MODIS-Landsat data fusion for estimating vegetation dynamics - a case study for two ranches in West Texas

Vegetation remote sensing has been largely focused on the utilization of the Vegetation Indices (VIs), especially when it is related to vegetation dynamics in low vegetation areas. Meanwhile, precipitation is often seen as an indicator and driver of vegetation dynamics and climate change. Consistent and continuous database is necessary and critical for surface and climate change modeling. Time series of Vegetation indices, such as the satellite data sets of Normalized Difference Vegetation Index (NDVI) and Leaf Area Index (LAI) represent the vegetation dynamics of land surfaces in both time and space. In this research, I proposed a data fusion approach, which leads to merge the MODIS data with Landsat TM data to create a dataset of vegetation dynamics with both high spatial resolution and a fine temporal resolution. To verify the database, one random sample of red band images was tested and compared with MODIS data with a high correlation coefficient 0.907 and RMSE 0.0245. Additionally, I compared the Savitzky – Golay (SG) filter with the fixed interval-smoothing filter to get the smooth time series NDVI and LAI result, and found the SG is more sensitive of the vegetation growth response. In this thesis, both the temporal and spatial changes of vegetation dynamics in Faith ranch and Comanche ranch of Dimmit County, Texas, as well as the relationship between NDVI and precipitation, were discussed. LAI and NDVI time series database of each ranch was created separately, by using MODIS 15 (LAI) and MODIS 13 (NDVI) data from the year of 2000 to 2011. The relationship between NDVI (smoothed by SG filter) and precipitation were discussed based on the precipitation database from the year from 2009 to 2011. Keywords: Data fusion, Vegetation Dynamics, Landsat TM, Moderate Resolution Imaging Spectroradiometer (MODIS), Normalized Difference Vegetation Index (NDVI)), Surface reflectance, data smoothin