Table 3.
Comparisons of the different ET estimation methods.
| Methods | Applications with UAVs | Advantages | Disadvantages |
|---|---|---|---|
| OSEB | Vineyard [35], grassland [40] | (1) Treat the surface as big leaf and therefore as a simple uniform layer. | (1) Uses empirical parameters to explain differences in the aerodynamic and radiometric components; (2) Assumes the whole surface as a uniform layer, which does not take advantage of UAV high-resolution imagery; (3) Less sensitive to land surface temperature variations than the TSEB model. |
| HRMET | Peach, nectarine [16], and corn | (1) Only requires basic meteorological data, spatial surface temperature, and canopy structure data; (2) Does not depend on wet and dry reference features to calculate turbulent fluxes. | (1) Needs more validation for clumped canopy structure, such as trees and vines. |
| ML/ANN | Vineyard [56] | (1) Capture non-linear crop characteristics | (1) Requires large amount of data for training models and validation |
| TSEB | Barley [21], vineyard [35,36,37,38], olive [31], sorghum and corn [39], grassland [40] | (1) The calculation of sensible heat flux and latent heat flux for canopy and soil are separate; (2) Parameterization of resistances is easier compared with a single layer model | (1) Sensitive to the temperature difference between the land surface and air; (2) The measurement of the absolute land surface temperature is inaccurate |
| DTD | Barley [21], corn and soybean | (1) One more input dataset, the land surface temperature retrieved one hour after sunrise; (2) Minimizes the bias in the temperature estimation; (3) Separates the land surface temperature into vegetation and soil temperatures | (1) Requires flights at two times during the morning hours, thus complicating flight missions |
| SEBAL | Corn and soybean [32] | (1) Requires minimum ground-based data; (2) Automatic internal correction | (1) Selecting hot or cold pixels is subjective, which can cause variations in ET estimation |
| METRIC | Vineyard [33,34] | (1) Eliminates the need for absolute surface temperature calibration; (2) Requires minimum ground-based data; (3) Automatic internal correction | (1) Selecting hot or cold pixels is subjective, which can cause variations in ET estimation |