Imaging spectroscopy reveals topographic variability effects on grassland functional traits and drought responses
Data files
Jan 03, 2025 version files 6.67 GB
-
Data_and_Codes.zip
6.67 GB
-
README.md
12.40 KB
Abstract
Functional traits and their variations are essential indicators of plant metabolism, growth, distribution, and survival and determine how a plant and an ecosystem function. Under the same climatic condition, traits can vary significantly between species and within the same species growing in different topographic conditions. When drought stress occurs, plants growing in these conditions may respond in various ways as their tolerance and adaptability are influenced by differences in topography. Insights into topographic variability-driven trait variation and drought response can improve our prediction of ecosystem functioning and ecological impacts. Imaging spectroscopy enables accurate identification of plant species, extraction of functional traits, and characterization of topography-driven and drought-related impacts on trait variation across spatial scales. However, applying this data in a heterogeneous grassland ecosystem is challenging as species are small, high mixed, spectrally and texturally similar, and highly varied with small-scale variation in topography. This paper presents the first study to explore the use of high-resolution airborne imaging spectroscopy for characterizing the variation of key traits—such as chlorophylls (Chl), carotenoids (Car), Chl/Car ratio, water content (WC), and leaf area index (LAI)—across topographic gradients and under drought stress at the species level in a heterogeneous grassland. The results demonstrate significant relationships between functional traits and topographic variability, with the strength of these relationships varying among species and across different environmental conditions. Additionally, drought-induced trait responses differed notably both within and between species, particularly between drought-tolerant invasive species and native species, as well as between lower and upper slope positions. The study makes a significant contribution to advancing our understanding of biological and ecological processes, enhancing the ability to predict plant invasion mechanism and ecosystem functioning under stressed environments.
README: Data and Codes: Imaging spectroscopy reveals topographic variability effects on grassland functional traits and drought responses
Data
Hyperspectral images (hyperspectral_images):
Hyperspectral images used in this study include 20160823_HSI.tif for the year of 2016 and 20170820_HSI_INT_N.tif for the year of 2017. Both datasets have 301 bands ranging from 400 nm to 1000 nm with a spectral interval of 2 nm. The reflectance values of pixels in the 20170820_HSI_INT_N.tif dataset are scaled up by multiplying the original reflectance by 10000 and stored in the integer data format. Hyperspectral images were acquired on 23 August 2016 and 20 August 2017 by a Mirco-Hyperspec VNIR push-broom sensor (Headwall Photonics Inc, USA), operated on a JetRanger helicopter.
Species classification training data (species_classification_training_data):
This dataset contains the training and test datasets in shapefile format for species classification model training and validation. The training dataset (80%) is stored in the train_split_Aug folder while the test dataset (20%) is stored in the test_split_Aug folder. In these folders, data of each species is stored in a separate shapefile.
Soil and shadow mask (soil_shadow_mask):
20170820_soil_shadow_mask.tif: a mask file for masking out soil and shadow background in the species classification results.
Spectral-trait data to build trait models (traits_spectra):
20160823_PlantProp_full_leaf.csv: leaf-level trait and spectral data collected in 2016
- Site: The name of the field plot
- Splname: The name of the leaf sample
- Species: The name of the species
- Greeness: The leaf greenness levels
- Chl: The total chlorophyll content of the leaf sample (µg/cm^2)
- Chl_ab: The chlorophyll-a to chlorophyll-b ratio of the leaf sample
- Car: The total carotenoid content of the leaf sample (µg/cm^2)
- Chl_car: The chlorophyll to carotenoid ratio of the leaf sample
- WC: The water content of the leaf sample (g/cm^2)
- SLA: The specific leaf area of the leaf sample (cm^2/g)
- B_0350 to B_2500: The spectral reflectance values from band 350 nm to band 2,500 nm
20160823_PlantProp_full_image.csv: image-level trait and spectral data collected in 2016
- Site: The name of the field plots
- Species: The name of the species
- Chl: The total chlorophyll content of the field plot (µg/cm^2)
- Chl_ab: The chlorophyll-a to chlorophyll-b ratio of the field plot
- Car: The total carotenoid content of the field plot (µg/cm^2)
- Chl_car: The chlorophyll to carotenoid ratio of the field plot
- WC: The water content of the field plot (g/cm^2)
- LAI: The Leaf Area Index of the field plot
- B_0400 to B_1000: The spectral reflectance values from band 400 nm to band 1,000 nm
20170820_PlantProp_full_leaf.csv: leaf-level trait and spectral data collected in 2017
- Site: The name of the field plot
- Splname: The name of the leaf sample
- Species: The name of the species
- Greeness: The leaf greenness levels
- Chl: The total chlorophyll content of the leaf sample (µg/cm^2)
- Chl_ab: The chlorophyll-a to chlorophyll-b ratio of the leaf sample
- Car: The total carotenoid content of the leaf sample (µg/cm^2)
- Chl_car: The chlorophyll to carotenoid ratio of the leaf sample
- WC: The water content of the leaf sample (g/cm^2)
- SLA: The specific leaf area of the leaf sample (cm^2/g)
- B_0350 to B_2500: The spectral reflectance values from band 350 nm to band 2,500 nm
20170820_PlantProp_full_image.csv: image-level trait and spectral data collected in 2017
- Site: The name of the field plots
- Species: The name of the species
- Chl: The total chlorophyll content of the field plot (µg/cm^2)
- Chl_ab: The chlorophyll-a to chlorophyll-b ratio of the field plot
- Car: The total carotenoid content of the field plot (µg/cm^2)
- Chl_car: The chlorophyll to carotenoid ratio of the field plot
- WC: The water content of the field plot (g/cm^2)
- LAI: The Leaf Area Index of the field plot
- B_0400 to B_1000: The spectral reflectance values from band 400 nm to band 1,000 nm
Leaf and field reflectance measurements were taken using a FieldSpec 3 spectroradiometer from Malvern PANalytical Company, United Kingdom. The sensor measured reflectance from 400 nm to 2500 nm at an interval of 1 nm.
Topographic position index data (topographic_position_index):
tpi_2m_rescale_clip.tif
This is a raster dataset of the topographic position index of the study area.
Ground sampling sites (ground_sampling_sites):
These datasets contain the locational data of the field sampling sites of both 2016 (GroundSites_2016.shp) and 2017 (GroundSites_2017.shp). These datasets allow us to correlate field and laboratory data with airborne hyperspectral images.
Tabular data (tabular_data):
plantprop_canopy_TPI_moisture.csv
This file contains soil moisture data measured at 42 field sites in 2016 and 2017 and its corresponding topographic position index (TPI) values. This dataset is used to explore the correlation between soil moisture and TPI in both drought year (2016) and normal year (2017).
- Site: The name of the field plot
- Species: The name of the species
- Year: The year of collection
- Soil_moisture: The soil moisture value of the field plot (%)
- TPI: The Topographic Position Index value of the field plot
plantprop_canopy_box.csv
This file contains the values of topographic position index (TPI) and five canopy-level traits, including chlorophyll content (Chl), chlorophylls to carotenoids ratio (Chl/Car), carotenoid content (Car), water content (WC) and leaf area index (LAI), at sampling field sites of the year of 2016, 2017, and 2016-2017 combined. This dataset is used to examine the correlation between TPI and these traits at the canopy level.
- Site: The name of the field plot
- Species: The name of the species
- TPI: The Topographic Position Index value of the field plot
- Year: The year of collection
- Chl: The total chlorophyll content of the field plot (µg/cm^2)
- Car: The total carotenoid content of the field plot (µg/cm^2)
- Chl_Car: The chlorophyll to carotenoid ratio of the field plot
- WC: The water content of the field plot (g/cm^2)
- LAI: The Leaf Area Index value of the field plot
plantprop_2016_2017_N_box.csv
This file contains the values of topographic position index (TPI), the six classified TPI categories (ridge (RD), upper slope (US), midslope (MS), toe slope (TS), flat (FL), and valey (VL)), and five canopy-level traits, including chlorophyll content (Chl), chlorophylls to carotenoids ratio (Chl/Car), carotenoid content (Car), water content (WC) and leaf area index (LAI), extracted from trait maps of the years of 2016 and 2017. This dataset is used to examine the trait variation both within and between species in both drought year (2016) and normal year (2017) across topographic variability categories.
- Species: The name of the species
- TPI_Cat: The topographic category of pixels extracted from the TPI category maps
- TPI: The Topographic Position Index of pixels extracted from the TPI maps
- Year: The year of collection
- Chl: The total chlorophyll content of pixels extracted from the Chl trait maps (µg/cm^2)
- Chl_Car: The chlorophyll to carotenoid ratio of pixels extracted from the Chl_Car trait maps
- Car: The total carotenoid content of pixels extracted from the Car trait maps (µg/cm^2)
- WC: The water content of pixels extracted from the WC trait maps (g/cm^2)
- LAI: The Leaf Area Index value of pixels extracted from the LAI trait maps
plantprop_2016_2017_N_dif.csv
This file contains the values of topographic position index (TPI), the six classified TPI categories (ridge (RD), upper slope (US), midslope (MS), toe slope (TS), flat (FL), and valey (VL)), and the differences in five canopy-level traits, including chlorophyll content (Chl), chlorophylls to carotenoids ratio (Chl/Car), carotenoid content (Car), water content (WC) and leaf area index (LAI) between 2016 and 2017, extracted from trait maps. This dataset is used to examine the trait variation both within and between species under drought stress across topographic variability categories.
- Species: The name of the species
- TPI_Cat: The topographic category of pixels extracted from the TPI category maps
- TPI: The Topographic Position Index of pixels extracted from the TPI maps
- Chl_diff: The difference in chlorophyll content (µg/cm^2) between 2016 and 2017 of each pixel
- Chl_Car_diff: The difference in chlorophyll/carotenoid ratio between 2016 and 2017 of each pixel
- Car_diff: The difference in carotenoid content (µg/cm^2) between 2016 and 2017 of each pixel
- WC_diff: The difference in water content (g/cm^2) between 2016 and 2017 of each pixel
- LAI_diff: The difference in Leaf Area Index between 2016 and 2017 of each pixel
Codes
Species_classification_RandomForest.ipynb
The Python notebook for species classification with the Random Forest classifier.
PLSR_leaf_model.R
The R code for building leaf-level trait model using the Partial Least Squares Regression (PLSR) method.
PLSR_image_model.R
The R code for building image-level trait model and deriving trait maps from hyperspectral images using the Partial Least Squares Regression (PLSR) method.
Statistical_analysis.R
The R code for all other statistical analyses of the study.
Usage notes
Handling shapefiles (.shp)
All the shapefiles (.shp) used in this study contain the geometry and attributes of geospatial features (e.g., points, lines, polylines, polygons). The file bundle contains the main file .shp and companion files including: .cpg, .dbf, .prj, .sbn, .sbx, .shx. Description of these file extensions is given as follows:
.shp: The main geospatial data file that contains feature geometry.
.cpg: The file specifying the codepage to identify the characterset.
.dbf: The dBASE that contains the attributes of features.
.prj: The file that contains the coordinate system and map projection information.
.sbn: The file containing the spatial index of features.
.sbx: The file containing the spatial index of features.
.shx: The file containing the index of feature geometry.
More details about these file extensions and the information they contain can be found at: https://desktop.arcgis.com/en/arcmap/latest/manage-data/shapefiles/shapefile-file-extensions.htm#:~:text=xml%E2%80%94Metadata%20for%20ArcGIS%E2%80%94stores,the%20characterset%20to%20be%20used.
The main .shp file can be opened and analyzed by Python, R, and many other programming languages, and open-source geospatial software such as QGIS, SAGA GIS, GRASS GIS, GeoDa, etc.
Handling GeoTIFF files (.tif)
All the shapefiles (.tif) used in this study to store raster data and images and its associated geospatial information. The file bundle contains the main file .tif and companion files including (not all files need all these companion files): .hdr, .aux, .cpg, .dbf, .enp, .ovr, .tfw, .xml. Description of these file extensions is given as follows:
.tif: Stores image information and raster graphics.
.hdr: The header file that stores information of an image or a raster, including: cell size, data type, compression technique, blocking factor, and tile.
.aux: Stores additional image or raster information that cannot be stored in the TIFF file.
.cpg: The file specifying the codepage to identify the characterset.
.dbf: The dBASE that contains the attributes of an image or a raster file.
.enp: Stores a pyramid file for display and visualization in ENVI software.
.ovr: Stores a pyramid file for display and visualization.
.tfw: Contains georeferencing information for a raster.
.xml: Metadata of an image or a raster file.
The main .tif file can be opened and analyzed by Python, R, and many other programming languages, and open-source geospatial software such as QGIS, SAGA GIS, GRASS GIS, GeoDa, etc, or proprietary software such as ArcGIS Pro, ENVI, ERDAS Imagine, etc.