Proposing an improved surface dryness index to estimate soil moisture based on the temperature vegetation dryness index

• Main Author: Luo, Lei
• Language: en_US
• Published: Kansas State University 2016
• Subjects: the “triangle” method; TVDI; soil moisture isolines; the dryness slope index; evaporative fraction
• Online Access: http://hdl.handle.net/2097/34605

Master of Arts === Department of Geography === Douglas Goodin === In this thesis, I proposed a new surface dryness index based on the slope of soil moisture isolines in the Land Surface Temperature/Normalized Difference Vegetation Index (LST/NDVI) feature space. This index, referred to here as Dryness Slope Index (DSI), overcomes the problem of Temperature Vegetation Dryness Index (TVDI) having different basis when calculating TVDI values across different images. This problem is rooted in the definition of TVDI whose calculation depends on the position of the “dry edge” and “wet edge” of pixels’ values in the LST/NDVI space of a specific image. The “wet edge” has a fairly stable physical meaning, which represents soil at field capacity or above, and it remains stable across a time series of images. However, the position of “dry edge” represents the driest condition in the image, which does not necessarily mean that the soil is completely dry. Therefore, the value of TVDI calculated from different images is not based on an invariant dry edge value as its baseline, and it is therefore likely to lead to incorrect conclusion if used without extra examination. This problem manifests itself when comparing TVDI values from different images with meteorological data. Results from similar analyses done with DSI showed more reasonable match with the validation data, indicating DSI is a more robust surface dryness index than TVDI. Having verified DSI can be effectively used in estimating soil moisture, I applied DSI on Landsat5 TM to study the relationship between soil moisture and land cover, slope, aspect, and relative elevation. Results showed that land cover accounts the most for variations of estimated soil moisture. I also applied DSI on a long time-series (2000 to 2014) of MODIS data trying to explore the temporal evolution of soil moisture in the entire Flint Hills ecoregion. Results showed little correlation between time and estimated soil moisture, indicating that no noticeable changes in soil moisture has been found through all these years.