Data from: Bringing cross-validation into the real world to evaluate transferability of satellite-based vegetation models
This tabular dataset supports findings reported in "Bringing cross-validation into the real world to evaluate transferability of satellite-based vegetation models" published in Scientific Reports. Data are from the Central Plains Experimental Range (CPER), near Nunn, Colorado, and contains nearly 10,000 ground samples collected from 2014 to 2023. Data are aggregated to the plot level (n=2,322), of which there are 4-6 in each pasture and also provided at the transect level (n=9,647), of which there are 4 in each plot. Data were used to train progressively more complex Machine learning algorithms (MLAs) to predict standing aboveground herbaceous biomass in grazinglands, such as those found at CPER. These data support an investigation of whether (1) do simpler geospatial MLAs have better transferability to ‘unseen’ conditions and ‘worst-case’ performance (i.e., under extreme conditions) compared to more complex MLAs and (2) are they more consistent when re-trained with new data. Data include (1) unique identifiers and blocking co-variates (2) in-situ ground-sampled herbaceous standing biomass estimated using visual obstruction (VO) methods, along with the VO readings themselves (3) geographic polygon bounding coordinates of each plot/transect (WKT format), and (4) satellite-derived indices and bands from the Harmonized Landsat Sentinel (HLS) data product. All in situ values and satellite derived indices are means for the respective plot and transect areas.This research used resources provided by the SCINet project and/or the AI Center of Excellence of the USDA Agricultural Research Service, ARS project numbers 0201-88888-003-000D and 0201-88888-002-000D.
Resources
5 resources available
-
Geodata ISO 19139 metadata
XML -
DataDictionary_kearney_scirep_plot_fnl.csv
CSV -
DataDictionary_kearney_scirep_transect_fnl.csv
CSV -
kearney_scirep_plot_fnl.csv
CSV -
kearney_scirep_transect_fnl.csv
CSV
Find Related Datasets
Search by Tags
Click any tag below to search for similar datasets
Complete Metadata
| @type | dcat:Dataset |
|---|---|
| accessLevel | public |
| accrualPeriodicity | R/P6M |
| bureauCode |
[
"005:18"
]
|
| contactPoint |
{
"fn": "Porensky, Lauren",
"hasEmail": "mailto:Lauren.Porensky@usda.gov"
}
|
| description | <p dir="ltr">This tabular dataset supports findings reported in "Bringing cross-validation into the real world to evaluate transferability of satellite-based vegetation models" published in <i>Scientific Reports</i>. Data are from the Central Plains Experimental Range (CPER), near Nunn, Colorado, and contains nearly 10,000 ground samples collected from 2014 to 2023. Data are aggregated to the plot level (n=2,322), of which there are 4-6 in each pasture and also provided at the transect level (n=9,647), of which there are 4 in each plot. Data were used to train progressively more complex Machine learning algorithms (MLAs) to predict standing aboveground herbaceous biomass in grazinglands, such as those found at CPER. These data support an investigation of whether (1) do simpler geospatial MLAs have better transferability to ‘unseen’ conditions and ‘worst-case’ performance (i.e., under extreme conditions) compared to more complex MLAs and (2) are they more consistent when re-trained with new data. Data include (1) unique identifiers and blocking co-variates (2) in-situ ground-sampled herbaceous standing biomass estimated using visual obstruction (VO) methods, along with the VO readings themselves (3) geographic polygon bounding coordinates of each plot/transect (WKT format), and (4) satellite-derived indices and bands from the Harmonized Landsat Sentinel (HLS) data product. All in situ values and satellite derived indices are means for the respective plot and transect areas.</p><p dir="ltr">This research used resources provided by the SCINet project and/or the AI Center of Excellence of the USDA Agricultural Research Service, ARS project numbers 0201-88888-003-000D and 0201-88888-002-000D.</p> |
| distribution |
[
{
"@type": "dcat:Distribution",
"title": "Geodata ISO 19139 metadata",
"format": "xml",
"accessURL": "https://geodata.nal.usda.gov/geonetwork/srv/api/records/de93dba9-4e90-49cf-95ec-b1ee4ccfed83/formatters/xml",
"mediaType": "text/xml",
"conformsTo": "https://www.isotc211.org/2005/gmd"
},
{
"@type": "dcat:Distribution",
"title": "DataDictionary_kearney_scirep_plot_fnl.csv",
"format": "csv",
"mediaType": "text/csv",
"downloadURL": "https://ndownloader.figshare.com/files/61674142"
},
{
"@type": "dcat:Distribution",
"title": "DataDictionary_kearney_scirep_transect_fnl.csv",
"format": "csv",
"mediaType": "text/csv",
"downloadURL": "https://ndownloader.figshare.com/files/61674145"
},
{
"@type": "dcat:Distribution",
"title": "kearney_scirep_plot_fnl.csv",
"format": "csv",
"mediaType": "text/csv",
"downloadURL": "https://ndownloader.figshare.com/files/61674148"
},
{
"@type": "dcat:Distribution",
"title": "kearney_scirep_transect_fnl.csv",
"format": "csv",
"mediaType": "text/csv",
"downloadURL": "https://ndownloader.figshare.com/files/61674151"
}
]
|
| identifier | 10.15482/USDA.ADC/31271986.v1 |
| keyword |
[
"RANGELANDS",
"biomass",
"grasslands",
"machine learning",
"satellite remote sensing;"
]
|
| license | https://creativecommons.org/publicdomain/zero/1.0/ |
| modified | 2026-02-20 |
| programCode |
[
"005:040"
]
|
| publisher |
{
"name": "Agricultural Research Service",
"@type": "org:Organization"
}
|
| spatial |
"{"type": "Polygon", "coordinates": [[[-104.78842163924, 40.800296198063], [-104.79006960522, 40.869287735048], [-104.68343352433, 40.869703138693], [-104.68240359798, 40.801647612446], [-104.78842163924, 40.800296198063]]]}"
|
| temporal | 2014-06-01/2023-10-31 |
| theme |
[
"geospatial"
]
|
| title | Data from: Bringing cross-validation into the real world to evaluate transferability of satellite-based vegetation models |
