I have twelve monthly MODIS datasets for a 1 year period (http://neo.sci.gsfc.nasa.gov/view.php?datasetId=MOD13A2_M_NDVI&year=2015) and wish to combine all the data into a single map highlighting average values but also representing seasonally high NDVI values in the Southern and Northern hemisphere summers. I have so far used the cell statistics tool in ArcGIS to create an annual average from the data and secondly an annual maximum (maximum values in each cell from all 12 datasets). I am not sure whether the average NDVI well represents high summer NDVI values, whilst maximum NDVI seems to over-represent them. Does anyone have any advice on how to best represent this data? I require just a single map as wish to produce a weighted overlay with a number of other raster datasets.
Potentials that I'd suggest that you look at are:
NDVI percentiles - to indicate the highest & lowest NDVI values, without having the issues associated with anomalous min & max values.
Range of NDVI values in a year - to indicate variability over the year. Potentially based on the percentiles, instead of min & max values.
Bi-modality, to indicate multiple growing seasons within a year - try Sarle's bimodality coefficient, which is based on skewness and kurtosis.