The USGS has developed a new and innovative suite of landscape metrics (i.e. “Climate-Landscape Response” or CLaRe metrics) that are proven effective in mapping when and where invasive buffelgrass is green in Saguaro National Park (SNP) near Tucson, Arizona. Buffelgrass fills the interstices between widely-spaced desert plants with a continuous mat of fine fuels, carrying fire throughout the non-fire adapted Sonoran Desert landscape, with the potential to transform the iconic, species-rich, ecosystem into a monotypic grassland. CLaRe phenometrics, derived from gridded climate data and 250-m MODIS data, capture the strength of the landscape greenness response to climate and expose buffelgrass due to its rapid and strong response to recent precipitation. Buffelgrass remains dormant much of the year, but has short windows when it is photosynthetically active and vulnerable to herbicide. By mapping when and where buffelgrass is green, land managers can optimize their treatment activities, saving money. In addition, current results suggest that it is possible to detect nascent populations of buffelgrass comprising less than 5% of the landscape by monitoring the trends of these CLaRe phenometrics.
The ability to monitor and interpret trends in land surface phenology informs science-based land management. The CLaRe metrics capture the landscape response to climate and will complement the current temperate-zone phenometrics; no doubt they will prove broadly useful for many applications, including mapping invasive species (many invasives are observed to respond more rapidly to climate – especially precipitation - than native vegetation, including buffelgrass, Lehman’s lovegrass and cheatgrass), monitoring ecosystems (capturing trends between temperature-driven and precipitation driven ecosystems), habitat modeling, and vegetation classification.
Modeled buffelgrass presence-absence based on the top 20 % of CLaRe ppt123 metrics. CLaRe ppt123 is the correlation between MODIS greenness and the cumulative precipitation for the prior three 8-day periods. For each year shown, all CLaRe ppt123 values were extracted for each mapped vegetation type (based on the SWreGAP classification) and the top 5th of the highest correlations were mapped as buffelgrass presence (left panels). Higher elevation vegetation types that are unsuited for buffelgrass (e.g., Madrean pine-oak forest and woodland) were masked. The composite model (right image) maps buffelgrass presence if 2 of 3 sub-models mapped buffelgrass presence. Validation results show the composite model has an overall accuracy of 83 % and correctly maps 46 % of known buffelgrass patches.