Urban Remote Sensing
Continuous Subpixel Monitoring (CSM)
Subpixel land cover change using Landsat Time Series
CSM allows continuous (spatially and temporally) monitoring at the subpixel level
Generated a time series cube with high accuracy to monitor subpixel impervious surface dynamics
CSM can automatically detect different patterns of urban transitions overtime
Training samples and trained random forest models were transferable with CSM
• CSM works well in one of the cloudiest regions in the United States
Urbanization; Deforestation; Agricultral land loss...
An urban pixel changed from low density developed land to high density developed land.
Time series model from a forest pixel, characterizing the vegetation phenological information . Red line illustrates the slightly growing trend of this forest pixel over time.
CSM is currently tested in different cities:
Binghamton, NY
Columbus, OH
Omaha, NE
SASMA: Spatially Adaptive Spectral Mixture Analysis
SASMA: Characterizing representative endmembers for unmixing where signatures from neighboring similar pixels
An automatic spatial adaptive spectral mixture analysis (SASMA) method was proposed
Endmember candidates were automatically chosen using a classification tree approach
Spectral and spatial information were applied to develop the classification rules
Endmember spectra were derived as spatially weighted spectra of adjacent candidates
The SASMA method performs well in estimating impervious surface fractions
Biophysical Component Index (BCI)
A simple but effective index, called biophysical composition index (BCI), is developed to characterize urban environment.
The performances of BCI are performed with NDVI, NDBI, and NDISI at three different spatial resolutions.
BCI is among the best in representing impervious surfaces abundance at each scale.
BCI is also effective in quantifying vegetation abundance when compared to NDVI.
With BCI, urban impervious surfaces and bare soil can be moderately separated.
Casting Shadow Removal
using high resolution satellite images
Original false-color Landsat images
Impervious surface fraction where overestimation is casued by casting shadows
Improved fractional impervious surface mapping by removing casting shadows
Cities in deserts
Mapping Cities in Mojave Desert, Middle East, and Africa (upcoming)
Subpixel mapping of 3 desert cities. Frrom Left to right: Las Vegas, USA; Amman, Jordan; Baghdad, Iraq
Urban Tree Canopy Extraction
using LiDAR and multispectral aerial photos
C3/C4 Grass Types in the Great Plain
using adapted SASMA algorithm