EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

A new Remote Sensing-based vegetation water stress index: -Temperature Vegetation Water Stress Index (TVWSI)

Rakesh Chandra Joshi1, Dongryeol Ryu2, Gary J. Sheridan1, and Patrick N.J. Lane1
Rakesh Chandra Joshi et al.
  • 1Forest Hydrology Research Group, School of Ecosystem and Forest Sciences, University of Melbourne, Australia (
  • 2Environmental Sensing and Modelling Lab., Department of Infrastructure Engineering, University of Melbourne, Australia

Remote sensing techniques are widely used to evaluate the biophysical status of vegetation, including water stress caused by soil water deficit. Based on the nominal links between water stress condition, transpiration and canopy temperature in the vegetation, numerous studies have used a trapezoidal relationship between Land Surface Temperature (LST) and Normalized Difference Vegetation Index (NDVI) over vegetated surfaces to develop the water stress metric, in which the level of stress could be identified by the spatial location of the pixels on the spectral space (Goetz and Goetz 1997; Lambin, Lambin, and Ehrlich 1996; Nemani et al. 1993; Nemani and Running 1989; Price 1990; Sandholt, Rasmussen, and Andersen 2002). However, the amount of change in canopy temperature could also vary spatially by the canopy water status at that time. Thus, LST-NDVI alone cannot construct an efficient metric to see the spatial patterns of water stress at ecosystem level unless they are coupled with water status of vegetation at that moment. This study hypothesizes that a metric which can combine LST-NDVI information with an indicator for canopy water status could give more accurate estimations of the real-time vegetation water stress. The remotely sensed plant canopy water status indicator (a metric based on canopy reflection in the Short-Wave Infrared region (SWIR)) could add the canopy water status information to the LST-NDVI based indices, which may better explain spatial/temporal water stress condition in the plants especially in densely forested areas where signal saturation is a major issue. In this study, the third-dimensional information of SWIR has been combined with LST-NDVI spectral space to create a new remotely sensed vegetation water stress index, TVWSI (Temperature Vegetation Water Stress Index) which seems to be more realistic to capture stress dynamics at large scale. 

Sixty grids (2 km X 2 km) each containing 16 pixels of daily MODIS-reflectance (band 1 – band 7, 500 m spatial resolution) and 4 pixels of daily MODIS-LST (1 km spatial resolution) were chosen over forested areas in Victoria representing most of the bioregions as classified by the Interim Biogeographic Regionalisation for Australia (IBRA7). From 2002 to 2018 daily TVWSI values of each grid were evaluated against the modelled daily available soil moisture content in the top 1 m of the soil profile, and rainfall data, from the Australian Bureau of Meteorology (BOM). TVWSI performed better than other dryness indices mentioned in the literature. A high correlation was obtained between TVWSI vs. soil moisture and TVWSI vs. rainfall with a coefficient of determination value of 0.6 (p<0.001) and 0.61 (p<0.001) respectively when data were combined spatially and temporally. Even improved correlations ranging (0.4-0.7, p<0.001) were obtained for individual grids over the mentioned period. While correlation ranging (0.15-0.48, p<0.001) were obtained using dryness indices like Perpendicular Drought Index (PDI), Modified PDI (MPDI), Temperature Vegetation Dryness Index (TVDI) and Vegetation Supply Water Index (VSWI). The result shows that the TVWSI can capture real-time ecosystem water stress well and the metric could be an efficient input parameter for many hydrological, drought and fire prediction models.


How to cite: Joshi, R. C., Ryu, D., Sheridan, G. J., and Lane, P. N. J.: A new Remote Sensing-based vegetation water stress index: -Temperature Vegetation Water Stress Index (TVWSI), EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-12308,, 2020