Short Virtual Presentation & Digital Poster 10th Australian Stream Management Conference 2021

Mapping water in dry flat environments: A study using Sentinel imagery in the northern Murray-Darling Basin (#73)

Paul Frazier 1 , Niva Verma 1 , Mark Southwell 2 , David Weldrake 3
  1. 2rog Consulting, Armidale, NSW, Australia
  2. Ecosystem management, University of New England, Armidale, NSW, Australia
  3. Murray-Darling Basin Authority, Canberra, ACT
  • Knowing where our water is as a function of flow, how far it extends and how it dries back is critical for managing environmental water and improving our environment.

  • We used Sentinel-2 satellite imagery to examine the accuracy and applications of water extent mapping in the northern Murray-Darling Basin. In the Warrego western floodplain and Gwydir wetlands we mapped water extent using eight common water indices to determine mapping accuracies. In the Darling River we examined in-channel pools to link pool recession to flow conditions.

  • All indices performed well with mapping accuracy generally over 90%. The modified normalised difference water index (MNDWI) performed consistently better than the others and we noted that the normalised difference water index (NDWI) was particularly poor at detecting inundation when water was highly turbid. Large pools in the Darling could be mapped using the Sentinel imagery and the increase in water surface area in the target reach resulting from environmental flow delivery could be detected. We also developed a set of outputs to describe inundation persistence and duration.

  • We can map surface water with a high degree of accuracy over large areas many times per year using opensource data. Having access to a simple map of water extent up to 50 times per year provides great insight into the outcomes of our environmental watering. We can target flow delivery more precisely and optimise our water use. In our dry and drying continent optimal water use is imperative for rebuilding our environment.

 

Download Full Paper