In loving memory of Larry Corp.
What is G-LiHT?
Goddard’s LiDAR, Hyperspectral, & Thermal Imager (G-LiHT) is an airborne instrument designed to map the composition of forested landscapes.
The G-LiHT instrument has a number of sensors that each serve a specific purpose. There are two LiDAR sensors that produce a series of LiDAR-derived forest structure metrics including a canopy height model, surface model, and digital terrain model. These models allow us to measure tree height and biomass volume.
Additionally, there are two cameras: one visible and one near-infrared (NIR). The visible and NIR bands acquired by the two cameras are paired to produce 4-band imagery. The 3-centimeter resolution photos taken by these cameras are aligned to build orthomosaics, which allow us to visually observe and identify changes in forest composition.
G-LiHT also has a hyperspectral sensor to acquire spectral information at a coarser resolution. These data can be used to identify vegetation composition and measure photosynthetic function as well as calculate vegetation indices at a fine spectral scale of 1 meter using radiometrically calibrated surface reflectance data.
The thermal sensor measures radiant surface temperature which allows us to create 3D temperature profiles derived from structure-for-motion. Thermal data provides us with information on the functional aspects of forest canopies. As photosynthetic function is related to evapotranspiration, we can observe that hotter canopies are more stressed relative to surrounding canopies.
Purpose of the field campaign
The G-LiHT airborne mission supports multiple groups including the U.S. Forest Service (USFS), the USFS Geospatial Technology and Applications Center (GTAC), and the University of Alaska Anchorage.
The USFS is creating a forest inventory for the state of Alaska, and G-LiHT measurements collected over Forest Inventory and Analysis (FIA) plots are a cost-effective method of forest inventory. G-LiHT data will also help to improve regional estimates of aboveground forest biomass and terrestrial ecosystem carbon stocks. GTAC uses G-LiHT data measurements for algorithm development. USFS Geospatial Technology and Applications Center will use G-LiHT data acquired over FIA- and GTAC-measured ground plots and between these plots to map forest characteristics on federally managed lands, including forest type, biomass, vegetation structure, tree and shrub cover, and more. Data will also be used to guide future inventory efforts in coastal Alaska using methods developed for interior Alaska.
This field campaign also acquired repeat data over Fairbanks, Alaska, to measure changes in permafrost.
G-LiHT image data was reacquired over spruce beetle monitoring transects stretching from the Kenai Peninsula in the south to Denali National Park in the north. These transects were last measured on the ground and with G-LiHT in 2018, during the peak of a spruce beetle outbreak, and changes in vegetation structure and spectral reflectance will be used to evaluate the long-term mortality and growth of these forests.
Integrative test flight
Our Alaskan field campaign started with an integrative test flight in June. Our team of three loaded up G-LiHT into a vehicle much too small and drove to Dynamic Aviation in Bridgewater, Virginia. We spent the first day installing the instrument into a 1960s King Air A90.
The second day was all about flying. We needed to make sure G-LiHT didn’t interfere with any of the aircraft’s systems. Additionally, the functional test flight over Harrisonburg, Virginia, allowed us to verify that G-LiHT was functioning properly. We flew in a grid pattern over the city which allowed us to geospatially align the data products from all of G-LiHT’s sensors.
The integrative test flight was a success. We installed G-LiHT properly with no issues and obtained the information we needed. Once we received the thumbs up to proceed with our campaign, the pilots loaded up the plane with supplies and headed out to Kodiak, where we would meet them the following week.
Our plan for the field campaign was to arrive in Kodiak, Alaska on July 6 and stay until the end of the month. We chose Kodiak as our hub because it was a convenient location to our flight lines. Unfortunately, despite the ideal location, poor weather prevented us from flying for the first three days of the campaign.
Once we were finally able to get in the air, we collected data over the forests near Iliamna.
Most of our days consisted of our team meeting in the hotel for breakfast at 8 a.m., discussing weather and flight plans for the day, and then driving to the airport to prepare the plane and G-LiHT for flying. Depending on how many flight lines we were able to complete, we often stopped in King Salmon or Iliamna to refuel the plane and then went back out to fly more lines before returning to Kodiak.
Our group was interested in measuring the effects of forest fires on vegetation in the Dillingham region. There were several burned areas to the west of the Nuyakuk River and east of Cook Inlet.
Toward the end of the campaign, we decided to transit to Fairbanks because the weather over the rest of our other flight lines didn’t look promising. If there were clouds below the plane at 1,100 feet, they would obstruct the instrument’s view and cast shadows on our data. We had to closely monitor the weather every morning. Additionally, we were unable to fly in rain or smoke as it would adversely affect the LiDAR sensors’ data returns.
One geological feature we saw extensively in the southwest was the oxbow lake. Also called cut-off lakes, these lakes have formed when meandering rivers erode at points of inflection because of sediments flowing along them to the point where two parts of the river will join together, creating a new straight part of the river—essentially “cutting off” the curved lake piece. This created an oxbow lake. Once the lake has fully dried out, it becomes a meander scar. We noted the difference in vegetation growing back within the oxbow lakes and meander scars and how this differs from surrounding vegetation patterns.
We had only planned to spend one night in Fairbanks, then transit back to Kodiak the following day. However, the weather had other plans for us. We ended up having to fly to Anchorage the following day because of extremely low cloud ceilings in Kodiak that made it too dangerous to land there. It worked out in the end, and the team was able to see more of beautiful Alaska and collect data over Anchorage and the Chugach region. It just goes to show how quickly things can change during a field campaign.
We collected data in the Campbell Creek region west of Anchorage. The data include visible and near-infrared photos which were composited into 4-band high-resolution orthomosaics and used to visually observe and identify changes in forest composition.
In addition to the high-resolution orthomosaics produced from the G-LiHT’s near-infrared and visible cameras, LiDAR data was processed to create various 1-meter resolution forest structure metrics including Digital Terrain Model (DTM), Digital Surface Model (DSM) and Canopy Height Model (CHM). These metrics are used to measure tree height and biomass volume. The CHM raster below was created by subtracting the DTM from the DSM.
After collecting data in Anchorage and the Chugach region of Alaska, the team flew back to Kodiak and finished data acquisition in the southwest.
And of course it wouldn’t be Alaska without some wildlife. The day before leaving Kodiak, I got to see not just one bear—but a family of four! Cars were honking to scare the bears out of the road, but luckily I had enough time to snap a picture before the bears ran off into the woods. It was the perfect end to an exciting field campaign.
Tags: airborne science, G-LiHT
As a Science Olympiad coach, I’m just curious: Could this survey have been conducted from satellites? Why not? Is it a matter of orbital paths, or expense, or clouds, or atmospheric opacity in the required wavelengths, or the need for better resolution than possible from space? In any event, what an adventure! Thanks for your work!