Measuring the dead component of mixed grassland with Landsat imagery

Detecting non-photosynthetic materials challenges multispectral remote sensing of vegetation. Dead component of grasslands, including litter and standing dead material as the accumulated phytomass of grassland productivity from previous years, is a primary connection in nutrient cycles in grasslands, and also conserves soil moisture and redistributes the grassland surface temperature. However, unlike green vegetation, it is hard to estimate the dead component with remote sensing approaches because the spectral signal of dead materials is similar to that of bare soil or soil crust (moss and lichen), with the only difference in the shortwave infrared region near 2000nm. In the literature, the Cellulose Absorption Index (CAI), an index based on hyperspectral imagery, is the most popular index for assessing dead cover or dead biomass. However, hyperspectral imagery is still not available for most grasslands worldwide. Therefore, a method to assess dead component using multispectral band imagery (e.g. Landsat imagery in this study) is needed. Grasslands National Park (GNP) in the southern part of Saskatchewan, Canada, has a large amount of accumulated dead vegetation because of long term conservation actions, which makes it a good study area for this project. This study aims to explore the relation between NDVI and dead cover, to investigate how different amounts of dead material change the relation of total biomass and NDVI, and also to test the potential to estimate dead cover using multispectral images. The results show that NDVI and dead cover have positive relationship when dead cover is less than 20%, no correlation when dead cover is between 20 and 80%, and significant negative relation when dead cover is more than 80%; further, the relation of total biomass and NDVI also changes with the same thresholds. The results also indicate that the dead component can be estimated with multispectral images using Normalized Burn Ratio (NBR) or Normalized Difference water index (NDWI), but the relationships are highly influenced by bare soil and soil crust.