Surface water masks over the Mekong Delta and Cambodia derived from visible/infrared MODIS observations



We introduce weekly surface water masks (1 mask every 8 days) over the Mekong Delta and Cambodia, derived from visible/infrared MODIS satellite observations. Surface water masks are built mainly based on a methodology developed by Sakamoto et al. [2007]1. This method uses the MODIS Terra surface reflectance 8-day Level-3 (L3) Global 500 m (MOD09A1) as the input. The method is applied to the Mekong basin located lower than latitude 15°N. More than 16-year MODIS data are processed to produce surface water masks for the studied area, from 2001 to current time, at 500 m spatial resolution. Figure 1 shows example of monthly surface water masks over the Mekong Delta and Cambodia, for 2011.
USTH News 04 800x419
Figure 1. Monthly surface water masks over the Vietnam Mekong Delta and Cambodia for 2011, derived from MODIS observations based on the methodology developed by Sakamoto et al. [2007]1

16-year surface water time series for the area in Figure 1 are shown in Figure 2, and the map of inundation frequency derived from 16-year data is shown in Figure 3
Figure 2. Time series of surface water extent over the Mekong Delta and Cambodia for the period 2001-2016.
Figure 3. Inundation frequency over the Mekong Delta and Cambodia, based on 16-year MODIS observations. 


The classification algorithm is presented in Figure 4. Details can be found in the paper1.
Figure 4. Block diagram of the classification algorithm (Figure is reprinted from Sakamoto et al. [2007]1).


MODIS-derived surface water masks over the Mekong Delta and Cambodia have been compared to:
  • Surface water maps derived from SAR Sentinel-1 observations
  • Topography-based floodability index
  • Water height over the Tonle Sap Lake, derived from altimetry data of the ENVISAT and JASON-2 satellites
  • Local precipitation data 
These comparisons showed good agreements with the MODIS-derived surface water estimates.


This product can be used to monitor and quantify surface water within the Mekong Delta and Cambodia. It can also be an input for regional hydrological models 

Data download

Data files are stored in the Google Driver. Data Folder:

A data file contains 4 states: 0 for non-water pixels, 1 for mixed pixels, 2 for water pixels, and -1 for non-analyzed pixels (including pixels located over ocean).
For mixed pixels, only part of the pixels is inundated. In Figure 2, it is assumed that 25% of the mixed pixels is inundated.


  1. Toshihiro Sakamoto, Nhan Van Nguyen, Akihiko Kotera, Hiroyuki Ohno, Naoki Ishitsuka, Masayuki Yokozawa. Detecting temporal changes in the extent of annual flooding within the Cambodia and the Vietnamese Mekong Delta from MODIS time-series imagery. Remote Sensing of Environment, Volume 109, Issue 3, 2007, Pages 295-313, ISSN 0034-4257,

This work is part of a PhD project prepared by Binh Pham Duc at LERMA – Observatoire de Paris, from 2015 to 2018.

Questions can be sent to PHAM Duc Binh ([email protected])