A global daily seamless 9-km Vegetation Optical Depth (VOD) product from 2010 to 2021

(I) DESCRIPTION:

· A global daily seamless 9-km Vegetation Optical Depth (VOD) product is generated through gap-filling and spatiotemporal fusion model. This daily products start from Jan 01, 2010 to Jul 31, 2021 (about 20GB memory after uncompressing all zip files).

· To further validate the effectiveness of these products, three verification ways are employed as follow: 1) Time series validation; 2) Simulated missing-region validation; And 3) Data comparison validation.

· It is important to note that the original data contain missing dates, and these corresponding gaps are also present in our dataset.

(II) DATA FORMATTING AND FILE NAMES 

For the convenience of our readers, we have two formats of data available for download.

1) MAT file (Version v1)

Data from 2010 to 2021 are stored separately into folders for the corresponding years, with each folder containing daily `.mat` files. The naming convention for the data is “YYYYXXZZ,” where YYYY is the 4-digit year, XX is the 2-digit month, and ZZ is the 2-digit date. The geographic scope is global and the grid size is 4000*2000.

MATFILES (.mat): The folders with matfiles contain individual files for:

1. Vegetation Optical Depth: VOD_seamless_9km_ YYYYXXZZ.mat

2. Latitude/Longitude: VOD_9km_Coordinates.mat

2) NetCDF file (Version v2)

The year-by-year daily data from 2010 to 2021 are stored in the ‘.nc’ files for the corresponding years. The daily data within each year into one NetCDF file. The variable names are named as VOD_xxxxyydd, where xxxx represents the year, yy represents the month, and dd represents the day. The longitude variable is named “lon” with a dimension of 4000×1, and the latitude variable is named “lat” with a dimension of 2000×1.

It should be noted that these NetCDF files are saved using the netCDF4 library in Python, with the dimension order being (lat, lon). When reading these NetCDF files in MATLAB, the default data dimension order is (lon, lat). Therefore, it is necessary to transpose the variables to match the correct dimension order.