Fractional Abundance Datasets for Salt Patch and Marshland Across the Delmarva Peninsula, v2

Abstract:

Coastal agricultural lands in the eastern USA are increasingly plagued by escalating soil salinity, rendering them unsuitable for profitable farming. Increasing soil salinization can lead to both land cover modification and conversion. Two notable instances of such transformations include the conversion of farmland to marshland or to barren salt patches devoid of vegetation. However, quantifying these land cover changes across large geographic regions poses a significant challenge due to their varying spatial granularity. To tackle this issue, we first developed a machine-learning based method using Sentinel-2 imagery for 2022 where we used a non-linear spectral unmixing approach utilizing a Random Forest (RF) algorithm (Sarupria et al., 2025). The RF models were constructed using 100 trees and 27,437 reference data points, resulting in two sets of ten models: one for salt patches and another for marshland. Validation metrics for sub-pixel fractional abundances revealed a moderate R-squared value of 0.50 for the salt model ensemble and a high R-squared value of 0.90 for the marsh model ensemble. Building upon this methodology, we then generated annual gridded datasets of fractional abundance for salt patch and marshland across the Delmarva Peninsula (14 coastal counties in Delaware, Maryland and Virginia, USA) for 2019-2023. In these datasets, we only report mean fractional abundance values ranging from 0.4 to 1 for salt patches and 0.25 to 1 for marshland, along with the standard deviation associated with each value.

Description: 

1. Salt Patch Fractional Abundance Mean (2019–2023): Five files, each representing the per-pixel fractional abundance mean of salt patches for the years 2019–2023. Values are derived from an ensemble of 10 Random Forest (RF) models. Only pixels with a salt patch fraction ≥ 0.40 were retained.

2. Salt Patch Fractional Abundance Standard Deviation (2019–2023): Five files representing the per-pixel standard deviation of the salt patch fractional abundance means for the years 2019–2023. Estimates are based on an ensemble of 10 RF models and include only pixels with a salt patch fraction ≥ 0.40.

3. Marsh Fractional Abundance Mean (2019–2023): Five files, each showing the per-pixel fractional abundance mean of marshland for the years 2019–2023, derived from an ensemble of 10 RF models. Only pixels with a marsh fraction ≥ 0.25 were retained.

4. Marsh Fractional Abundance Standard Deviation (2019–2023): Five files showing the per-pixel standard deviation of the marsh fractional abundance means for 2019–2023. Estimates are based on an ensemble of 10 RF models, limited to pixels with a marsh fraction ≥ 0.25.

Input Data:

This approach integrated Sentinel-2 surface reflectance imagery, a global land use/land cover dataset from ESRI (Karra et al., 2021), a NAIP-derived Delmarva land cover dataset (Mondal et al., 2022), high-resolution PlanetScope true color images (Planet Team, 2017), very high-resolution Unoccupied Aerial Vehicle (UAV) imagery, and ground truth data.

More details on input data can be found in Sarupria et al., 2025a and version 1 of this dataset (Sarupria et al., 2025b).

Method:

We utilized Sentinel-2 Level 2 A surface reflectance imagery for spectral unmixing. This multispectral dataset, corrected for atmospheric and radiometric effects, encompasses 13 spectral bands from visible to near-infrared wavelengths (0.443–2.190 micrometers). The imagery offers spatial resolutions ranging from 10 m to 60 m and is captured every 5 days. To aid in selecting reference points for model training and testing, high-resolution (60 cm) UAV images of specific farmlands in Dorchester and Somerset counties, Maryland, were acquired under optimal weather conditions. 

The study incorporated multiple datasets to refine the analysis. The Sentinel-2 derived global land use/land cover dataset from ESRI was employed to isolate relevant land cover classes such as 'Crops' and 'Rangeland'. A NAIP-derived Delmarva land cover dataset with eight classes helped exclude non-agricultural land cover types. High-resolution PlanetScope true color images with 3 m spatial resolution were used as reference data for model validation.

In this study, we applied a Random Forest (RF) classifier for nonlinear spectral unmixing. The RF classifier functions by utilizing an ensemble of decision trees that are independently trained on random subsets of training data through bootstrap aggregation. The final classification is determined by aggregating votes from all trees, with the endmember receiving the highest total votes being selected as the final output. To access soft voting information from the RF classifier in python, we used its probability prediction function called ‘predict_proba’. This function enables each decision tree to produce a probability distribution for each endmember instead of making a single class decision. The probability distribution from a decision tree indicates how likely it is that an input pixel belongs to each endmember. The final predicted probabilities are calculated by averaging these distributions across all decision trees for each of the five endmembers. As a result, each pixel in the final output is represented by five probability values that indicate the fractional abundance of each corresponding endmember within that pixel. These probabilities sum to one, effectively illustrating the spectral unmixing of a mixed pixel. For a specific endmember, a pixel with fractional abundance value of 0 signifies the absence of it, while a value of 1 indicates a pure pixel. Values between 0 and 1 reflect varying levels of mixed endmembers.

More details on the methods and accuracy assessment can be found in Sarupria et al., 2025a.

Data format:

The spatial resolution of all the derived datasets is 10 m. These georeferenced datasets are distributed in GEOTIFF format and are compatible with GIS and/or image processing software, such as R and ArcGIS Pro. The GIS-ready raster files can be used directly in mapping and geospatial analysis.

Code:

Sample python code for performing spectral unmixing is available at: https://github.com/Manan-prog/Non-linear-Spectral-Unmixing. To run this code successfully, the user must provide training data for the desired land cover classes and an input raster image for spectral unmixing.

Datasets for download: 

1. Five layers for salt patch mean values for 2019-2023: SaltPatch_FrAb_Mean_<Year>
2. Five layers for salt patch standard deviation values for 2019-2023: SaltPatch_FrAb_StdDev_<Year>
3. Five layers for marsh mean values for 2019-2023: Marsh_FrAb_Mean_<Year>
4. Five layers for marsh standard deviation values for 2019-2023: Marsh_FrAb_StdDev_<Year>