My primary research interests aim to investigate how plants respond to climate through time in both natural and urban landscapes. To answer such questions, I primarily use imagery from remote sensing platforms such as satellites (e.g., Landsat) and planes (e.g, AVIRIS) at a range of spatial, spectral, and temporal resolutions. I also use in situ data from eddy covariance flux tower measurements, terrestrial biosphere model outputs, satellite-derived products (e.g., vegetation productivity), meteorological datasets, geospatial vector data, and more.
My research to date fits within a few interlocking themes:
Seasonal changes and phenology
Land cover change
Feel free to email me for any article PDFs linked below, and many of my recent publications are open access.
Multiple years of drought (2014 in red is severe) affecting green vegetation fractional cover in different urban tree species and turfgrass present in Santa Barbara, California (e.g., QUAG = Quercus agrifolia, coast live oak) (Miller et al. 2020).
Enhanced droughts can be considered as a leading edge of future impacts of climate change, with drought conditions becoming increasingly warm and dry.
Urban areas are often even hotter than the surrounding natural landscape, and so can serve as a testbed for extreme droughts that may occur in the future in natural landscapes. In my PhD, I studied how urban vegetation responded to recent severe drought in southern California using time series of remote sensing data from airborne imaging spectrometers (also known as hyperspectral imagery) (Miller et al., 2020; Miller et al., 2022 ISPRS; Miller et al., 2022 Urban Climate). This is to address concerns about long-term potential for sustainability of different trees and grass in urban areas and the ecosystem services, such as cooler temperatures, that they can provide during potential warmer and drier conditions.
In my current postdoc work, I am studying how vegetation productivity responds to drought across different seasons using eddy covariance measurements (Miller et al. in prep). This is to resolve potential outlying cases of vegetation response during specific drought conditions that have been observed in the literature. I am also involved with research related to rapid-onset flash droughts and their impacts on productivity (Osman et al., 2022).
High spatial resolution map (4 m) of urban vegetation types in the Santa Barbara, California urban region classified from AVIRIS-NG hyperspectral imagery using random forest (Miller et al., 2022 ISPRS).
We live on an increasingly urban planet. Most of humanity now lives in cities and this is only expected to grow in the future, and so the vast majority of individual interactions with the natural world will be formed within urban environments. Much of my research has focused on urban ecosystems, using many remote sensing techniques that are applicable in natural and agricultural areas as well. This also requires working at high spatial resolutions and within different constraints than many natural environments, but affords many opportunities to answer understudied questions needed in urban areas: What urban vegetation will be available and how to maintain ecosystem services in the future? How does urban vegetation vary spatially across cities, and how does it provide benefits (or nuisances) for urban residents?
I have addressed many of these questions in my PhD work, investigating drought in southern California (Miller et al., 2020; Miller et al., 2022 ISPRS; Miller et al., 2022 Urban Climate). I also estimated urban productivity in an upscaling study in Minneapolis-St. Paul, Minnesota in my master's thesis work (Miller et al., 2018), and as an undergraduate performed key analyses for how the urban heat island affected deciduous tree spring phenology (leaf-out) in Boston, Massachusetts (Melaas et al., 2016).
Monthly changes in NDVI greenness (x) and land surface temperature reduction compared to impervious surfaces (y) across different urban vegetation types in the Santa Barbara, California urbanized area from 10 years of Landsat satellite data (Miller et al., 2022 ISPRS).
Seasonal changes and phenology
One of the primary responses to climate change is a shifting in the seasons, with warming temperatures throughout the year. This can affect the timing of how ecosystems take up carbon and provide ecosystem services in urban regions, to name a couple of examples that motivate my research.
In my postdoc research, I have investigated this relates the timing of ecosystem photosynthesis and drought response across different seasons (Miller et al. in prep). In my PhD, I compared how urban vegetation expressed drought response in different seasons in a Mediterranean climate, and compared this timing to the duration of drought conditions from a drought index (SPEI) (Miller et al., 2022 ISPRS). I focused on Landsat satellite-derived urban tree phenology (start of spring) in Boston during undergraduate research (Melaas et al., 2016).
Estimated summertime ecosystem photosynthesis (GPP) upscaled from eddy covariance and sap flux measurements using high spatial resolution satellite imagery (WorldView-2) around a tall eddy covariance tower in a suburban neighborhood of Minneapolis-St. Paul, Minnesota (Miller et al., 2018).
Quantifying the ability of the land surface to take up carbon from the atmosphere (e.g., vegetation photosynthesis) and estimating how much carbon it can and will be able to hold (e.g., biomass) are key goals related to climate change.
In my postdoc research, I am studying relationships between gross primary productivity (GPP, the initial product of ecosystem photosynthesis) and droughts using a network of eddy covariance flux tower data (Miller et al. in prep). I also estimated upscaled urban GPP rates at high spatial resolution during my master's thesis (Miller et al., 2018). Urban GPP has been understudied because urban vegetation is difficult to spatially resolve and then parameterize in global satellite observations, but forms a growing flux of carbon across the land surface as urban areas expand. I have also worked with labeling biomass changes due to fires, insects, clearcuts, etc. using Landsat satellite data (Kennedy et al. 2018).
Land cover change
The land surface is changing rapidly both as a result of the impacts of climate change and directly as a result of human activities.
In my PhD research, I evaluated how tree and grass cover changed across Los Angeles during 2013-2018, a period marked by exceptionally severe droughts (Miller et al., 2022 Urban Climate). This study used a time series of rarely-available annual data from hyperspectral imagery, allowing for separation of trees, grass, non-photosynthetic vegetation (NPV, or dead vegetation), and non-vegetated urban surfaces at the sub-pixel scale.
I have also worked with land cover change attribution from Landsat imagery across the west coast of the US, using random forest models as part of the Landtrendr project (Kennedy et al. 2018).