QTP Cloud-Free Snow Cover Validation Dataset for MODIS Snow Cover Product Evaluation

Data description The QTP cloud-free snow cover validation dataset provides high-resolution reference snow maps and corresponding cloud-covered validation samples for evaluating the performance of five MODIS-based daily cloud-free snow cover extent (SCE) products over the Qinghai–Tibet Plateau (QTP). The evaluated products include Dysno_Cloudfree, HMRF-MODIS, I-MODIS, F-MODIS, and NIEER-MODIS. The dataset contains high-resolution snow cover reference data derived from GF-1, Landsat-8, and Sentinel-2 imagery, together with MODIS cloud masks from Collections 5, 6, and 6.1. These cloud masks were used to identify MODIS cloud-contaminated areas, where the corresponding snow-cover information from high-resolution reference data and cloud-free MODIS SCE products was extracted for validation. The dataset supports both pixel-level accuracy assessment and snow-cover morphology evaluation, including fractal dimension (FD) and structural similarity index (SSIM) analysis. High-resolution snow-cover validation samples: 1.High_resolution_validation_samples Folder: GFLandsat8ST_2 Filename: named by acquisition date, sample number, reference sensor, MODIS Collection version, or snow-cover product Examples: 20131024(1)GFC520131024(1)GFC620131024(1)GFC6120131024(1)fsc20131024(1)hmrf20131024(1)ims20131024(1)v6120130524L820130524NIEER Format: tif Variable: High-resolution reference snow cover MODIS cloud masks from Collections 5, 6, and 6.1 High-resolution reference snow cover extracted within MODIS cloud-covered regions Snow cover results from F-MODIS Snow cover results from HMRF-MODIS Snow cover results from I-MODIS Snow cover results from Dysno_Cloudfree Snow cover results from NIEER-MODIS ps: (1) GF, L8, and ST2 indicate GF-1, Landsat-8, and Sentinel-2 reference data, respectively. (2) The first eight digits in the filename indicate the acquisition date in the format YYYYMMDD. (3) Numbers in parentheses, such as (1) and (2), distinguish multiple high-resolution samples acquired on the same date. (4) C5, C6, and C61 indicate MODIS Collections 5, 6, and 6.1, respectively. (5) Files containing cl, such as MODC5cl or MYDC6cl, represent cloud masks extracted from the corresponding MODIS Collection. (6) Files such as GFC5, GFC6, and GFC61 represent GF reference snow cover extracted within the cloud-covered regions identified by MODIS Collections 5, 6, and 6.1. Equivalent naming rules are used for Landsat-8 and Sentinel-2 samples. (7) fsc represents F-MODIS. The original fractional snow cover data were converted into binary snow cover extent data using a 50% fractional snow cover threshold. (8) hmrf represents HMRF-MODIS. (9) ims represents I-MODIS. (10) v61 represents Dysno_Cloudfree. (11) NIEER represents NIEER-MODIS. Data evaluation algorithms The provided pseudocodes include: Confusion Matrix Based Accuracy AssessmentThis pseudocode describes the calculation of pixel-level classification metrics, including overall accuracy (OA), precision, recall, F1-score, and bias, based on comparisons between reference snow maps and evaluated snow cover products. Structural Similarity Index (SSIM) CalculationThis pseudocode describes the procedure for evaluating the spatial structural consistency between reference snow cover maps and cloud-free snow cover products using the SSIM metric. Fractal Dimension (FD) CalculationThis pseudocode describes the box-counting method used to calculate the fractal dimension of snow cover maps, which characterizes the spatial complexity and morphological features of snow distribution. These pseudocodes provide a general representation of the processing workflow and evaluation methodology used in this dataset.