Analysis of MODIS 250 m NDVI Using Different Time-Series Data for Crop Type Separability

Abstract

The primary objectives of this research were to: (1) investigate the use of different compositing periods of NDVI values of time-series MODIS 250 m data for distinguishing major crop types on the central Great Plains of the U.S. and (2) analyze collection 5 versus collection 4 time-series MODIS 250 m NDVI data to separate crop types. NDVI profiles extracted from different compositing periods for 2001 and 2005 were analyzed to see whether 8-day (and dual-8-day) composited NDVI data, as compared to 16-day composited NDVI data, would show finer scale spectral-temporal variability that would result in improved crop separability. NDVI value profiles were also extracted from different collection versions (4 and 5) for 2001 and 2005 (collection 5 only). Phenological curves for all crops and all datasets were created and visually inspected and JM distance statistical analysis was performed to compare separability of the crops for both the compositing period analysis and the collection version analysis. Major conclusions and findings for the compositing period analysis include: (1) there are statistical differences among the different compositing period datasets, (2) time-series data that have shorter compositing periods are more effective in separating crop types, and (3) any observed differences should be interpreted with care and in the context of variations in environmental conditions for a given growing season. For the collection version analysis the major finding was that, contrary to expectations, the most recent version of time-series MODIS 250 m data (version 5) was inferior to version 4 in terms of crop separability; however the analysis did not suggest reasons for the outcome. As a result of this research, it is tentatively recommended (subject to further research) that MODIS NDVI data (a) from a shorter 8-day compositing period and (b) from collection 4 should be used where possible for crop-type mapping in the study region.