|Title||A technique for monitoring ecological disturbance in tallgrass prairie using seasonal NDVI trajectories and a discriminant function mixture model|
|Publication Type||Journal Article|
|Year of Publication||1997|
|Authors||Goodin, DG, Henebry, GM|
|Journal||Remote Sensing of the Environment|
Natural and anthropogenic disturbance in tallgrass prairie communities can induce changes in plant species composition, including shifts in the relative abundance of C3 and C4 lifeforms. The asynchronous seasonality in greenness exhibited by C3 and C4 species enables monitoring of their relative abundance using temporal trajectories of sensor-derived vegetation indices, such as NDVI. We use close-range measurements made over 22 experimental plots at the Konza Prairie Research Natural Area (KPRNA) to evaluate seasonal trajectories in NDVI as a function of Full-size image (<1 K) ratio. The NDVI data were collected from each plot at approximately 10-day intervals throughout the 1995 growing season. Metrics for summarizing the temporal behavior of NDVI trajectories were derived from two transformations of the data: 1) a conventional approach plotting NDVI against day of year and 2) an alternative approach plotting normalized cumulative integrated NDVI against growing degree day. Discriminant function mixture models derived from each set of inetrics were used together with species composition data from the experimental plots to derive relative Full-size image (<1 K) abundances. Results show that both methods can classify the majority of cases into their correct Full-size image (<1 K) abundance category; however, the transformation using normalized integrated NDVI against growing degree day was a significantly better discriminator (p=0.0102). The techniques presented are effective for monitoring relative Full-size image (<1 K) abundance in tallgrass prairie; however, more investigation is needed to assess their performance at different spatial resolutions and in different geographic settings.