Google Earth Engine (GEE) - Highly Scalable Time-Adaptive Reflectivity Fusion Model (HISTARFM) Database

Highly Scalable Time-Adaptive Reflectance Fusion Model (HISTARFM) Database
The HISTARFM database is a high spatial resolution monthly albedo time series corrected for cloud data gaps. The dataset was created at 30-meter resolution by fusing Landsat and Moderate Resolution Imaging Spectroradiometer (MODIS) time series. The method consists of using two estimators that work together to remove random noise and minimize the bias in Landsat spectral reflectance. The first estimator is an optimal interpolator that generates Landsat reflectance estimates using Landsat historical data and fused MODIS and Landsat reflectance from the nearest corridor. The fusion process employs a pixel-level linear regression model. The second estimator is a Kalman filter that corrects for any bias in the reflectivity produced by the first estimator. Foreword – HISTARFM provides improved albedo values ​​and a unique and useful by-product, albedo uncertainty, which is helpful for realistic error calculations (e.g., errors in calculating vegetation indices or biophysical variables strip). For a more detailed explanation of the HISTARFM algorithm, please refer to the 2020 manuscript by Moreno-Martinez et al.

https://www.sciencedirect.com/science/article/pii/S0034425720302716

 

 

Citation¶

Moreno-Martínez, Álvaro, Emma Izquierdo-Verdiguier, Marco P. Maneta, Gustau Camps-Valls, Nathaniel Robinson, Jordi Muñoz-Marí, Fernando Sedano,
Nicholas Clinton, and Steven W. Running. "Multispectral high resolution sensor fusion for smoothing and gap-filling in the cloud." Remote Sensing of
Environment 247 (2020): 111901.

Earth Engine Snippet¶

Different editions and fields of study are already processes:

The US CONUS database contains 154 images stored as assets. It corresponds to the second edition and covers the time period from January 2009 to October 2021. Each image in the ImageCollection covers the entire continental United States, and each image has "version", "month" and "year" properties. This information is also present in their filenames. For example, an image named Gap_Filled_Landsat_CONUS_month_10_2009_v2 is an image of the CONUS region in October 2009. The CONUS database is available in this asset and the imagery loads in Earth Engine as follows:

var histarfm_conus = ee.ImageCollection("projects/KalmanGFwork/GFLandsat_V1")

 Databases for Europe, major regions of East Asia, and Somalia are currently being generated using version 5 of the algorithm. Version 5 contains 26916 images. Europe covers 9 years from 2013 to 2021, East Asia covers 3 years from 2019 to 2021, and Somalia covers 5 years from 2010 to 2014. All study areas are divided into tiles and stored as cloud-optimized Geotiffs on Google Cloud Platform. The name of the image includes month, year, specific study area, and tile. As an example, an image named GF_2018_10_EUROPA_1 represents the image on the first tile in Europe in October 2018. Version 5 of the database is available here, and images can be loaded in Earth Engine with the following code:

var histarfm_ic = ee.ImageCollection("projects/ee-kalman-gap-filled/assets/histarfm_v5")

Sample code : https://code.earthengine.google.com/?scriptPath=users/sat-io/awesome-gee-catalog-examples:analysis-ready-data/HISTARFM-V5-EXAMPLE 

For more information on how to work with HISTARFM, and some examples of how you can use the HISTARFM database to improve your research and applications, visit the tutorial here.

The HISTARFM database was used in the following papers

• Martínez-Ferrer, L., et al. "Quantifying uncertainty in high resolution biophysical variable retrieval with machine learning." Remote Sensing of Environment 280 (2022): 113199.

• Salerno, L., et al. "Satellite Analyses Unravel the Multi-Decadal Impact of Dam Management on Tropical Floodplain Vegetation." Frontiers in Environmental Science (2022): 357.

• Kushal, K. C., and Sami Khanal. "Agricultural productivity and water quality tradeoffs of winter cover crops at a landscape scale through the lens of remote sensing." Journal of Environmental Management 330 (2023): 117212.

License¶

The dataset is licensed under a Creative Commons Attribution NonCommercial 4.0 International license.

Curated by: Álvaro Moreno-Martínez, Emma Izquierdo-Verdiguier, Jordi Muñoz-Marí and Nicolas Clinton.

Keywords: MODIS, Landsat, Land reflectance images, gap-filled temporal series, vegetation

Last updated: 06-03-2023