The Science and Technology Society (STS) of Sarasota-Manatee Counties, Florida is working with the NASA STELLA (Science and Technology Education for Land/Life Assessment) outreach program as a part of our STEM initiative. According to their site,
- "NASA STELLA instruments are portable low-cost do-it-yourself (DIY) instruments that support science education, and outreach through scientific engagement, inquiry, and discovery while helping you understand Landsat better".
STELLA instruments are developed under the influence and inspiration of Landsat. This alignment not only fulfills our project needs but also serves as a compelling addition to our STEAM initiatives:
-
To train the minds young Floridians to be more aware of our wetlands, to care for them and about them. Our program will bring more community publicity to the issue of wetlands change, as well.
-
To expose our middle- and high- school aged students to real science, using real data. That means how to use instrumentation and understand how the data is collected, and how the data can be used in the real world. It means not only to create beautiful charts and images that form the good results, but also to understand that data must be collected in a proper and reproducible way, that there physics reasons for lack of accuracy and lack of precision that one must understand and minimize in order to achieve meaningful results.
The NASA STELLA-Q2 is capable of making 18 different spectral measurements from the violet/blue portions of the electromagnetic spectrum out to near infrared regions (beyond our range of vision).The following figure (1) shows the visible spectrum by wavelength, and the yellow box indicates the STELLA-Q2 frequency range.
More can be found on the STELLA DIY instruments at the following link.
The following is a sample-by-sample animation of the type of data acquired from STELLA-Q2 Spectrometer built by STS. STS is providing the python code in a Jupyter Notebook that can be used as an example of how to display the data from the STELLA-Q2 device. We have also provided some sample data to be used with this notebook. It should be noted that we did change the name of some of the headers created from our instrument to add colors to each of the wavelength reading that are made in order to display each wavelength as a corresponding color. The near infrared wavelength readings are colored in grays, wheat and gold where the normal visible spectrum colors are in vivid colors that they represent.
The NASA site offers detailed instructions on how to build the STELLA-Q2 instrument at the following link:
https://landsat.gsfc.nasa.gov/stella/stella-q2/stella-q2-build-instructions/
The following image is the provided STELLA parts list that we used to order our components. If you download the parts list file from this repository, then the links will direct you to each component from one of the two vendors.
This is the STS STELLA-Q2 instrument that STS has built using the details supplied by the NASA:
There is excellent documentation on programming this device that can be found at the following link:
https://landsat.gsfc.nasa.gov/stella/stella-q2/stella-q2-programming-instructions/
We are just starting on the applications from this spectrometer, but they appear to be immense. The following figure provides some information on just one application we are delving into (1)
Bianca Clento of Rochester Institute of Technology has an excellent poster on the Quantifying Plant Biodiversity Using Different Spectrometers, Spectral Unmixing, and UAV Imagery that serves as an example of the type of application that we would like to apply for SW Florida Gulf Coast region related to the vegetative health along our coastline.
Our long-term plan is to deploy the STELLA-Q2 on a drone capturing spectral data along our Florida Gulf Coast. In a recent meeting with Sherri Swanson, Ecological and Marine Resources Division Manager for Manatee County Natural Resources, Sherri has been extremely helpful providing us shapefiles for the Sarasota Bay Estuary Program (SBEP) Watershed boundaries and a very interesting paper on Identifying and Diagnosing Locations of Ongoing and Future Saltwater Wetland Loss (2) available in this repository too.
and is image is from the SBEP Watershed Boundaries Shapefile:
In this report they state:
"Landsat green and near-infrared bands can be used to identify mangroves of varying conditions. Landsat imagery of two different periods can be used to measure mangrove forest improvement or decline.
This use also provides a sensitive expression of mangrove health in areas that are difficult to access on site and which are only revealed via aerial photography at very large scale (for example under 1000 scale).
A map of mangrove status and change was used to provide a detailed review the study area and identify 90 restoration opportunities, 121 sites of natural decline, 13 sites where there is no remedy for the decline and 3 sites where restoration is in progress but the mangroves have to yet rebounded to their earlier vigor."
STS plans on providing the ground truth data to be used in the calibration of these Landsat images.
As stated in their paper:
"Mangrove Conditions can be assessed from the work performed by Pastor-Guzman et al (2015) compared 20 hyperspectral and broad band vegetation indices to relative mangrove canopy chlorophyll measured at 12 sites along the northwest coast of the Yucatan Peninsula, Mexico. The sites were 30m by 30m to represent Landsat spatial resolution. The purpose of the work was to develop indicators of mangrove condition using remotely sensed data. Of the indices, normally distributed vegetation index green (NDVIgreen) was the most sensitive to canopy chlorophyll at the site level (r2 = 0.805.) The formula for NDVIgreen uses the near infrared and green bands. We found the NVDIgreen index to be an excellent indicator of mangrove condition in the Charlotte Harbor area.
The formula for NDVIgreen using Landsat 8 bands is:
NDVIgreen = (NIR − Green)/(NIR + Green)
Where:
- NIR (Near Infrared) corresponds to Landsat 8 band 5 Green corresponds to Landsat 8 band 3.
- The result is a value between 0 and 1.
The formula for NDVIgreen using Landsat 5 and 7 bands is:
NDVIgreen = (NIR − Green)/(NIR + Green)
Where:
- NIR (Near Infrared) corresponds to Landsat 8 band 4 Green corresponds to Landsat 8 band 2.
- The result is a value between -1 and 1.
Pastor-Guzman et al (2015) further explain that the linear model to construct a mangrove canopy chlorophyll map is:
y =−54.545 + 149.396x
x = pixel value of the Landsat 8 NDVIgreen calculation.
Applying the equation to the Landsat 8 NDVIgreen values yielded a generally narrower range of canopy chlorophyll values for Charlotte Harbor compared to the Yucatan. Because the equation could not be applied to earlier Landsat missions and more work needed to be done to confirm the relationships between Charlotte Harbor mangrove canopy chlorophyll and NDVIgreen, the Principal Investigator settled on simply using NDVIgreen the indicator of mangrove condition.
The B7-NIR band for Landsat 8 is between 0.85 and 0.88 micrometers wavelength, compared to 0.76 and 0.9 for Landsat 4 and 5. The Landsat 4 and 5 missions together provide a period of record from July 16, 1982 through June 5, 2012. Between the differences in wavelength and data formats, direct comparisons are difficult. The next section describes the method used to compare NDVIgreen between Landsat missions in order to detect change in mangrove condition from before the Landsat 8 launch in 2013. The image below is from this report showing the mangrove conditions assessed from the NDVIgreen calculations."
Now that we have our STELLA-Q2 device in working order, we have started testing from know color panels to actual vegetative species and states of health in different shades of light (Full Sun, Shade...). We have created 2 Jupyter Notebooks using the data from our STELLA-Q2 device for 4 different examples:
STELLA_brief_ver2_backyard_grass_shoreline.ipynb
with all of the required data files being provided. In this notebook we perform the following steps:
- We first load our STELLA-Q2 raw data as well as the White-Card readings that we made for calibration.
- We then proceed to plot the raw STELLA data and then the White-Corrected data to evaluate the data.
- We calculate NDVI and NIRv as discussed below. They are used in the prediction of all of our test patterns based on their spectral attributes.
- We calculate a Decision Tree to evaluate the data and observe the code that is required to predict our test patterns in a determistic method.
- We then define our spectral End Members for each of our test patterns.
- We use Knn to predict each test pattern from our data. Knn is a straight forward technique that is very transparent and not a black box.
We also have included a Jupyter Notebook called:
convert_clean4_clean.ipynb
This program will convert our typical STELLA files into easy to use Excel files. The program first reads in the raw STELLA data and your white-card data, and then makes the white-card corrections for each STELLA wavelength as well as calculating NDVI with plots too. We used python xlswriter to create our easy to use Excel files.
NDVI = ( NIR - Red ) / ( NIR + Red )
NDVI = (860nm - 645nm) / (860nm + 645nm)
We are testing use of NIRv in certain situations related to phenology of plant cycles (photosynthesis vs. transpiration). The NDVI reading for grass are nearly the same, but the NIRv for grass in shade is much lower than the NIRv of grass in bright sunlight.
NIRv = NDVI * NIR = NDVI * 860nm
NDVI and now NIRv can be very good attributes to be used to differentiate Vegetative species.
Most examples notebooks have the raw STELLA readings along with readings from our phtotographic White Card that is used for calibration. The entire process, including calibration to the white card readings, is demonstrated in these examples. The following animation is from our White Card readings in Full Sun, grass readings in full sun, grass readings in shade and White Card readings in shade too. The White Card calibration allows us to normalize all of these readings in these different lighting conditions.
As of this date we have been using photographic 18% gray and white cards for our calibration, but we will be going to white polystyrene foam per the advice from Paul Mirel which is reported to be a "broadband spectrally flat reflector". This sounds ideal and much cheaper than Spectralon. White PTFE (Polytetrafluoroethylene) Sheets are also an option.
The following are the python libraries that are required to run these notebooks:
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import re
import ipywidgets as widgets
from IPython.display import display
We have been using an Anaconda installation of python, and all of these libraries were readily available, but we are trying to make these methods available on Google CoLab where the data will be pulled from this GitHub repository and all calculations can then be run from your web browser. A Google account will be required for this and then we can link you to our notebooks upon your request.
We did changed the header portion of the code.py file on the STELLA-Q2 for the near IR wavelengths to include a color name as a part of the column description. Our version of the code.py file works well and now has matplotlib recognized color names embedded in the column names of this header file for plotting purposes. Please find in this repository the original code_original.py and our new STS version of code.py that we are using on our STELLA-Q2. The near IR regions can be distinguished on our spectral plots since these wavelengths have a lavender background and are located at the long end of the spectrum. We hope this is not too confusing.
We extend our heartfelt appreciation to the NASA STELLA Team for their unwavering support of our STS STELLA project. Paul Mirel, the visionary creator and lead Engineer of the STELLA project, along with Mike Taylor, the esteemed Team Leader, and Petya Campbell, the Lead Scientist, alongside all contributors to the STELLA endeavor, have been invaluable in propelling our project forward. Their provision of tools, guidance, and expertise has been instrumental in our journey. STS is profoundly grateful for their steadfast technical support, which has been pivotal in our progress.
For more information, please go to the following website:
https://landsat.gsfc.nasa.gov/stella/
A link to the updated STELLA-Q2 drone housing 3D print files with instructions and photos from Mike Taylor can be found below:
https://landsat.gsfc.nasa.gov/wp-content/uploads/2024/02/Q2-drone-parts-1.zip
(1) North Carolina Geographic Information Coordinating Council by the Statewide Mapping Advisory Committee, Working Group for Orthophotography Planning, July 2011.
(2) Charlotte Harbor National Estuary Program Technical Report 16-3, Identifying and Diagnosing Locations of Ongoing and Future Saltwater Wetland Loss: Mangrove Heart Attack, 3/9/2017