Estimation of evapotranspiration and energy fluxes using a deep-learning-based high-resolution emissivity model and the two-source energy balance model with sUAS information

Abstract The utility of a snow–vegetation energy balance model for estimating surface energy fluxes is evaluated with field measurements at two sites in a rangeland ecosystem in southwestern Idaho during the winter of 2007: one site dominated by aspen vegetation and the other by sagebrush. Model parameterizations are adopted from the two-source energy balance (TSEB) modeling scheme, which estimates fluxes from the vegetation and surface substrate separately using remotely sensed measurements of land surface temperature. Modifications include development of routines to account for surface snowmelt energy flux and snow masking of vegetation. Comparisons between modeled and measured surface energy fluxes of net radiation and turbulent heat showed reasonable agreement when considering measurement uncertainties in snow environments and the simplified algorithm used for the snow surface heat flux, particularly on a daily basis. There was generally better performance over the aspen field site, likely due to more reliable input data of snow depth/snow cover. The model was robust in capturing the evolution of surface energy fluxes during melt periods. The model behavior was also consistent with previous studies that indicate the occurrence of upward sensible heat fluxes during daytime owing to solar heating of vegetation limbs and branches, which often exceeds the downward sensible heat flux driving the snowmelt. However, model simulations over aspen trees showed that the upward sensible heat flux could be reversed for a lower canopy fraction owing to the dominance of downward sensible heat flux over snow. This indicates that reliable vegetation or snow cover fraction inputs to the model are needed for estimating fluxes over snow-covered landscapes.

Download Full-text

Scale analysis of evapotranspiration estimates from an energy-water balance model and remotely sensed LST

10.5194/egusphere-egu21-12870 ◽

2021 ◽

Author(s):

Nicola Paciolla ◽

Chiara Corbari ◽

Giuseppe Ciraolo ◽

Antonino Maltese ◽

Marco Mancini

Keyword(s):

Energy Balance ◽

High Resolution ◽

Water Balance ◽

Surface Temperature ◽

Remotely Sensed ◽

Water Balance Model ◽

Balance Model ◽

Source Surface ◽

Scale Analysis ◽

Energy Fluxes

Remote Sensing (RS) information has progressively found, in recent years, more and more applications in hydrological modelling as a valuable tool for easy and frequent collection of geophysical data. However, this kind of data should be handled carefully, minding its characteristics, spatial resolution and the heterogeneity of the target area.In this work, a scale analysis on evapotranspiration estimates over heterogeneous crops is performed combining a distributed energy-water balance model (FEST-EWB) and high-resolution remotely-sensed Land Surface Temperature (LST) and vegetation data.The FEST-EWB model is calibrated on measured LST, based on a procedure where every single pixel is modified independently one from the other; hence in each pixel of the analysed domain the minimum of the pixel difference between modelled RET and satellite observed LST is searched over the period of calibration.The case study is a Sicilian vineyard, with test dates in the summer of 2008. Meteorological and energy fluxes data are available from an eddy-covariance station, while LST and vegetation data are obtained from low-altitude flights at the high resolution of 1.7 metres.After a preliminary calibration on LST data and validation on energy fluxes, the scale analysis is performed in two ways: model input aggregation and model output aggregation. Four coarser scales are selected in reference to some common satellite products resolution: 10.2 m (in reference to Sentinel&#8217;s 10 m), 30.6 m (Landsat, 30 m), 244.8 m (MODIS visible, 250 m) and 734.4 m (MODIS, 1000 m). First, modelled surface temperature and evapotranspiration are aggregated to each scale by progressive averaging. Then, model inputs are upscaled to the same spatial resolutions and the model is calibrated anew, obtaining independent results directly at the target scale.The results of the two procedures are found to be quite similar, testifying to the capacity of the model to provide accurate products for a heterogeneous area even at low resolutions. The robustness of the analysis is strengthened by a further comparison with two well-established energy-balance algorithms: the one source Surface Energy Balance Algorithm for Land (SEBAL) and the Two-Source Energy Balance (TSEB) model.

Download Full-text

Modelling surface energy fluxes over a dehesa ecosystem using a two-source energy balance model and medium resolution satellite data

10.1117/12.2029235 ◽

2013 ◽

Cited By ~ 2

Author(s):

A. Andreu ◽

M. P. González-Dugo ◽

W. P. Kustas ◽

M. J. Polo ◽

M. C. Anderson

Keyword(s):

Energy Balance ◽

Surface Energy ◽

Satellite Data ◽

Energy Balance Model ◽

Balance Model ◽

Energy Fluxes ◽

Surface Energy Fluxes ◽

Medium Resolution

Download Full-text

On the Utility of High-Resolution Soil Moisture Data for Better Constraining Thermal-Based Energy Balance over Three Semi-Arid Agricultural Areas

Remote Sensing ◽

10.3390/rs13040727 ◽

2021 ◽

Vol 13 (4) ◽

pp. 727

Author(s):

Bouchra Ait Hssaine ◽

Abdelghani Chehbouni ◽

Salah Er-Raki ◽

Said Khabba ◽

Jamal Ezzahar ◽

...

Keyword(s):

Soil Moisture ◽

Energy Balance ◽

High Resolution ◽

Energy Balance Model ◽

Balance Model ◽

Wheat Crop ◽

Field Scale ◽

Thermal Data ◽

Agricultural Areas ◽

Semi Arid

Over semi-arid agricultural areas, the surface energy balance and its components are largely dependent on the soil water availability. In such conditions, the land surface temperature (LST) retrieved from the thermal bands has been commonly used to represent the high spatial variability of the surface evaporative fraction and associated fluxes. In contrast, however, the soil moisture (SM) retrieved from microwave data has rarely been used thus far due to the unavailability of high-resolution (field scale) SM products until recent times. Soil evaporation is controlled by the surface SM. Moreover, the surface SM dynamics is temporally related to root zone SM, which provides information about the water status of plants. The aim of this work was to assess the gain in terms of flux estimates when integrating microwave-derived SM data in a thermal-based energy balance model at the field scale. In this study, SM products were derived from three different methodologies: the first approach inverts SM, labeled hereafter as ‘SMO20’, from the backscattering coefficient and the interferometric coherence derived from Sentinel-1 products in the water cloud model (WCM); the second approach inverts SM from Sentinel-1 and Sentinel-2 data based on machine learning algorithms trained on a synthetic dataset simulated by the WCM noted ‘SME16’; and the third approach disaggregates the soil moisture active and passive SM at 100 m resolution using Landsat optical/thermal data ‘SMO19’. These SM products, combined with the Landsat based vegetation index and LST, are integrated simultaneously within an energy balance model (TSEB-SM) to predict the latent (LE) and sensible (H) heat fluxes over two irrigated and rainfed wheat crop sites located in the Haouz Plain in the center of Morocco. H and LE were measured over each site using an eddy covariance system and their values were used to evaluate the potential of TSEB-SM against the classical two source energy balance (TSEB) model solely based on optical/thermal data. Globally, TSEB systematically overestimates LE (mean bias of 100 W/m2) and underestimates H (mean bias of −110 W/m2), while TSEB-SM significantly reduces those biases, regardless of the SM product used as input. This is linked to the parameterization of the Priestley Taylor coefficient, which is set to αPT = 1.26 by default in TSEB and adjusted across the season in TSEB-SM. The best performance of TSEB-SM was obtained over the irrigated field using the three retrieved SM products with a mean R2 of 0.72 and 0.92, and a mean RMSE of 31 and 36 W/m2 for LE and H, respectively. This opens up perspectives for applying the TSEB-SM model over extended irrigated agricultural areas to better predict the crop water needs at the field scale.

Download Full-text