Remote sensing and believe in that direction, I just do and then want to do the timing of land classification, preliminary intend to do is to use MOD13Q1 NDVI value, then using unsupervised classification, but then encountered a problem when processing MOD13Q1 image, is drying out some poor quality like cloud images like yuan yuan and ice and snow, can empty out a lot of null values, I checked the documents, others is to use linear interpolation method to the entire value of the image gap to plug in values, but this step is how to achieve? I check the documents and baidu for a long time, not solve, so this turn, again!!!!!! Thank you...
Specific literature content is:
The NDVI products utilized here were initially preprocessed bystacking 23 annual images into a time - series. Or, as these raw
NDVI time - series are more disturbed by background noises due to cloud contamination and atmospheric variability, we removed those flagged as either clouds, water, or ice -based on MODIS quality assurance information and then applied a linear interpolation method to complete the missing data. We then employed a Savitzky - Golay filter (Savitzky & amp; Golay, 1964) to reconstruct a smoothed NDVI time series and utilized the TIMESAT tool (J ¨ onsson & amp; Eklundh, 2004) to compute a set of 11 annual vegetation phenology features.
Don't know whether I understand the wrong... If there is a cognitive deviation, please correct me,