Several studies have analyzed the impact of Urbanization on LST using different Remote Sensing products. Often, the objective is to find how urban land cover /urbanization (through replacement of natural land cover often with asphaltic, concrete, building/roofs, and generally non-evaporative surfaces) which results in elevated temperatures in urban areas relative the rural/non-urban environment, commonly referred to as Urban Heat Island (UHI). Common methodology adopted in such studies is to derive different land cover indices such as Normalized Difference Vegetation Index (NDVI), Normalized Difference Bareness Index (NDBaI), Normalized Difference Build-up Index (NDBI) and Impervious Surface Area (ISA), etc. and regress such indices on LST.
However, soil moisture (not an urban land cover index) appears to influence LST more than anything else. That is why, as Dr. Tenenbuam mentions, ..for a given NDVI, (or indeed any land cover index) there is usually a range of *LST* values (as a function of moisture condition varying at the same NDVI, NDBI, ISA, etc.). Thus, is the LST-NDVI, NDBI, ISA, NDBal etc relationship well modeled as a linear function?
The answer to the rhetorical question is definitely not…. because LST varies even at fixed NDVI, NDBI, NDBal, ISA and other land cover indices- showing a strong relationship to soil moisture condition. The logic goes like this, as Dr. Tenenbaum goes on to explain that …If NDVI, NDBal, NDBI, ISA, etc, is the same at two pixels, and one is hotter and the other is cooler, the hotter one is likely more dry (because available energy is being partitioned to have a higher amount of sensible heat, which accounts for the higher temperature), whereas the cooler one is likely less dry (because available energy is being partitioned to have a higher amount of latent heat, which accounts for the lower temperature)… I agree with him.
My question is this: How accurate is it therefore to model/explain LST in terms of NDVI, NDBI, ISA, NDBal and related land cover indices which are used as a proxy for urban environment?
What urban land cover indices will best serve as proxy to explain LST, especially when the objective is to explain LST dynamics as a result urbanization processes.
Thanks for your anticipated contributions
I think your prompt is the general form of the question as to what causes the urban heat island and how can these factors be measured, especially using remote sensing.
The forcing factors that have to do with changes in evapotranspiration (e.g. due to vegetation clearance or soil sealing) will probably be more difficult to directly detect remotely, as they have to do with file spatial- and temporal-scale changes in water flux and surface humidity. But there are dozens of studies that use vegetation indices (take your favorite -- NDVI, EVI, etc) to show that cities are hotter where they are less vegetated, or that cities are hotter where they are more paved, and that this effect likely has at least something to do with evaporative cooling. Other land cover metrics -- I've been looking at albedo -- are also going to have some forcing effect on producing UHI conditions. I think measures of urban "canopy" or building texture might also supply valuable insight into how convection processes may be altered over cities as well. But no single index of "urban-ness" seems likely to capture all the biophysical forcing factors at play in producing an urban heat island.
Even the means of accurately measuring urban heat island temperatures is made more difficult by land cover heterogeneity across the urban landscape. A colleague of mine at BU has very recently published work that shows that LST and near-surface air temperature do not track consistently across gradients in impervious cover. So even our ability to remotely detect the near-surface urban heat island may be more prone to error than we would like.
LST IS INVERSLY PROPORTIONAL TO NDVI
LST IS DIRECTLY PROPORTIONAL TO URBAN SETTLEMENT
http://www.sciencedirect.com/science/article/pii/S2212095514001126
http://eprints.qut.edu.au/64565/1/remotesensing-05-05969.pdf
I think your prompt is the general form of the question as to what causes the urban heat island and how can these factors be measured, especially using remote sensing.
The forcing factors that have to do with changes in evapotranspiration (e.g. due to vegetation clearance or soil sealing) will probably be more difficult to directly detect remotely, as they have to do with file spatial- and temporal-scale changes in water flux and surface humidity. But there are dozens of studies that use vegetation indices (take your favorite -- NDVI, EVI, etc) to show that cities are hotter where they are less vegetated, or that cities are hotter where they are more paved, and that this effect likely has at least something to do with evaporative cooling. Other land cover metrics -- I've been looking at albedo -- are also going to have some forcing effect on producing UHI conditions. I think measures of urban "canopy" or building texture might also supply valuable insight into how convection processes may be altered over cities as well. But no single index of "urban-ness" seems likely to capture all the biophysical forcing factors at play in producing an urban heat island.
Even the means of accurately measuring urban heat island temperatures is made more difficult by land cover heterogeneity across the urban landscape. A colleague of mine at BU has very recently published work that shows that LST and near-surface air temperature do not track consistently across gradients in impervious cover. So even our ability to remotely detect the near-surface urban heat island may be more prone to error than we would like.
What you need is access to high resolution TIR data that is corrected for vegetation and emissivity that covers entire cities - see www.myheat.ca
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing