Remote Sensing of Invasive Annual Grasses -- Greater Yellowstone Ecosystem
Exotic annual grasses such as cheatgrass (Bromus tectorum) have heavily invaded portions of the western United States, rapidly degrading habitats and increasing wildfire risk. Cheatgrass and other ESIs (desert alyssum [Alyssum desertorum], and annual wheatgrass [Eremopyrum triticeum]) are an emerging threat to the Greater Yellowstone Ecosystem (GYE); climatic changes including earlier snowmelt/run-off and warmer spring temperatures have made the region more suitable for invasive annuals, and disturbances from human recreation and wildlife movement/migration can act as vectors for spread and invasion. ESIs often out-compete native species, have recently expanded at a rapid rate, and pose a threat to wildlife and biodiversity in the internationally cherished GYE.
We are developing multi-scale methods to capture the phenology of cheatgrass with harmonized Landsat and Sentinel satellite imagery, high resolution Planet imagery, and experimenting with centimeter-scale uncrewed areal systems (UAS) multispectral imagery. In other studies, time-series analysis techniques show great utility in identifying infestations through the differencing of early-season peak greenness and mid-summer cheatgrass senescence. With field-based training and validation data we will train machine learning models on our seasonal variety of imagery sources to identify ESI infestations and compare the efficacy of different imagery sources on model performance. With input from National Park Service (NPS), US Forest Service, and Country weed managers we are co-producing early detection tools, high-resolution map products to quantify change at management-relevant scales, and giving assistance prioritizing management actions considering changing climatic conditions and the spread of invasive annuals.
With a better understanding of the spatial distribution of ESI infestations, we will proceed into our decision science tasks. This includes understanding ESI management options available and their effectiveness. Using the Resist-Accept-Direct (RAD) framework recently adopted by the NPS (Schuurman et al. 2020), we will work with our partners to coproduce a suite of possible spatially explicit actions to address ESI and understand the consequences of invasion. In the Gardiner Basin in northern Yellowstone NP for example, our partners have expressed a ‘hold the line’ strategy (resist) and explained that cheatgrass cannot be allowed to expand further upslope into the core areas of Yellowstone. However, given climate change, dispersal vectors, and continued recreational visitation and park use, are there certain actions or strategic locations that would lead to better outcomes if planned strategically (i.e., minimizing invasive spread in newly disturbed areas)? Our intent is not to prescribe actions to managers, but to develop effective strategies that fit within the real-world constraints land managers face.
Exotic annual grasses such as cheatgrass (Bromus tectorum) have heavily invaded portions of the western United States, rapidly degrading habitats and increasing wildfire risk. Cheatgrass and other ESIs (desert alyssum [Alyssum desertorum], and annual wheatgrass [Eremopyrum triticeum]) are an emerging threat to the Greater Yellowstone Ecosystem (GYE); climatic changes including earlier snowmelt/run-off and warmer spring temperatures have made the region more suitable for invasive annuals, and disturbances from human recreation and wildlife movement/migration can act as vectors for spread and invasion. ESIs often out-compete native species, have recently expanded at a rapid rate, and pose a threat to wildlife and biodiversity in the internationally cherished GYE.
We are developing multi-scale methods to capture the phenology of cheatgrass with harmonized Landsat and Sentinel satellite imagery, high resolution Planet imagery, and experimenting with centimeter-scale uncrewed areal systems (UAS) multispectral imagery. In other studies, time-series analysis techniques show great utility in identifying infestations through the differencing of early-season peak greenness and mid-summer cheatgrass senescence. With field-based training and validation data we will train machine learning models on our seasonal variety of imagery sources to identify ESI infestations and compare the efficacy of different imagery sources on model performance. With input from National Park Service (NPS), US Forest Service, and Country weed managers we are co-producing early detection tools, high-resolution map products to quantify change at management-relevant scales, and giving assistance prioritizing management actions considering changing climatic conditions and the spread of invasive annuals.
With a better understanding of the spatial distribution of ESI infestations, we will proceed into our decision science tasks. This includes understanding ESI management options available and their effectiveness. Using the Resist-Accept-Direct (RAD) framework recently adopted by the NPS (Schuurman et al. 2020), we will work with our partners to coproduce a suite of possible spatially explicit actions to address ESI and understand the consequences of invasion. In the Gardiner Basin in northern Yellowstone NP for example, our partners have expressed a ‘hold the line’ strategy (resist) and explained that cheatgrass cannot be allowed to expand further upslope into the core areas of Yellowstone. However, given climate change, dispersal vectors, and continued recreational visitation and park use, are there certain actions or strategic locations that would lead to better outcomes if planned strategically (i.e., minimizing invasive spread in newly disturbed areas)? Our intent is not to prescribe actions to managers, but to develop effective strategies that fit within the real-world constraints land managers face.