture of the phenology in the middle year is generated. If no data files exist for years before and after the time-period to be processed we recommend the user to duplicate the first and last years of data and add first and last in the time series. We have added a section in the new users manual to clarify this point.
- Why do I get large areas containing only missing data, when I process my TIMESAT output with seas2img?
This may occur if you have not specified a time interval wide enough to include the full growing season in seas2img. The interval sometimes needs to be very wide, perhaps spanning one and a half years. Try widening the interval and see if you obtain data for all your area. Another reason may be that the original data is too noisy for TIMESAT to identify any meaningful sesons in your areas. You can distinguish case this by setting separate missing value codes in seas2img.
- Does TIMESAT run under all operating systems and Matlab versions?
No, we have developed the latest version on Windows XP and tested it with Matlab 2007b and 2008b. Users have reported problems with Windows 2000, and with older Matlab versions. There also seem to be problems connected to different language versions. Unfortunately we do not have the capacity to handle these problems. However, we intend to modify the code so that it will run under Linux.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Web design: Dr Maggie
|
|
Site maintained by: Lars Eklundh.
This page updated: 17 February 2012
|
|
|
