Articles
Using ALEXI-DisALEXI for estimation of satellite-derived water use in a California almond orchard under spatially heterogeneous conditions
Article number
1409_20
Pages
143 – 150
Language
English
Abstract
A study was carried out to evaluate modifications to the ALEXI (Atmosphere-Land Exchange Inverse) and DisALEXI (associated disaggregation technique) modeling framework to estimate water use equivalent to actual evapotranspiration (ETa) for a drip-irrigated almond orchard located in the Central Valley Region of California, USA. Modifications included the creation of a “synthetic” ALEXI ETa by redistributing coarse resolution (4 km) ALEXI ETa to higher spatial resolutions (2, 1, 0.5 km) using leaf area index derived from Harmonized Landsat and Sentinel-2 data sets.
This was done to provide more representative estimates of ETa for DisALEXI when applied over thermally heterogeneous landscapes.
For the estimation of ETa using ALEXI/DisALEXI, 26 satellite images (Landsat 8 OLI/TIRS and Landsat 9 OLI-2/TIR-2) acquired during clear sky days were used during 2022. The performance of synthetic ALEXI and subsequently DisALEXI was evaluated using measurements of ETa from an Eddy covariance system (EC). Analysis indicated that synthetic ALEXI provided more representative ETa estimates when applied over a region where a 4 km ALEXI pixel included mostly barren land and a small percentage of irrigated agriculture.
The synthetic ALEXI version at 1 km performed best when used in DisALEXI and compared to observed ETa. However, the difference in mean absolute error remained <0.2 mm day‑1 between approaches, suggesting Landsat-scale input to DisALEXI remains the most important factor in the ALEXI/DisALEXI modeling scheme.
This was done to provide more representative estimates of ETa for DisALEXI when applied over thermally heterogeneous landscapes.
For the estimation of ETa using ALEXI/DisALEXI, 26 satellite images (Landsat 8 OLI/TIRS and Landsat 9 OLI-2/TIR-2) acquired during clear sky days were used during 2022. The performance of synthetic ALEXI and subsequently DisALEXI was evaluated using measurements of ETa from an Eddy covariance system (EC). Analysis indicated that synthetic ALEXI provided more representative ETa estimates when applied over a region where a 4 km ALEXI pixel included mostly barren land and a small percentage of irrigated agriculture.
The synthetic ALEXI version at 1 km performed best when used in DisALEXI and compared to observed ETa. However, the difference in mean absolute error remained <0.2 mm day‑1 between approaches, suggesting Landsat-scale input to DisALEXI remains the most important factor in the ALEXI/DisALEXI modeling scheme.
Authors
K. Knipper, N.E. Bambach, M.C. Anderson, Y. Yang, W.P. Kustas, A.J. McElrone, M.A. Nocco, A. Torres-Rua, F. Gao, C. Hain, S.J. Castro, O. Crompton, S. Saa
Keywords
remote sensing, satellite, evapotranspiration, advection, T-REX
Online Articles (64)