ABoVE: Land Cover, Methane Flux, and Environmental Data, Big Trail Lake, Fairbanks AK

This dataset provides gridded land cover classifications for the area north of Big Trail Lake, an active thermokarst lake in Goldstream Valley (near Fairbanks, Alaska, USA) at 10 cm resolution. Land cover classifications were derived from a supervised random forest classification using RedEdgeMX and multiSPEC4C multispectral drone imagery, normalized difference vegetation index (NDVI), and the USGS 3D Elevation Program (3DEP) digital surface model (DSM) elevation data. RedEdgeMX data was collected in July 2021 on and multiSPEC4C on 2019-08-04. Additionally, gridded microtopography and slope estimates were derived at 1m resolution. Field samples were collected at 15 sites across the study area in June 2021 and August 2022. Field data include methane (CH4) and carbon dioxide (CO2) fluxes, gross primary production (GPP), SIMPER microbial community analysis, methanogen abundance, soil characteristics, and meteorological characteristics. Gridded products are provided in cloud-optimized GeoTIFF format, field data are provided in comma-separated values format, and sample location photos in JPEG format are contained within a compressed file.

This dataset includes 14 data files: 10 in comma-separated values format (*.csv), three in cloud-optimized GeoTIFF (*.tif) format, and JPEG images in one compressed (*.zip) file.

This dataset provides gridded land cover classifications for the area north of Big Trail Lake, an active thermokarst lake in Goldstream Valley (near Fairbanks, Alaska, USA) at 10 cm resolution. Land cover classifications were derived from a supervised random forest classification using RedEdgeMX and multiSPEC4C multispectral drone imagery, normalized difference vegetation index (NDVI), and the USGS 3D Elevation Program (3DEP) digital surface model (DSM) elevation data. RedEdgeMX data was collected in July 2021 on and multiSPEC4C on 2019-08-04. Additionally, gridded microtopography and slope estimates were derived at 1m resolution. Field samples were collected at 15 sites across the study area in June 2021 and August 2022. Field data include methane (CH4) and carbon dioxide (CO2) fluxes, gross primary production (GPP), SIMPER microbial community analysis, methanogen abundance, soil characteristics, and meteorological characteristics. 

Project:  Arctic-Boreal Vulnerability Experiment (ABoVE)

The Arctic-Boreal Vulnerability Experiment (ABoVE) is a NASA Terrestrial Ecology Program field campaign being conducted in Alaska and western Canada, for 8 to 10 years, starting in 2015. Research for ABoVE links field-based, process-level studies with geospatial data products derived from airborne and satellite sensors, providing a foundation for improving the analysis, and modeling capabilities needed to understand and predict ecosystem responses to, and societal implications of, climate change in the Arctic and Boreal regions.

Related Publication:

Farina, M., W. Christian, N. Hasson, T. McDermott, S. Powell, R. Hatzenpichler, H. Webb, G. LaRue, K. Okano, E. Sproles, J. Watts. 2025. Methane flux spatial heterogeneity driven by surface and subsurface conditions in boreal forest-fen mosaic. [Manuscript submitted to Environmental Research Letters for publication.]

Related Data:

Barnes, N., H. Webb, M.K. Farina, S. Powell, and J.D. Watts. 2021. Multispectral Imagery, NDVI, and Terrain Models, Big Trail Lake, Fairbanks, AK, 2019. ORNL DAAC, Oak Ridge, Tennessee, USA. https://doi.org/10.3334/ORNLDAAC/1834

• The multiSPEC4C imagery and NDVI data from Barnes et al. (2021) data were used to produce the land cover classification data in this dataset.

Spatial Coverage: Site approximately 150 m north of Big Trail Lake, located in Goldstream Valley near Fairbanks, Alaska, US

Spatial Resolution: 

• DTM_DSM_Microtopography_Slope_1m.tif: 1 m
• Land_Classification_Inputs_10cm.tif and Land_Classification_Outputs_10cm.tif: 10 cm
• Field samples: point locations

Temporal Coverage: 2019-08-04 to 2022-08-23

Temporal Resolution: Flux samples were collected 1-2 times during the July 2021 and August 2022 field surveys. Drone imagery was collected twice, on 2019-08-04 (RedEdgeMX) and in July 2021 (multiSPEC4C). Meteorological variables were collected hourly from 2021-07-07 to 2021-07-27. Six soil loggers collected soil temperature data half-hourly from 2021-07-10 to 2021-07-28.

Study Areas: Latitude and longitude are given in decimal degrees.

Site	Westernmost Longitude	Easternmost Longitude	Northernmost Latitude	Southernmost Latitude
Goldstream Valley, Alaska, US	-147.82	-147.82	64.92	64.92

Data File Information

This dataset includes 14 data files: 10 in comma-separated values format (*.csv), three in cloud-optimized GeoTIFF (*.tif) format, and JPEG images in one compressed (*.zip) file.

Missing numeric data are indicated by -9999 and missing text data are represented by N/A.

GeoTIFF Files

This dataset contains three cloud-optimized GeoTIFF (*.tif) files:

• DTM_DSM_Microtopography_Slope_1m.tif contains four bands of lidar-derived or estimated elevation characteristics over the study area at one meter resolution. Projection: UTM zone 6N (EPSG: 32606). Map units: meters.
  • Band 1 (USGS_3DEP_2017_Fairbanks_Lidar_DTM_aggregated): Digital terrain model (DTM) showing the elevation of the underlying terrain of the earth’s surface in meters. Data are from USGS 3D Elevation Program DTM (USGS, 2017).
  • Band 2 (USGS_3DEP_2017_Fairbanks_Lidar_DSM_aggregated): Digital surface model (DSM) showing elevation values of landscape features on the earth's surface in meters. Data are from USGS 3D Elevation Program DSM (USGS, 2017).
  • Band 3 (Microtopography_meters): Gridded estimates of microtopography
  • Band 4 (Slope_8pixel_neighborhood_degrees): Gridded estimates of slope calculated from 8-pixel neighborhood
• Land_Classification_Inputs_10cm.tif contains 12 bands of land classification inputs used to produce land cover gridded outputs (Land_Classification_Outputs_10cm.tif) at 10 cm resolution. Projection: WGS 84 (EPSG: 4326). Map units: decimal degrees
  • Band 1 (Blue_REMX): Blue band from drone RedEdge-MX sensor.
  • Band 2 (Green_REMX): Green band from drone RedEdge-MX sensor.
  • Band 3 (Red_REMX): Red band from drone RedEdge-MX sensor.
  • Band 4 (RedEdge_REMX): Red-edge band from drone RedEdge-MX sensor.
  • Band 5 (NIR_REMX): Near-infrared (NIR) band from drone RedEdge-MX sensor.
  • Band 6 (DSM_REMX): Digital surface model (DSM; ellipsoidal elevation in meters) from drone RedEdge-MX sensor.
  • Band 7 (Green_MS4C): Green band from drone multiSPEC-4C sensor (resampled to match pixel grid of RedEdge-MX imagery).
  • Band 8 (Red_MS4C): Red band from drone multiSPEC-4C sensor (resampled to match pixel grid of RedEdge-MX imagery).
  • Band 9 (NIR_MS4C): Near-infrared (NIR) band from drone multiSPEC-4C sensor (resampled to match pixel grid of RedEdge-MX imagery).
  • Band 10 (DSM_MS4C): Digital surface model (DSM; orthometric elevation in meters) from drone multiSPEC-4C sensor (resampled to match pixel grid of RedEdge-MX imagery).
  • Band 11 (DTM_MS4C): Digital terrain model (DTM; orthometric elevation in meters) from drone multiSPEC-4C sensor (resampled to match pixel grid of RedEdge-MX imagery).
  • Band 12 (USGS_3DEP_DSM_resampled): USGS 3DEP digital surface model (DSM; orthometric elevation in meters), resampled to match pixel grid of RedEdge-MX imagery.
• Land_Classification_Outputs_10cm.tif contains three bands of land cover classifications derived using supervised random forest classification at 10 cm resolution. Land cover classifications are represented by integer values where 1=Graminoid; 2=Moss; 3=Shadow; 4=Water; and 5=Woody. Projection: WGS 84 (EPSG: 4326). Map units: decimal degrees.
  • Band 1 (Combined_landcover_prediction): The combined land cover classification from the models using RedEdgeMX and multiSPEC4C multispectral drone imagery.
  • Band 2 (Predicted_landcover_using_RedEdgeMX_imagery): Land cover classification based on RedEdge-MX multispectral drone imagery.
  • Band 3 (Predicted_landcover_using_multiSPEC4C_imagery): Land cover classification based on multiSPEC-4C multispectral drone imagery.

CSV Files

This dataset contains 10 comma-separated values (CSV) files:

• Table_1_Chamber_Fluxes_Insitu_Data_July_2021.csv: in situ methane flux observations collected in July 2021 and modeled variables.
• Table_2_Detailed_Chamber_Flux_Estimates_July_2021_August_2022.csv: Contains chamber methane (CH₄) and carbon dioxide (CO₂) flux estimates and GPP estimates from field surveys conducted in July 2021 and August 2022.
• Table_3_Methanogen_Methanotroph_Relative_Abundance.csv: relative abundance observations of methanogens and methanotrophs at for four soil depths (0-12.5 cm, 12.5-25 cm, 25-37.5 cm, 37.5 - 50 cm).
• Table_4_Soil_Sample_Chemical_Analysis.csv: Contains soil chemical analyses that were conducted on soil samples from 17 locations at various depths.
• Table_5_Simper_Microbial_Community_Analysis.csv: SIMPER microbial community analysis results for each amplicon sequence variant (ASV).
• Table_6_Qiime2_taxonomy_output.csv: the raw output from Qiime2 which includes the total number of reads assigned to each ASV in each sample. The header rows of the file are structured such that row 1 contains the Sample_ID, row 2 contains the sample depth in cm (depth_cm), row 3 is blank, and row 4 contains the Soil sample ID number (#OTU ID). The following rows contain the taxaon in column 1 and total count of reads assigned to each ASV in each soil sample in subsequent columns.
• Table_7_Hourly_Meteorology_Weather_Sensor.csv: meteorology data from weather station at hourly resolution from 2021-07-07- to 2021-07-27. The weather station was located at 64.91984738, -147.81891737.
• Table_8_Halfhourly_Soil_Temperature_Logger_Data.csv: Contains soil temperature data at half-hourly resolution 2021-07-10 to 2021-07-28.
• Table_9_Training_Testing_Data_RedEdgeMX_Classification.csv: training and testing datasets used to calibrate and validate the supervised random forest land cover classification based on RedEdge-MX multispectral drone imagery.
• Table_10_Training_Testing_Data_multiSPEC4C_Classification.csv: training and testing data used to calibrate and validate the supervised random forest land cover classification based on RedEdge-MX multispectral drone imagery.

Table 1. Data dictionary for Table_1_Chamber_Fluxes_Insitu_Data_July_2021.csv

Variable	Units	Description
Sample_ID		Sample location ID (56 unique IDs)
Site_ID		Site ID (15 unique IDs)
Latitude	degrees north	Sample location latitude in decimal degrees
Longitude	degrees east	Sample location longitude in decimal degrees
Vegetation_class		Vegetation class of sample: 1. Moss, sedge, field horsetail; 2. Sedge, grass, moss; 3. Moss, field horsetail, grass, sedge; 4. Variegated horsetail, moss, sedge; 5. Grass, sedge, moss; 6. Field horsetail, moss, grass, sedge; 7. Moss, field horsetail; 8. Dwarf shrub, moss, field horsetail, grass, sedge; 9. Moss.
Vegetation_details		Vegetation details at individual sample locations.
Mean_CH4_flux	mg m-2 d-1	Mean CH4 flux across three repetitions
CH4_flux_category		CH4 flux category (VH=Very High; High=High; Int=Intermediate; Low=Low; Neg=Negative)
VWC_6cm	percent	Soil volumetric water content at 6 cm depth
VWC_12cm	percent	Soil volumetric water content at 12 cm depth
VWC_20cm	percent	Soil volumetric water content at 20 cm depth
SoilT_5cm	degrees C	Soil temperature at 5 cm depth
SoilT_10cm	degrees C	Soil temperature at 10 cm depth
Soil_pH_3cm		Soil pH at 2.5 cm depth

 

Table 2. Data dictionary for Table_2_Detailed_Chamber_Flux_Estimates_July_2021_August_2022.csv

Variable	Units	Description
Sample_ID		ID of sample location (56 unique sample locations). An asterisks (*) indicates flux samples that were taken in exactly the same locations as the July 2021 samples. Double asterisks (**) indicates flux samples that were taken in approximately the same locations as the July 2021 samples.
Chamber		LI-COR 8200-01S Smart Chamber (SC); Polycarbonate chamber with opaque cover (PC_cover); Polycarbonate chamber without opaque cover (PC_no_cover)
Rep		Repetition (1-3)
Date	YYYY-MM-DD	Date of flux sampling.
Time_AKDT		Time of flux sampling (Alaska Daylight Time).
Orig_CH4_flux	nmol m-2 s-1	CH4 flux before adjusting for chamber headspace volume filled by vegetation.
Orig_CO2_flux	umol m-2 s-1	CO2 flux before adjusting for chamber headspace volume filled by vegetation.
Veg_vol	percent	Estimated percent headspace volume filled by vegetation.
Adj_CH4_flux	nmol m-2 s-1	CH4 flux after adjusting for chamber headspace volume filled by vegetation.
Adj_CO2_flux	umol m-2 s-1	CO2 flux after adjusting for chamber headspace volume filled by vegetation.
GPP	umol m-2 s-1	Gross Primary Productivity.

 

Table 3. Data dictionary for Table_3_Methanogen_Methanotroph_Relative_Abundance.csv

Variable	Units	Description
Sample_ID		Sample location ID (56 unique IDs)
Depth_cm	cm	Soil sample depth
Methanogen_Rel_Abundance	percent	Methanogen relative abundance
Methanotroph_Rel_Abundance	percent	Methanotroph relative abundance

 

Table 4. Data dictionary for Table_4_Soil_Sample_Chemical_Analysis.csv

Variable	Units	Description
Sample_ID		Sample location ID (56 unique IDs)
Depth_cm	cm	Soil sample depth
pH		Soil pH
OM	percent	Organic matter
EC	mmhos cm-1	Electrical conductivity
Nitrate_N	ppm	Nitrate-nitrogen concentration
NH4_N	ppm	Ammonium-nitrogen concentration
P_Olsen	ppm	Phosphorus (Olsen test) concentration
K	ppm	Potassium concentration
Ca	ppm	Calcium concentration
Mg	ppm	Magnesium concentration
Na	ppm	Sodium concentration
S	ppm	Sulfate-sulfur concentration
CEC	meq (100 g)-1	Cation exchange capacity in milliequivalents per 100 grams of soil
K_base_sat	percent	Base saturation K
Ca_base_sat	percent	Base saturation Ca
Mg_base_sat	percent	Base saturation Mg
Na_base_sat	percent	Base saturation Na
CCE	percent	Calcium carbonate equivalent

 

Table 5. Data dictionary for Table_5_Simper_Microbial_Community_Analysis.csv

Variable	Units	Description
SIMPER_Order		Simper Order
ASV		Amplicon sequence variant (ASV)
More_abundant_in_Intermediate_CH4_flux_samples		Was the AVS more abundant in Intermediate CH4 flux samples? (yes/no)
Domain		Domain
Phyla		Phyla
Class		Class
Order		Order
Family		Family
Genus		Genus
Cumulative_Sum	1	The cumulative contribution of each ASV to the difference between two groups (samples from the intermediate CH4 flux category, and samples from all other CH4 flux categories). The column is ordered with the highest contributors at the top.Thus the cumulative sum in the first row is the contribution from the 1st most distinguishing organism, the second row is the sum of rows 1 and 2, and so on. Units = proportion.
p_value	1	Permutation-based p-value. The probability of getting a larger or equal average contribution for each ASV if the grouping factor (CH4 flux category) was randomly permuted

 

Table 6. Table_6_Qiime2_taxonomy_output

Variable	Units	Description
Sample_ID		Sample location ID (56 unique IDs)
Depth_cm	cm	Soil sample depth (cm)
OTUID		Soil sample ID number. Note that AK64 , AK65, AK66, AK67 are DNA extraction blanks and PCRB1, PCRB1D, PCRB1Z, PCRB5, PCRB5Z PCR blanks

 

Table 7. Data dictionary for Table_7_hourly_meteorology_weather_sensor.csv

Variable	Units	Description
Date_time_AKDT	YYYY-MM-DD hh:mm:ss	Local date and time (Alaska Daylight Time, UTC−08:00)
RECORD		Record number
BattV_Avg	V	Battery voltage (Avg)
BattV_Max	V	Battery voltage (Max)
BattV_Min	V	Battery voltage (Min)
BattV	V	Battery voltage (Smp - the current/instantaneous value at the time of measurement)
BattV_Std	V	Battery voltage (Std)
PTemp_C_Avg	degrees C	Average Wiring Panel Temperature of the Campbell Scientific CR300 data logger (Avg)
PTemp_C_Max	degrees C	Average Wiring Panel Temperature of the Campbell Scientific CR300 data logger (Max)
PTemp_C_Min	degrees C	Average Wiring Panel Temperature of the Campbell Scientific CR300 data logger (Min)
PTemp_C	degrees C	Average Wiring Panel Temperature of the Campbell Scientific CR300 data logger (Smp - the current/instantaneous value at the time of measurement)
PTemp_C_Std	degrees C	Average Wiring Panel Temperature of the Campbell Scientific CR300 data logger (Std)
SlrFD_W_Avg	W m-2	Solar lux density (Avg)
SlrFD_W_Max	W m-2	Solar flux density (Max)
SlrFD_W_Min	W m-2	Solar flux density (Min)
SlrFD_W	W m-2	Solar flux density (Smp - the current/instantaneous value at the time of measurement)
SlrFD_W_Std	W m-2	Solar flux density (Std)
Rain_mm_Tot	mm	Total rainfall (Tot)
Strikes_Tot	1	Count of lightning strikes (Tot)
Dist_km_Avg	km	Distance to lightning strikes (Avg)
Dist_km_Max	km	Distance to lightning strikes (Max)
Dist_km_Min	km	Distance to lightning strikes (Min)
Dist_km	km	Distance to lightning strikes (Smp - the current/instantaneous value at the time of measurement)
Dist_km_Std	km	Distance to lightning strikes (Std)
WS_ms_Avg	m s-1	Wind speed (Avg)
WS_ms_Max	m s-1	Wind speed (Max)
WS_ms_Min	m s-1	Wind speed (Min)
WS_ms	m s-1	Wind speed (Smp - the current/instantaneous value at the time of measurement)
WS_ms_Std	m s-1	Wind speed (Std)
WS_ms_S_WVT	m s-1	Wind speed (WVc)
WindDir_D1_WVT	degrees	Wind direction (WVc)
WindDir_SD1_WVT	degrees	Wind direction (WVc)
WindDir	degrees	Wind direction (Smp)
MaxWS_ms_Avg	m s-1	Max wind speed (Avg)
MaxWS_ms_Max	m s-1	Max wind speed (Max)
MaxWS_ms_Min	m s-1	Max wind speed (Min)
MaxWS_ms	m s-1	Max wind speed (Smp - the current/instantaneous value at the time of measurement)
MaxWS_ms_Std	m s-1	Max wind speed (Std)
AirT_C_Avg	degrees C	Air temperature (Avg)
AirT_C_Max	degrees C	Air temperature (Max)
AirT_C_Min	degrees C	Air temperature (Min)
AirT_C	degrees C	Air temperature (Smp - the current/instantaneous value at the time of measurement)
AirT_C_Std	degrees C	Air temperature (Std)
VP_mbar_Avg	mbar	Vapor pressure (Avg)
VP_mbar_Max	mbar	Vapor pressure (Max)
VP_mbar_Min	mbar	Vapor pressure (Min)
VP_mbar	mbar	Vapor pressure (Smp - the current/instantaneous value at the time of measurement)
VP_mbar_Std	mbar	Vapor pressure (Std)
BP_mbar_Avg	mbar	Barometric pressure (Avg)
BP_mbar_Max	mbar	Barometric pressure (Max)
BP_mbar_Min	mbar	Barometric pressure (Min)
BP_mbar	mbar	Barometric pressure (Smp - the current/instantaneous value at the time of measurement)
BP_mbar_Std	mbar	Barmetric pressure (Std)
RH_Max	percent	Relative humidity (Max)
RH_Min	percent	Relative humidity (Min)
RH	percent	Relative humidity (Smp - the current/instantaneous value at the time of measurement)
RHT_C_Avg	degrees C	Temperature reading from the relative humidity & temperature sensor (Avg)
RHT_C_Max	degrees C	Temperature reading from the relative humidity & temperature sensor (Max)
RHT_C_Min	degrees C	Temperature reading from the relative humidity & temperature sensor (Min)
RHT_C	degrees C	Temperature reading from the relative humidity & temperature sensor (Smp - the current/instantaneous value at the time of measurement)
RHT_C_Std	degrees C	Temperature reading from the relative humidity & temperature sensor (Std)
TiltNS_deg_Avg	degrees	North/South orientation of the ClimaVue instrument (Avg)
TiltNS_deg_Max	degrees	North/South orientation of the ClimaVue instrument Tilt (Max)
TiltNS_deg_Min	degrees	North/South orientation of the ClimaVue instrument Tilt (Min)
TiltNS_deg	degrees	North/South orientation of the ClimaVue instrument Tilt (Smp - the current/instantaneous value at the time of measurement)
TiltNS_deg_Std	degrees	North/South orientation of the ClimaVue instrument Tilt (Std)
TiltWE_deg_Avg	degrees	West/East orientation of the ClimaVue instrument Tilt (Avg)
TiltWE_deg_Max	degrees	West/East orientation of the ClimaVue instrument Tilt (Max)
TiltWE_deg_Min	degrees	TWest/East orientation of the ClimaVue instrument Tilt (Min)
TiltWE_deg	degrees	West/East orientation of the ClimaVue instrument Tilt (Smp - the current/instantaneous value at the time of measurement)
TiltWE_deg_Std	degrees	West/East orientation of the ClimaVue instrument Tilt (Std)
SlrTF_MJ_Tot	MJ m-2	Total solar radiation over the measurement interval
CVMeta		Data logger ID
Invalid_Wind_Avg	1	Count of invalid wind measurements (Avg).
Invalid_Wind_Max	1	Count of invalid wind measurements (Max).
Invalid_Wind_Min	1	Count of invalid wind measurements (Min).
Invalid_Wind	1	Count of invalid wind measurements (Smp - the current/instantaneous value at the time of measurement).
Invalid_Wind_Std	1	Count of invalid wind measurements (Std).
Invalid_Wind_Tot	1	Count of invalid wind measurements (Tot).

 

Table 8. Data dictionary for Table_8_Halfhourly_Soil_Temperature_Logger_Data.csv

Variable	Units	Description	Logger Latitude	Logger Longitude
Date_time_AKDT	YYYY-MM-DD hh:mm:ss	Date and time (Alaska Daylight Time)
SoilT_C_logger1	degrees C	Soil temperature logger-1	64.92185221	-147.8203832
SoilT_C_logger2	degrees C	Soil temperature logger-2	64.92252362	-147.8204773
SoilT_C_logger3	degrees C	Soil temperature logger-3	64.9219876	-147.8192036
SoilT_C_logger4	degrees C	Soil temperature logger-4	64.92197142	-147.8212274
SoilT_C_logger5	degrees C	Soil temperature logger-5	64.92243131	-147.8210107
SoilT_C_logger6	degrees C	Soil temperature logger-6	64.9223092	-147.8198968

 

Table 9. Data dictionary for Table_9_Training_Testing_Data_RedEdgeMX_Classification.csv

Variable	Units	Description
Latitude	degrees north	Latitude of training/testing point in decimal degrees.
Longitude	degrees east	Longitude of training/testing point in decimal degrees.
train_test		Indicates training or testing point.
Class		Ground-truth based on visual assessment of high-resolution drone imagery (1=Graminoid; 2=Moss; 3=Shadow; 4=Water; 5=Woody).
remx_blue	1	RedEdge-MX blue band; brightness value as a digital number.
remx_green	1	RedEdge-MX green band; brightness value.
remx_red	1	RedEdge-MX red band; brightness value..
remx_rededge	1	RedEdge-MX rededge band; brightness value.
remx_NIR	1	RedEdge-MX near-infrared (NIR) band; brightness value.
remx_NDVI	1	Normalized vegetation difference index (NDVI) calculated from RedEdge-MX imagery.
USGS_DSM	m	USGS 3DEP digital surface model (DSM; orthometric elevation)

 

Table 10. Data dictionary for Table_10_Training_Testing_Data_multiSPEC4C_Classification.csv

Variable	Units	Description
Latitude	degrees north	Latitude of training/testing point in decimal degrees.
Longitude	degrees east	Longitude of training/testing point in decimal degrees.
train_test		Indicates training or testing point.
Class		Ground-truth based on visual assessment of high-resolution drone imagery (1=Graminoid; 2=Moss; 3=Shadow; 4=Water; 5=Woody).
ms4c_NIR	1	multiSPEC-4C green band; reflectance
ms4c_red	1	multiSPEC-4C red band; reflectance.
ms4c_green	1	multiSPEC-4C near-infrared (NIR) band; reflectance.
ms4c_NDVI	1	Normalized vegetation difference index (NDVI) calculated from multiSPEC-4C imagery.
USGS_DSM	m	USGS 3DEP digital surface model (DSM; orthometric elevation)

Compressed Zip File

The dataset contains one compressed file, Sample_Location_Photos_July_2021.zip. Enclosed are 326 photos in JPEG (*jpg) format that were taken at soil sampling locations in 2021. There are 15 subdirectories, one for each Site_ID. The JPEG files are named YYYYMMDD_hhmmss_ Sample_ID.jpg, where YYYYMMDD_hhmmss is the timestamp and Sample_ID­ is the sample location ID.

This dataset provides high-resolution multispectral drone imagery (spatial resolution ~10 cm) and land cover maps (spatial resolution 1 m) over a boreal forest-fen mosaic ecosystem. Most variables were collected (or modeled) over multiple points in space (single time point), which may offer insight in the spatial relationships between carbon fluxes and driver variables (e.g., spatial relationships between CH₄ flux and soil moisture, or between CH₄ flux and soil microbial community composition). Additionally, soil microbial community analyses provide information about the community structure across space and depth.

Please see Farina et al. 2025 for additional details

Chamber flux estimates: Three of the 168 samples had poor model fits with either the CH₄ or CO₂ concentration data and were removed from the dataset. Among the remaining 165 CH4 flux estimates, 115 were fitted with exponential models, and 50 were fitted with linear models. The majority of these fitted models (96%) had R² values greater than 0.9, and the remaining 4% of fitted models had R² values between 0.77 and 0.87. The three replicate CH₄ fluxes were averaged to calculate a mean flux per sample location. Among the 165 CO₂ flux (ecosystem respiration; Reco) estimates, 158 samples were fitted with exponential models, and seven were fitted with linear models (R² values ranging 0.94 – 0.99).

Soil moisture measurements: Across the 56 sampling locations, the Hydrosense II measurements of period (τ) ranged 2.339 – 3.832 μs for the 20 cm rod inserted vertically; 1.273 – 2.402 μs for the 12 cm rod inserted vertically; and 1.172 – 2.367 μs for the 12 cm rod inserted at a 30° angle. These measurements fell within valid ranges of τ. For the 20 cm rod, the valid range of τ is 1.38 μs (in air) to 3.89 μs (in water); for the 12 cm rod, τ ranges from approximately 1.0 μs (in air) to 2.47 μs (in water) (Campbell Scientific).

Other in situ measurements: The HI98331 probe has a soil temperature accuracy of ±1 °C, and the HI981030 Soil pH tester has an accuracy of ±0.05 pH.

Please see Farina et al. 2025 for additional details.

GeoTIFF files

The USGS 3DEP DTM and DSM rasters (tile ID 4520; 0.5 m spatial resolution; USGS, 2017) were georeferenced to match the RedEdge-MX drone imagery (~10 cm spatial resolution; collected July 2021) as precisely as possible. Georeferencing was done using a third degree polynomial transformation algorithm and nearest neighbor resampling. The georeferenced DTM raster was aggregated to 1 m spatial resolution and projected to the WGS 84 / UTM Zone 6N coordinate reference system (EPSG:32606). The aggregated raster was used for subsequent modeling. The local mean DTM elevation was calculated for a 31 by 31 m moving window, and per-grid cell microtopography was calculated as the local mean subtracted from the 1-m DTM values. Gridded slope estimates were calculated using 8-pixel neighborhoods.

There are limitations to the approach used to model microtopography. USGS 3DEP digital terrain model (DTM) data was smoothed by calculating the moving window (31 m x 31 m) mean value. Per-grid cell microtopography was calculated as the local mean subtracted from the original DTM. However, the optimal degree of smoothing (i.e., moving window size) is uncertain. The 31-m window size was selected because the resulting smoothed DTM seemed to capture the broad scale gradient, but a larger or smaller window size could have a substantial impact on modeled topography (Erdbrügger et al., 2021).

The multiSPEC-4C imagery is described in more detail by Barnes et al. (2021).

Land cover classes (graminoid, moss, woody vegetation, shadow, and surface water) were mapped across the study area using supervised random forest classification informed by the drone multispectral imagery, normalized difference vegetation index (NDVI), and the USGS 3D Elevation Program (3DEP) digital surface model (DSM) elevation. Separate random forest classification models were calibrated and validated using the multiSPEC-4C imagery collected in August 2019 and the RedEdge-MX imagery collected in July 2021. In the multiSPEC-4C classification, predictor variables included the green, red, near-infrared (NIR), and NDVI layers from the multiSPEC-4C sensor, and 3DEP DSM. In the RedEdge-MX classification, predictor variables included the blue, green, red, red edge, NIR, and NDVI layers from the RedEdge-MX sensor, and 3DEP DSM.

Each classification model was calibrated using a dataset of 90 observations (20 graminoid, 10 moss, 20 woody vegetation, 20 shadow, 20 surface water), and validated using an independent dataset of 90 observations. The training and testing datasets were generated using visual assessment of the high-resolution imagery in QGIS mapping software (QGIS,2024). The supervised random forest classifications were run using the randomForest package in R (Liaw and Wiener, 2002). The two resulting classifications were combined to minimize the number of pixels classified as shadow. Because the RedEdge-MX imagery was collected during the same field campaign as the chamber flux and in situ data collection (July 2021), land cover predictions from the RedEdge-MX classification were selected as the baseline. Pixels that were predicted as shadow by the RedEdge-MX classification, and predicted as non-shadow by the multiSPEC-4C classification, were assigned the multiSPEC-4C prediction in the combined classification.

CSV Files:

For 56 methane flux sample locations, in situ observations (Table_1_Chamber_Fluxes_Insitu_Data_July_2021.csv) were collected in July 2021 (3 replicates per location, totaling 168 samples). Mean CH₄ flux is the mean of three repetitions of chamber flux estimates. Modeled microtopography (m) and slope (degrees) were extracted from the gridded datasets for each flux sample location. Flux data were collected with a LI-COR Smart Chamber.

A Hydrosense II sensor with 20 cm and 12 cm probes was used to sample soil moisture (Campbell Scientific, Logan, UT, USA). To measure soil moisture at 20 cm and 12 cm depths, the 20 cm and 12 cm probes were inserted into the ground vertically. To measure soil moisture at 6 cm depth, the 12 cm probe was inserted into the ground at a 30° angle. To estimate soil moisture as volumetric water content (VWC), the Hydrosense II measurements of period (τ) were converted to percent VWC (Table_1_Chamber_Fluxes_Insitu_Data_July_2021.csv) using a calibration equation developed for organic boreal soils. The Soil Type A equation developed by Bourgeau-Chavez and Loboda (2018) was used for non-burned, organic soils. After calibration to percent VWC, the soil moisture values ranged from 29.0% to 88.2% (20 cm depth), 24.8% to 54.6% (12 cm depth), and 14.8% to 54.6% (6 cm depth).

Chamber methane (CH₄) and carbon dioxide (CO₂) flux estimates from field surveys conducted in July 2021 and August 2022 (Table_2_Detailed_Chamber_Flux_Estimates_July_2021_August_2022.csv) are included. For each flux estimate repetition, the original flux estimate, the estimated chamber headspace occupied by vegetation (percent volume), and the adjusted flux estimate, which accounts for the vegetation volume, are reported. The percent volume of chamber headspace filled by vegetation was estimated visually. CH₄ and CO₂ flux data were collected with either a LI-COR 8200-01S Smart Chamber (SC); Polycarbonate chamber with opaque cover (PC_cover); or Polycarbonate chamber without opaque cover (PC_no_cover). Each method used three replicates per sample. CO₂ flux estimates without the opaque cover represent net ecosystem exchange (NEE), whereas CO₂ flux estimates with the opaque cover represent ecosystem respiration (Reco). Gross primary productivity (GPP) was estimated as Reco - NEE. If the estimated GPP was a negative value, then the GPP was set to zero; and (c) CH₄ and CO₂ flux data collected with the LI-COR Smart Chamber at a subset of sample locations during the August 2022 field survey.

Relative abundance observations of methanogens and methanotrophs (Table_3_Methanogen_Methanotroph_Relative_Abundance.csv) were conducted at for four soil depths (0-12.5 cm, 12.5-25 cm, 25-37.5 cm, 37.5 - 50 cm) for subset of sample locations for which soil samples were collected (n = 17).

Soil chemical analyses (Table_4_Soil_Sample_Chemical_Analysis.csv) were conducted on soil samples from 17 locations at various depths. Measured variables included: 1.) general soil properties including pH, percent organic matter (OM), soluble salts (electrical conductivity, EC 1:1), sodium concentration (Na); 2.) macronutrient concentrations including nitrate-nitrogen, ammonium-nitrogen (NH4), phosphorous (P; Olsen method), potassium (K), calcium (Ca), magnesium (Mg), and sulfate-sulfur (S); and 3.) base saturation (K, Ca, Mg, Na).

SIMPER Microbial Community Analysis (Table_5_Simper_Microbial_Community_Analysis.csv) was used to compare microbial communities from the medium methane flux sites to the microbial communities from the high, low, and negative methane flux sites.

Meteorological data (Table_7_Hourly_Meteorology_Weather_Sensor.csv) were collected in July 2021 from a weather station that was located at 64.91985 latitude, -147.8189 longitude. Meteorological data were collected with a Campbell Scientific ClimaVue50 digital weather sensor, which collected measurements at 5 second resolution. Values in the data file are summarized in hourly increments.

At each sample location, soil temperature at 5 cm and 10 cm depth (Table_8_Halfhourly_Soil_Temperature_Logger_Data.csv), and soil pH at 2.5 cm and 6.4 cm depth (Table_1_Chamber_Fluxes_Insitu_Data_July_2021.csv), were collected using HI98331 and HI981030 GroLine soil probes (Hanna Instruments, Woonsocket, RI, USA). The soil pH probe was calibrated using pH 4.01 and 7.01 calibration buffer solutions.

Please see Farina et al. 2025 for additional details

Barnes, N., H. Webb, M.K. Farina, S. Powell, and J.D. Watts. 2021. Multispectral Imagery, NDVI, and Terrain Models, Big Trail Lake, Fairbanks, AK, 2019. ORNL DAAC, Oak Ridge, Tennessee, USA. https://doi.org/10.3334/ORNLDAAC/1834

Bourgeau-Chavez, L., and T. Loboda. 2018. Organic soil calibration algorithms for the Campbell Scientific handheld Hydrosense-I and II units.

Erdbrügger, J., I. van Meerveld, K. Bishop, and J. Seibert. 2021. Effect of DEM-smoothing and-aggregation on topographically-based flow directions and catchment boundaries. Journal of Hydrology 602:126717. https://doi.org/10.1016/j.jhydrol.2021.126717

Farina, M., W. Christian, N. Hasson, T. McDermott, S. Powell, R. Hatzenpichler, H. Webb, G. LaRue, K. Okano, E. Sproles, J. Watts. 2025. Methane flux spatial heterogeneity driven by surface and subsurface conditions in boreal forest-fen mosaic. [Manuscript submitted to Environmental Research Letters for publication.]

Liaw, Andy, and Matthew Wiener. 2002. Classification and regression by randomForest. R News 2:18-22. https://journal.r-project.org/articles/RN-2002-022/RN-2002-022.pdf

U.S. Geological Survey (USGS). 2017. 2017 USGS 3DEP Lidar: Fairbanks, AK (QL1 & QL2) from 2010-06-15 to 2010-08-15. NOAA National Centers for Environmental Information, https://www.fisheries.noaa.gov/inport/item/55358

QGIS. 2024. QGIS Geographic Information System. Open Source Geospatial Foundation Project. http://qgis.org