Mapping an invasive grass in the northwestern US with fused satellite time series and biophysical features

The introduction and spread of the invasive annual grass Ventenata dubia (ventenata) has incited concern from land managers in the Inland Northwestern United States. Maps describing ventenata's distribution would be a valuable management asset but have not been developed. Techniques using satellite time-series have been used to detect annual grasses' unique phenological qualities (timing of biological events); however, detection is complicated by the co-occurrence of phenologically similar species. This study aimed to examine the capability of land surface phenology derived from fused satellite imagery to map and gain insight into the ventenata invasion. We evaluated the influence of land surface phenology, climate, and topo-edaphic predictors on ventenata classification and examined differences in the distributions predicted from three random forest models: 1) hybrid (phenology, climate, topo-edaphic), 2) bioclimatic (climate, topo-edaphic), and 3) phenology (phenology). The hybrid model indicated that 7.7 % (5454 km 2 ) of the Blue Mountains Ecoregion may contain heavy ventenata invasion and that many populations were located in forest/non-forest transition zones and forest openings. The phenology model predicted ventenata populations in regions occupied by other annual grasses, suggesting that land surface phenology characteristics alone could not differentiate ventenata from other invasive annual grasses. The bioclimatic model identified suitable habitat but overpredicted invasion extent in heavily treed areas. These results suggested that incorporating phenology with climatic predictors effectively differentiates invasive annual grasses when phenological patterns are similar, but habitat requirements differ. • Land surface phenology and biophysical predictors resulted in improved maps of an invasive grass • 30-m resolution phenology enabled detection of smaller patches of an invasive annual grass • Notable populations of an invasive annual grass are identified in forest openings and ecotones