Detecting invasive sericea lespedeza (Lespedeza cuneata) in Mid-Missouri pastureland using hyperspectral imagery

Missouri ranks second in cow-calf numbers in the United States and its pastureland has experienced invasion of various plant species. As one of the invasive weeds, sericea lespedeza is becoming a serious threat to pasturelands in this state. The in-situ field survey in these vast pastures is time consuming and often impossible because of accessibility. Typical aerial survey is also difficult to detect sericea because the plant is of similar size and color as natural grass and, thus, cannot be effectively discriminated in broadband aerial color photos. This study used an airborne hyperspectral image to map sericea and its invasiveness in a public grass field in Mid-Missouri. The maximal 1st-order derivative in red-near infrared region (650-800nm) was derived to separate sericea from fescue, the dominant grass in pastures in Missouri. With a simple threshold approach, sericea of various sizes were identified in the study area. It was also found that the maximal 1st-order derivatives of sericea patches were log-linearly related to sericea volume, a quasi 3-dimensional biophysical variable as an approximate measure of sericea invasiveness. The squared correlation coefficient (r(2)) of the regression was 0.65 and the estimation error of sericea volume estimation was 11% based on ground measurements at 27 sample sites. With this empirical regression model, the quantitative distribution of sericea volume was mapped, which could serve as a first step in alerting landowners and the general public about the seriousness of sericea invasion in Missouri pasturelands.