Brief description
We describe a method to estimate δ15N from Landsat spectra. The development could inform the improvement of nitrogen cycle models and the assessments of impacts of past environmental changes.The predicted foliar nitrogen isotope ratio (δ15N) can provide valuable insights into terrestrial nitrogen cycles and their responses to environmental changes. Maps generated using this method can fill the gaps in existing observations of δ15N suffering from spatial bias and temporal discontinuity. The data could help explain the contradictory findings across different biomes, and help build capacity to detect and attribute drivers of change.
Lineage: Briefly, the Landsat spectra were linked to ground based δ15N data with a random forest model.
We used surface reflectance from all six spectral bands of Landsat 5 and 8 from 1984-2022. Landsat data in croplands, urban areas, permanent water, ice, or bare ground were removed.
Each ground-based δ15N was matched to a peak-growing season Landsat retrieval with highest green vegetation cover. A random forest model was then fitted to the ground-based δ15N and Landsat spectra. We used multiple validation methods to avoid overfitting and autocorrelation. The code is readily applicable to large scale mapping. The full description of the method and the evaluation can be found in Yang et al. (in review). Global maps derived using the framework are provided in Yang et al. (2024).
FILE DESCRIPTION
main.R has the code to conduct the fitting and prediction; r folder has the required functions; cache folder has the processed data. Note that Landsat spectra need to be downloaded from proper sources.
EXAMPLE
An example dataset is given with the example code in the main.R file. The dataset is a small subset of Craine et al. (2018).
REFERENCES
Yang et al. (in review) Mapping the multidecadal trends of terrestrial plant nitrogen stable isotope ratios globally.
Yang et al. (2024). Global terrestrial plant nitrogen stable isotope ratio samples: 30m centroid of each 0.1 degree grid cell from 1984 to 2022. https://data.csiro.au/collection/csiro:62622
Available: 2024-06-20
Data time period: 2024-05-31 to 2024-05-31
Subjects
Biological Sciences |
CO2 |
Earth Sciences |
Earth System Sciences |
Ecological Physiology |
Ecology |
Geochemistry |
Geochemistry Not Elsewhere Classified |
Nitrogen cycles |
Other Earth Sciences |
Plant Biochemistry |
Plant Biology |
climate change |
nitrogen stable isotope |
vegetation change |
User Contributed Tags
Login to tag this record with meaningful keywords to make it easier to discover
