Comparing the utility of microwave and thermal remote-sensing constraints in two-source energy balance modeling over an agricultural landscape

A two-source (soil + vegetation) energy balance model using microwave-derived near-surface soil moisture as a key boundary condition (TSM_SM) and another scheme using thermal-infrared (radiometric) surface temperature (TSM_TH) were applied to remote sensing data collected over a corn and soybean production region in central Iowa during the Soil Moisture Atmosphere Coupling Experiment (SMACEX)/Soil Moisture Experiment of 2002 (SMEX02). The TSM_SM was run using fields of near-surface soil moisture from microwave imagery collected by aircraft on six days during the experiment, yielding a root mean square difference (RMSD) between model estimates and tower measurements of net radiation (Rn) and soil heat flux (G) of approximately 20 W m^− 2, and 45 W m^− 2 for sensible (H) and latent heating (LE). Similar results for H and LE were obtained at landscape/regional scales when comparing model output with transect-average aircraft flux measurements. Flux predictions from the TSM_SM and TSM_TH models were compared for two days when both airborne microwave-derived soil moisture and radiometric surface temperature (T_R) data from Landsat were available. These two days represented contrasting conditions of moderate crop cover/dry soil surface and dense crop cover/moist soil surface. Surface temperature diagnosed by the TSM_SM was also compared directly to the remotely sensed T_R fields as an additional means of model validation. The TSM_SM performed well under moderate crop cover/dry soil surface conditions, but yielded larger discrepancies with observed heat fluxes and T_R under the high crop cover/moist soil surface conditions. Flux predictions from the thermal-based two-source model typically showed biases of opposite sign, suggesting that an average of the flux output from both modeling schemes may improve overall accuracy in flux predictions, in effect incorporating multiple remote-sensing constraints on canopy and soil fluxes.     

Comparison of integrating LAS/MODIS data into a land surface model for improved estimation of surface variables through data assimilation

In this article, land surface temperature (LST) and sensible heat flux (H) data assimilation schemes were developed separately using the ensemble Kalman filter (EnKF) and the common land model (CoLM). Surface measurements of ground temperature, H, and latent heat flux (LE) collected at the Yucheng (longitude: 116° 36′ E; latitude: 36° 57′ N) and Arou (longitude: 100° 27′ E; latitude: 38° 02′ N) experimental stations were compared with the predictions by assimilating different observation sources into the CoLM. The results showed that both LST and H data assimilation schemes could improve the estimation of ground temperature and H. The root mean square error (RMSE) compared between the predictions and in situ measurements decreased more significantly with the assimilation of values of H measured by a large aperture scintillometer (LAS). Assimilating Moderate Resolution Imaging Spectroradiometer (MODIS) LST only slightly improved the predictions of H and ground temperature. Daytime to night-time comparison results using both assimilation schemes also indicated that accurately quantifying model, prediction, and observation error would improve the efficiency of the assimilation systems. The newly developed land data assimilation schemes have proved to be a feasible and practical method to improve the predictions of heat fluxes and ground temperature from CoLM. Moreover, integrating multisource data (LAS and MODIS LST) simultaneously into the land surface model is believed to result in an efficient and robust way to improve the accuracy of model predictions from a theoretical point of view.     

Evapotranspiration calculated from remote multispectral and ground station meteorological data

Evapotranspiration was evaluated by combining remotely sensed reflected solar radiation and surface temperatures with ground station meteorological data (incoming solar radiation, air temperature, windspeed, and vapor pressure) to calculate net radiation and sensible heat flux. Soil heat flux was estimated as a fraction of the net radiation. Instantaneous values of ET were calculated for 18 wheat plots for 44 cloudless days over a growing season. Three of the 18 plots contained lysimeters which provided data to compare against the instantaneous values. For the remaining plots, daily ET was estimated from the instantaneous data and compared with values calculated from soil water contents measured with a neutron moisture meter. For generally clear sky conditions, the comparisons indicated that ET could be adequately evaluated using a combination of remotely sensed and ground based meteorological data. The results suggest that ET maps of relatively large areas could be made using this method with data from airborne sensors. The extent of the area covered appears to be limited by the distance that air temperature and windspeed data can be extrapolated.     

A new method for estimating of evapotranspiration and surface soil moisture from optical and thermal infrared measurements: the simplified triangle

Earth Observation (EO) provides a promising approach towards deriving accurate spatiotemporal estimates of key parameters characterizing land surface interactions, such as latent (LE) and sensible (H) heat fluxes as well as soil moisture content. This paper proposes a very simple method to implement, yet reliable to calculate evapotranspiration fraction (EF) and surface moisture availability (M_o) from remotely sensed imagery of Normalized Difference Vegetation Index (NDVI) and surface radiometric temperature (T_ir). The method is unique in that it derives all of its information solely from these two images. As such, it does not depend on knowing ancillary surface or atmospheric parameters, nor does it require the use of a land surface model. The procedure for computing spatiotemporal estimates of these important land surface parameters is outlined herein stepwise for practical application by the user. Moreover, as the newly developed scheme is not tied to any particular sensor, it can also be implemented with technologically advanced EO sensors launched recently or planned to be launched such as Landsat 8 and Sentinel 3. The latter offers a number of key advantages in terms of future implementation of the method and wider use for research and practical applications alike.     

Modeling and assimilation of root zone soil moisture using remote sensing observations in Walnut Gulch Watershed during SMEX04

Soil moisture status in the root zone is an important component of the water cycle at all spatial scales (e.g., point, field, catchment, watershed, and region). In this study, the spatio-temporal evolution of root zone soil moisture of the Walnut Gulch Experimental Watershed (WGEW) in Arizona was investigated during the Soil Moisture Experiment 2004 (SMEX04). Root zone soil moisture was estimated via assimilation of aircraft-based remotely sensed surface soil moisture into a distributed Soil-Water-Atmosphere-Plant (SWAP) model. An ensemble square root filter (EnSRF) based on a Kalman filtering scheme was used for assimilating the aircraft-based soil moisture observations at a spatial resolution of 800 m × 800 m. The SWAP model inputs were derived from the SSURGO soil database, LAI (Leaf Area Index) data from SMEX04 database, and data from meteorological stations/rain gauges at the WGEW. Model predictions are presented in terms of temporal evolution of soil moisture probability density function at various depths across the WGEW. The assimilation of the remotely sensed surface soil moisture observations had limited influence on the profile soil moisture. More specifically, root zone soil moisture depended mostly on the soil type. Modeled soil moisture profile estimates were compared to field measurements made periodically during the experiment at the ground based soil moisture stations in the watershed. Comparisons showed that the ground-based soil moisture observations at various depths were within ± 1 standard deviation of the modeled profile soil moisture. Density plots of root zone soil moisture at various depths in the WGEW exhibited multi-modal variations due to the uneven distribution of precipitation and the heterogeneity of soil types and soil layers across the watershed.     

Evaluation of remotely sensed and modelled soil moisture products using global ground-based in situ observations

In situ soil moisture data from more than 200 stations located in Africa, Australia, Europe and the United States are used to determine the reliability of three soil moisture products, one analysis from the ECMWF (European Centre for Medium-Range Weather Forecasts) numerical weather prediction system (SM-DAS-2) and two remotely sensed soil moisture products, namely ASCAT (Advanced scatterometer) and SMOS (Soil Moisture Ocean Salinity). SM-DAS-2 is produced offline at ECMWF and relies on an advanced surface data assimilation system (Extended Kalman Filter) used to optimally combine conventional observations with satellite measurements. ASCAT remotely sensed surface soil moisture is provided in near real time by EUMETSAT. At ECMWF, ASCAT is used for soil moisture analyses in SM-DAS-2, also. Finally the SMOS remotely sensed soil moisture data level two product developed at CESBIO is used. Evaluation of the times series as well as of the anomaly values, shows good performances of the three products to capture surface soil moisture annual cycle and short term variability. Correlations with in situ data are very satisfactory over most of the investigated sites located in contrasted biomes and climate conditions with averaged values of 0.70 for SM-DAS-2, 0.53 for ASCAT and 0.54 for SMOS. Although radio frequency interference disturbs the natural microwave emission of the Earth observed by SMOS in several parts of the world, hence the soil moisture retrieval, performances of SMOS over Australia are very encouraging.     

The influence of meteorological conditions and topographic parameters on the beech forest microclimate of Simbruini Mountains,central Italy

The relation between vegetation surface temperature and remotely sensed spectral vegetation indices has been examined by a number of authors. The observed linear decrease in surface temperature with the increase in vegetation cover density has generally been explained in terms of the increase in latent heat flux associated with greater amounts of transpirationally active vegetation. However, these investigations have initially concentrated in spatially uniform crop or pasture targets on level terrain, excluding more complex forested environments with variable Sun-sensor-surface geometry. In irregular terrains, the vegetation surface temperature may be strongly influenced by topographic parameters, such as altitude and insulation angle, so that the actual forest microclimate is often difficult to evaluate. Moreover, in the thermal regime, the emission of radiative flux within the canopy element is very tightly coupled to the environment through driving mechanisms such as meteorological conditions. In fact, the allocation of absorbed solar radiation into sensible heat flux and latent heat flux is dominated by the availability of water at the Earth's surface and thus by precipitations and air temperature conditions. In this paper, which uses remotely sensed inputs of surface temperature and vegetation fractional cover, the effects of topographic parameters and vegetation cover density on surface temperature of vegetation are investigated based on Landsat 5 satellite images obtained in the daytime of two clear summer days with different antecedent meteorological conditions. For both scenes analysed, results indicate that altitude as well as the orientation of the surface relative to the Sun were the most important factors controlling surface temperatures of beech forests of Simbruini Mountains, in central Italy.     

A thermal-based remote sensing technique for routine mapping of land-surface carbon,water and energy fluxes from field to regional scales

Robust yet simple remote sensing methodologies for mapping instantaneous land-surface fluxes of water, energy and CO₂ exchange within a coupled framework add significant value to large-scale monitoring networks like FLUXNET, facilitating upscaling of tower flux observations to address questions of regional carbon cycling and water availability. This study investigates the implementation of an analytical, light-use efficiency (LUE) based model of canopy resistance within a Two-Source Energy Balance (TSEB) scheme driven primarily by thermal remote sensing inputs. The LUE model computes coupled canopy-scale carbon assimilation and transpiration fluxes, and replaces a Priestley–Taylor (PT) based transpiration estimate used in the original form of the TSEB model. In turn, the thermal remote sensing data provide valuable diagnostic information about the sub-surface moisture status, obviating the need for precipitation input data and prognostic modeling of the soil water balance. Both the LUE and PT forms of the model are compared with eddy covariance tower measurements acquired in rangeland near El Reno, OK. The LUE method resulted in improved partitioning of the surface energy budget, capturing effects of midday stomatal closure in response to increased vapor pressure deficit and reducing errors in half-hourly flux predictions from 16 to 12%. The spatial distribution of CO₂ flux was mapped over the El Reno study area using data from an airborne thermal imaging system and compared to fluxes measured by an aircraft flying a transect over rangeland, riparian areas, and harvested winter wheat. Soil respiration contributions to the net carbon flux were modeled spatially using remotely sensed estimates of soil surface temperature, soil moisture, and leaf area index. Modeled carbon and water fluxes from this heterogeneous landscape compared well in magnitude and spatial pattern to the aircraft fluxes. The thermal inputs proved to be valuable in modifying the effective LUE from a nominal species-dependent value. The model associates cooler canopy temperatures with enhanced transpiration, indicating higher canopy conductance and carbon assimilation rates. The surface energy balance constraint in this modeling approach provides a useful and physically intuitive mechanism for incorporating subtle signatures of soil moisture deficiencies and reduced stomatal aperture, manifest in the thermal band signal, into the coupled carbon and water flux estimates.     

Experiments of one-dimensional soil moisture assimilation system based on ensemble Kalman filter

Ensemble Kalman filter is a new sequential data assimilation algorithm which was originally developed for atmospheric and oceanographic data assimilation. It can be applied to calculate error covariance matrix through Monte-Carlo simulation. This approach is able to resolve the nonlinearity and discontinuity existed within model operator and observation operator. When observation data are assimilated at each time step, error covariances are estimated from the phase-space distribution of an ensemble of model states. The error statistics is then used to calculate Kalman gain matrix and analysis increments. In this study, we develop a one-dimensional soil moisture data assimilation system based on ensemble Kalman filter, the Simple Biosphere Model (SiB2) and microwave radiation transfer model (AIEM, advanced integration equation model). We conduct numerical experiments to assimilate in situ soil surface moisture measurements and low-frequency passive microwave remote sensing data into a land surface model, respectively. The results indicate that data assimilation can significantly improve the soil surface moisture estimation. The improvement in root zone is related to the model bias errors at surface layer and root zone. The soil moisture does not vary significantly in deep layer. Additionally, the ensemble Kalman filter is predominant in dealing with the nonlinearity of model operator and observation operator. It is practical and effective for assimilating observations in situ and remotely sensed data into land surface models.     

An intercomparison of the Surface Energy Balance Algorithm for Land (SEBAL) and the Two-Source Energy Balance (TSEB) modeling schemes

An intercomparison of output from two models estimating spatially distributed surface energy fluxes from remotely sensed imagery is conducted. A major difference between the two models is whether the soil and vegetation components of the scene are treated separately (Two-Source Energy Balance; TSEB approach) or as a lumped composite (one-source approach; Surface Energy Balance Algorithm for Land; SEBAL) in the parameterization of radiative and turbulent exchanges with the overlying air. Comparisons are performed using data from two largescale field experiments covering sub-humid grassland (Southern Great Plains '97) and semi-arid rangeland (Monsoon '90) having very different landscape properties. In general, there was reasonable agreement between flux output from both models versus a handful of flux tower observations. However, spatial intercomparisons of model output over the full modeling domains yielded relatively large discrepancies (on the order of 100 W m^− 2) in sensible heat flux (H) that are related to land cover. In particular, bare soil and sparsely vegetated areas yielded the largest discrepancies, with TSEB fluxes being in better agreement with tower observations. Modifications to SEBAL inputs that reduced discrepancies with TSEB and observations for bare soil and shrub classes tended to increase differences for other land cover classes. In particular, improvements to SEBAL inputs of surface roughness for momentum tended to exacerbate errors with respect to observed fluxes. These results suggest that some of the simplifying assumptions in SEBAL may not be strictly applicable over the wide range in conditions present within these landscapes. An analysis of TSEB and SEBAL sensitivity to uncertainties in primary inputs indicated that errors in surface temperature or surface-air temperature differences had the greatest impact on H estimates. Inputs of secondary importance were fractional vegetation cover for TSEB, while for SEBAL, the selection of pixels representing wet and dry moisture end-member conditions significantly influenced flux predictions. The models were also run using input fields derived from both local and remote data sources, to test performance under conditions of varying ancillary data availability. In this case, both models performed similarly under both constraints.