Bad Bromes in the Badlands: Monitoring Invasives in the Great Plains

Category: Identifying Invasive Species Extent & Critical Species Habitat
Project Team: Northern Great Plains Ecological Forecasting
Team Location: NASA Goddard Space Flight Center – Greenbelt, Maryland

MODIS ForWarn (top), Landsat 8 (middle), and Sentinel-2 (bottom) data were used with in situ data to identify invasive annual brome. Image Credit: Northern Great Plains Ecological Forecasting Team

MODIS ForWarn (top), Landsat 8 (middle), and Sentinel-2 (bottom) data were used with in situ data to identify invasive annual brome. Image Credit: Northern Great Plains Ecological Forecasting Team

Authors:
Amanda Clayton
Jessica Fayne
Carl Green
Jared Tomlin

Mentors/Advisors:
Dr. Kenton Ross (NASA Langley Research Center)

Past/Other Contributors:
Sean McCartney (Center Lead)

Abstract:

Cheatgrass (Bromus tectorum) and Japanese brome (Bromus japonicus) are widespread invasive annual brome grasses that are distributed throughout the western United States. In the Northern Great Plains (NGP), the presence of these invasive brome species has led to a decrease in native plant diversity, reduced soil water content, and altered fire regimes—leading to more frequent fires of higher intensity. The National Park Service Northern Great Plains Network (NGPN) Inventory & Monitoring Program conducts vegetation monitoring throughout national park units to develop management practices for habitat restoration and invasive species control. Incorporating remotely sensed NASA Earth observations data enables the NGPN to efficiently monitor the regional extent of brome abundance over time. The spatial and temporal resolution of Aqua/Terra Moderate Resolution Imaging Spectroradiometer (MODIS), Landsat 8 Operational Land Imager (OLI), Landsat 5 Thematic Mapper (TM), and Sentinel-2 Multispectral Instrument (MSI) were leveraged to capture the vegetation phenology of these invasive brome species across the NGP. The satellite-derived Normalized Difference Vegetation Index (NDVI) was used to determine the distinct and early green up and senescence patterns of brome. Using unsupervised classification methods, the invasive bromes were identified from the surrounding native grassland species and associations were made between brome abundance and NDVI phenology metrics. Understanding the behavior of these invasive species through space and time will aid managers in developing a successful annual brome management strategy for NGP park units and identify areas for targeted management efforts.

Main Page | Next story

14 Comments

Daryl Ann Winstead (Mekong River Basin Agriculture) 18-08-2016, 12:30

Great video! Did the team encounter any errors when working with the Sentinel 2 imagery? Thanks in advance!

Reply
Northern Great Plains Ecological Forecasting 19-08-2016, 01:24

Thanks for the question Daryl Ann,

Could you please elaborate on what you mean by ‘errors working with the Sentinel 2 imagery’? There are a multitude of sensor errors that we can expect when we work with satellite imagery, but the key is being able to understand the metadata to appropriately identify these sensor, scanning, or reflectance errors in order to not skew the analysis. As far as the needs of our team were met by the Sentinel 2 imagery, there were no outstanding errors that prevented us from being able to use the data. It would be good to know if there are things to look out for in other applications, were you troubled by Sentinel 2 imagery errors in your research? If could you talk more about that it would be great to learn and prepare in case we run into those issues.

All the best with your research!

NPG-EF

Reply
Northern Great Plains Ecological Forecasting 19-08-2016, 01:25

Sorry! I signed the acronym out of order,
NGP-EF**

Reply
Daryl Ann Winstead (Mekong River Basin Agriculture) 19-08-2016, 12:16

Hi there!

I probably should have stated “limitations” rather than “errors”. I have previously worked with the Sentinel-2 data when it was much newer and experienced that the file size and file nomenclature were both difficult to work with. So I was just curious if the project experienced anything of that sort when working with the data. Thanks again for your response!

Reply
Northern Great Plains Ecological Forecasting 22-08-2016, 13:15

Hi Daryl Ann –

We found that you could easily visualize and download data using Amazon Web Service – easier than downloading direct from the ESA portal. ESA also has their own Sentinel-2 Toolbox complete with it’s own processing software SNAP (Sentinel Application Platform) that we found helpful!

Brian Woodward 16-08-2016, 02:20

Great job, team! I’d love to hear about your experience using Sentinel 2 imagery– how did it do in comparison with Landsat in your model? I look forward to integrating S2 into our projects in the future.

All my best,
Brian Woodward

Reply
Northern Great Plains Ecological Forecasting 16-08-2016, 12:58

Hi Brian!

Thanks for asking! The Sentinel-2 Imagery was great to work with! Because of the large footprint of the tile, small pixel size, and the increased spectral resolution for land surveying compared to Landsat, the Sentinel-2 Imagery was able to tell us a lot about the landscape that we would not be able to see using Landsat or MODIS. However, because the area is very sensitive to phenology, it was more important to focus on datasets with a longer data record, which meant that we had to sacrifice the great qualities of Sentinel for a sensor with a longer data record, Landsat. The Sentinel-2 satellite provided us data for this year (2016) to compare to our Landsat classifications as well as NDVI differencing techniques, which was extremely useful.

I also look forward to working with Sentinel-2 data in the future, this is a great sensor! But I think we’ll really start seeing the benefits of it in a few years when we have more historical data.

All the best with your research!

NGP-EF

Reply
Northern Great Plains Ecological Forecasting 15-08-2016, 15:09

Hi Alison!

Thanks for your comment! That’s a terrific question. Because of the phenology of the brome species, brome generally only increases with time. This is because brome’s ‘green-up’ is earlier than most native species. This allows brome to use up the valuable nutrients from the ground before the native species ever has a chance. This in turn inhibits the growth of the native species, shifting the dominant vegetation type. It is therefore very important to identify regions where invasive plant species are prevalent to reduce invasive plant populations, giving native species an opportunity to flourish.

Thanks again!
Jessica

Reply
Amy Symstad 08-09-2016, 15:34

Hi Alison,

Actually, I don’t think we have enough data (yet) to state whether or not annual bromes, once in a site, will consistently result in a vegetation type shift in the northern Great Plains. In fact, our in-situ (on the ground) data show that there is a lot of year-to-year variation in the relative abundance of annual bromes at a given location. Some of this variation is definitely tied to weather (generally less brome in drought years), but weather is only part of the complicated story.

Amy

Reply
Emily Gotschalk 15-08-2016, 14:27

What platforms did you use to analyze your data, and how useful was Sentinel-2 data given that there was a limited number of scenes available?

Reply
Northern Great Plains Ecological Forecasting 15-08-2016, 16:12

Thanks for the question Emily!

We used Landsat 5&8 and MODIS primarily. Sentinel-2 is a great sensor because it is very high resolution at 10 meters, and has wonderful red wavelengths such as ‘red-edge’ which is great for vegetation health and classification. However, as you mentioned, there were a limited number of scenes available to use in our analysis. We were able to use the Sentinel-2 data to help identify the spectral response signal from the bromes to ensure we were looking in the right locations when we did our initial dive into Landsat classifications. The Landsat 5 & 8 sensors, along with ground survey data from the National Park Service, allowed us to classify percent brome abundance from 2008-2016.

Thanks again for the great question!

Reply
Emma Baghel 12-08-2016, 13:52

Wow, pretty unique VPS style. Loved the interviews.

Reply
Northern Great Plains Ecological Forecasting 15-08-2016, 13:44

Hi Emma!

Thanks for your comment! We are very glad that the National Park Service were so cooperative and willing to share important data and resources with us, as well as devoting their time to working on these interviews with us!

Reply
Alison Thieme 11-08-2016, 13:18

Great video! Once a brome invades a new area, is it consistently high abundance in the following years? What affects their distribution?

Reply

Leave a Reply