Data and Code for paper "A satellite-based ice fraction record for small water bodies of the Arctic Coastal Plain"

What is this? This file contains both the ice fraction dataset for small water bodies on the Arctic Coastal Plain and the associated production code. Data Introduction The ice fraction dataset records the fractional ice cover of small water bodies (900 m² to 25 km²) within each 1-km grid cell in the Arctic Coastal Plain of Alaska (ACP) from 2017 to 2023. The 1-km ice fraction was aggregated from 10-m ice cover maps. The 10-m maps were generated using random forest classification models based on Sentinel-1 SAR imagery, texture features, and temperature data. For more details, please refer to the paper. The final released dataset is provided in GeoTIFF format with a spatial resolution of 1 km, a temporal resolution of about 6 days, and is projected in the Alaska Albers Equal Area projection (EPSG: 3338). Each GeoTIFF image, named as YYYYMMDD.tif, represents the ice fraction of small water bodies in the ACP on a given day, observed by both ascending and descending Sentinel-1 passes, and recorded as the fraction of ice-covered area within small water bodies in each 1-km grid cell. Each image contains two bands: (1) ice fraction, with values ranging from 0 to 1, and (2) the proportion of small water bodies within each 1-km grid cell, also ranging from 0 to 1. The quality flag layer is also provided as a GeoTIFF at a spatial resolution of 1 km, with the band named “RRMSE” representing the data quality of each 1-km grid cell. Code Introduction This code is used to generate an ice fraction dataset for small water bodies of the Arctic Coastal Plain. It includes code for generating ice fraction in GEE and processing it locally using Python. These codes support the manuscript currently under review: "A satellite-based ice fraction record for small water bodies of the Arctic Coastal Plain". Code Execution Guide --------------Part 1: GEE Section--------------Step 0: Copy and paste the code from the txt file into the GEE console.Step 1: Collect training samples for the model. Run 1. Sampling (model training).Step 2: Perform hyperparameter tuning of the Random Forest classifier. Run 2. Hyperparameter tuning of the Random Forest classifier (ascending) and 2. Hyperparameter tuning of the Random Forest classifier (descending).Step 3: Export the 1 km ice fraction as CSV files to Google Drive. Run 3. Export ice fraction to Drive (ascending) and 3. Export ice fraction to Drive (descending). --------------Part 2: Python Section--------------Step 0: Save the ice fraction CSV files downloaded from GEE to data/S1_iceFraction_DESCENDING or data/S1_iceFraction_ASCENDING. Adjust the ascending or descending variables in the code as needed.Step 1: Merge ice fraction CSV files from the same day. Run 1. Merge CSV files from the same day.Step 2: Convert ice fraction CSV files to GeoPackage format. Run 2. Convert ice fraction CSV to GeoPackage.Step 3: Perform post-processing on the ice fraction data. Run 3. Ice fraction post-processing.Step 4: Convert ice fraction GeoPackage to GeoTIFF format. Run 4. Convert ice fraction GeoPackage to GeoTIFF.