Discriminating invasive crested wheatgrass (Agropyron cristatum) in northern mixed grass prairie using remote sensing technology
| dc.contributor.author | Zhou, W. | |
| dc.contributor.author | Guo, X. | |
| dc.date.accessioned | 2018-08-04T17:42:08Z | |
| dc.date.available | 2018-08-04T17:42:08Z | |
| dc.date.issued | 2007-03-01 | |
| dc.description.abstract | Invasive crested wheatgrass in the Grasslands National Park cause biodiversity decrease and irreparable damage to prairie ecosystems. Controlling and managing invasive species require new methods to map and monitor their presence and spread. Traditional mapping techniques based on field observation and data collection are considered time-consuming, subjective, and always very limited in spatial extent and economically for relatively large areas. Remote sensing techniques provide a potential solution to this problem. However, previous work has been limited because of low spatial and spectral resolution of some data sources. The principal challenges in using remote sensors to detect invasive species lie in the spectral similarity across species and invasive species often mixing with the native species. This paper discusses how SPOT-5 imagery with 10-m resolution can be used to detect invasive crested wheatgrass in the mixed prairie. Several vegetation indices, including Normalized Difference Vegetation Index (NDVI), Normalized Difference Moisture Index (NDMI), Soil Adjusted Vegetation Index (SAVI), Modified Soil Adjusted Vegetation Index (MSAVI), Simple Ratio (SR), and Triangulated Vegetation Index (TVI), were initially selected and their spectral separability in separating crested wheatgrass and natives was examined. A new vegetation index, ExpNDMI, was derived from NDMI by incorporating an adjustment factor (L) to enlarge the difference among classes; and, further, performing a exponential transformation upon the modified index to suppress the variations in all classes. An artificial Neural Netwok (ANN) classifier based on back propagation (BP) algorithm was employed to classify crested wheatgrass and native grasslands in this study. The results indicated that ExpNDMI could significantly increase the spectral separability between crested wheatgrass and native grasslands and improve the classification accuracy. The highest overall accuracy of 79% was obtained. Band/VI combination with ExpNDMI improved the classification accuracy by more than 4% than the combination without ExpNDMI. The result of this study suggests that single-date SPOT 5 image with 10 m resolution could be useful in discriminating crested wheatgrass from natives in the mixed grasslands, and thus may reduce the dependence on the multitemporal data. | en_US |
| dc.description.version | Non-Peer Reviewed | |
| dc.identifier.uri | http://hdl.handle.net/10388/9369 | |
| dc.language.iso | en | en_US |
| dc.relation.ispartof | Soils and Crops Workshop | |
| dc.rights | Attribution-NonCommercial-NoDerivs 2.5 Canada | * |
| dc.rights.uri | http://creativecommons.org/licenses/by-nc-nd/2.5/ca/ | * |
| dc.subject | spectral separability | en_US |
| dc.subject | ExpNDMI | en_US |
| dc.title | Discriminating invasive crested wheatgrass (Agropyron cristatum) in northern mixed grass prairie using remote sensing technology | en_US |
| dc.type | Presentation | en_US |