Using covariates to spatially interpolate moisture availability in the Murray–Darling Basin: A novel use of remotely sensed data

Moisture availability is estimated in the 1.1 million km2 Murray–Darling Basin (MDB) in southeast Australia. Remotely sensed data from the Advanced Very High Resolution Radiometer (AVHRR) are combined with meteorological data to estimate the Normalised Difference Temperature Index (NDTI). The NDTI provides a measure of the moisture availability, the ratio of actual to potential evapotranspiration. Surface temperature minus air temperature, percent vegetation cover and net radiation explained 85% of variation in the modelled NDTI. Using these three covariates across the network of meteorological stations allows NDTI images, which maps changes in moisture availability across the MDB, to be calculated. This method uses a calculate then interpolate (CI) approach that uses the per-pixel variation present in the AVHRR data as the backbone for the spatial interpolation. Using the spatially dense AVHRR-based covariates in a CI approach avoids errors that occur between measurement points when interpolating variables for regional hydrologic modelling, most significantly the spatial pattern of rainfall. The NDTI provides a link into regional water balance modelling which does not require daily rainfall to be spatially interpolated. Assessing spatial and temporal interactions between the NDTI and the Normalised Difference Vegetation Index (NDVI) provides useful information about regional hydroecological processes, including agricultural management, within the context of Australiaʹs highly variable climate and sparse network of meteorological stations.