Estimation of air temperature and reference evapotranspiration using MODIS land surface temperature over Greece

Moderate Resolution Imaging Spectroradiometer (MODIS), land surface temperature data, during daytime (LST_day) or night-time (LST_night), were employed for predicting maximum (T_max) or minimum (T_min) air temperature measured at ground stations, respectively, in order to be used as alternative inputs in minimum data-based reference evapotranspiration (ET) models in 28 stations in Greece during the growing season (May–October). The deviations between daily LST_night and T_min were found to be small, but they were greater between LST_day and T_max. Furthermore, the temperature vegetation index (TVX) method was employed for achieving more accurate T_max values from LST_day, after determining the normalized difference vegetation index of a full canopy (NDVI_max). The TVX method was validated on ‘temporal’ basis, but when the method was tested spatially, the improvement on the T_max estimates from LST_day was not encouraging, for being used operationally over Greece. Thus, LST_day or LST_night MODIS data were used as inputs in three ET models [Hargreaves–Samani, Droogers–Allen, and Reference Evapotranspiration Model for Complex Terrains (REMCT)] and their estimations, as compared with ground-based Penman–Monteith estimates, indicated that the REMCT model achieved the most accurate ET predictions (r = 0.93, mean bias error = 0.44 mm day^–1 and root mean square error = 0.74 mm day^–1), which can allow the spatial analysis of ET at higher spatial resolutions in areas with lack of ground temperature data.     

Evaluation of remotely sensed actual evapotranspiration products from COMS and MODIS at two different flux tower sites in Korea

Estimating the evapotranspiration (ET) is a requirement for water resource management and agricultural productions to understand the interaction between the land surface and the atmosphere. Most remote-sensing-based ET is estimated from polar orbiting satellites having low frequencies of observation. However, observing the continuous spatio-temporal variation of ET from a geostationary satellite to determine water management usage is essential. In this study, we utilized the revised remote-sensing-based Penman–Monteith (revised RS-PM) model to estimate ET in three different timescales (instantaneous, daily, and monthly). The data from a polar orbiting satellite, the Moderate Resolution Imaging Spectroradiometer (MODIS), and a geostationary satellite, the Communication, Ocean, and Meteorological Satellite (COMS), were collected from April to December 2011 to force the revised RS-PM model. The estimated ET from COMS and MODIS was compared with measured ET obtained from two different flux tower sites having different land surface characteristics in Korea, i.e. Sulma (SMC) with mixed forest and Cheongmi (CFC) with rice paddy as dominant vegetation. Compared with flux tower measurements, the estimated ET on instantaneous and daily timescales from both satellites was highly overestimated at SMC when compared with the flux tower ET (Bias of 41.19–145.10 W m^?2 and RMSE of 69.61–188.78 W m^?2), while estimated ET results were slightly better at the CFC site (Bias of –27.28–13.24 W m^?2 and RMSE of 45.19–71.82 W m^?2, respectively). These errors in results were primarily caused due to the overestimated leaf area index that was obtained from satellite products. Nevertheless, the satellite-based ET indicated reasonable agreement with flux tower ET. Monthly average ET from both satellites showed nearly similar patterns during the entire study periods, except for the summer season. The difference between COMS and MODIS estimations during the summer season was mainly propagated due to the difference in the number of acquired satellite images. This study showed that the higher frequency of COMS than MODIS observations makes it more ideal to continuously monitor ET as a geostationary satellite with high spatio-temporal coverage of a geostationary satellite.     

Estimation of daily actual evapotranspiration from remotely sensed data under complex terrain over the upper Chao river basin in North China

Daily actual evapotranspiration over the upper Chao river basin in North China on 23 June 2005 was estimated based on the Surface Energy Balance Algorithm for Land (SEBAL), in which the parameterization schemes for calculating the instantaneous solar radiation and daily integrated radiation were improved by accounting for the variations in slope and azimuth of land surface and terrain shadow in mountainous areas. The evapotranspiration (ET) estimated from satellite data in this study for the whole watershed ranges from 0 mm to 7.3 mm day^?1 with a mean of 3.4 mm day^?1, which was validated by Penman–Monteith approaches for water body and paddy land. The comparison of ET estimates for a wide range of land cover types reflected distinct mechanisms of energy partition and water removal of various land cover types, showing differences in the spatial distribution pattern of ET, which could be not only the reflection but also the driving force of advection and local circulation that may violate the surface energy balance equation in the vertical direction. The spatial variation in daily solar radiation and ET estimates under the complex terrain of forest land were elaborated and evaluated by exploring the relationship between ET estimates and elevations for wood land and grass land. In addition, the utility and limitations of SEBAL's applicability to watersheds with various land cover types and complex terrain were analysed.     

Improving remotely sensed actual evapotranspiration estimation with raster meteorological data

Evapotranspiration is a process driven by weather, vegetation, and soil conditions. The complex interrelations among these parameters have been modelled by numerous remote-sensing energy balance algorithms. When estimating evapotranspiration on a regional scale, the spatial variability of the weather parameters is important and thus closer attention to the meteorological input data is required. The aim of this work is to improve the accuracy of estimating actual evapotranspiration by integrating outputs from a meteorological model into a remotely sensed energy balance model. In order to achieve this, a time series of Terra Moderate Resolution Imaging Spectroradiometer (MODIS) satellite images were processed to retrieve daily evapotranspiration values using raster meteorological data. The ITA-MyWater tool implementing the ReSET-Raster algorithm was used in the Tâmega trans-boundary watershed shared by Portugal and Spain. The results were compared to the global MODIS evapotranspiration products for validation, achieving a coefficient of correlation of 0.61 and a root mean square error of 0.92 mm day^–1. Compared with an actual evapotranspiration map that was generated using weather station data, there were improvements in the spatial distribution, especially in dry areas where differences between evapotranspiration estimations of up to 1.88 mm day^–1 were noticed. The proposed methodology contributes to the improved estimation of water use, an important parameter of water cycles, using satellite remote-sensing data.     

Multi-scale validation of GLEAM evapotranspiration products over China via ChinaFLUX ET measurements

Land evapotranspiration (ET) is a key component of terrestrial ecosystems, as it is the nexus of hydrological, energy, and carbon cycles. Satellite-based observations are commonly utilized to provide high-resolution, large-scale ET estimates. The ground-based validation of such large-scale estimates is necessary to ensure that remotely sensed ET characteristics are accurate, and to extend their various applications. The Global Land-surface Evaporation Amsterdam Methodology (GLEAM) combines a wide range of multi-satellite observations to estimate daily actual evaporation through a process-based methodology. In this study, we focused on evaluating a daily GLEAM 0.25° ET product using in-situ eddy covariance (EC) ET data (2003–2005) as a benchmark at eight sites from the Chinese Flux Observation and Research Network (ChinaFLUX), which contains several biome types (croplands, grasslands, shrublands, savannas, and forests) across China at a range of temporal scales (from daily, to monthly, to annual). The results indicated that the ET products of the Global Land-surface Evaporation Amsterdam Methodology (GLEAM ET) over different time scales can estimate actual ET with reasonable accuracy. GLEAM showed high skill scores for most of the land-cover types except at the Xishuangbanna forest site (XSBN), where significantly systematic bias was detected at each individual temporal scale. Overall, GLEAM ET products were closer to the EC observations at the three grassland sites than at the four forest sites or the cropland site. GLEAM significantly overestimated the EC measurements at the four forest sites and one cropland site, while a slight underestimation occurred at the three grassland sites; there was a year-long systematic overestimation for GLEAM at the four forest sites. The daily GLEAM ET aggregated by monthly and annual data agreed more closely with EC measurements than those taken at the daily timescale. The results also showed a high average correlation coefficient (r) with in-situ EC observations at all sites, at daily (r = 0.71), monthly (r = 0.86), and annual (r = 0.79) time scales in addition to ET season-dependent characteristics for satellite estimation errors. The results presented here contribute to further assessment of the quality and uncertainty of GLEAM ET products, which may benefit future advancements in the ET algorithm and its product quality.     

Integrating two layers of soil moisture parameters into the MOD16 algorithm to improve evapotranspiration estimations

To integrate soil moisture into the algorithm of the Moderate Resolution Imaging Spectroradiometer (MODIS) global evapotranspiration (ET) project (MOD16), two improvements were implemented: two layers of relative soil moisture parameters were combined with a surface resistance model; and the complementary relationship was replaced with the Penman-Monteith (P-M) method to estimate the dry soil surface evaporation. In the vegetation surface resistance model, a multiplier R_sm1 was added, and the influence of the relative soil moisture in the root zone was accounted for. In the soil surface resistance model, an empirical exponential relationship was used. To calculate the relative soil moisture parameters, soil hydraulic parameters, such as field capacity (Fc), wilting point (Wp), and saturation point (Sp), were estimated according to the soil texture information; these parameters were used as critical values to estimate the relative soil moisture. Both the MOD16 method and improved method were validated using ET flux data collected at nine flux-tower sites in the USA from 2000 to 2009. The mean absolute BIAS and the root mean square error (RMSE) decreased from 0.36 to 0.30 mm day^–1 and from 1.14 to 0.97 mm day^–1, respectively, after integrating the soil moisture parameters. Meanwhile, the mean correlation coefficient (R) for the nine sites increased from 0.54 to 0.70. Therefore, the improved method performed better than the MOD16 method. Furthermore, the uncertainties associated with the MODIS leaf area index (LAI) products, flux-tower measurements, soil texture, soil moisture, and model parameters were analysed. The outlook for future modifications was also discussed.     

Development of real-time reference evapotranspiration at the regional scale using satellite-based observations

Real-time data of reference evapotranspiration (ET₀) at different space-time scales are essential to regional agricultural drought assessment, water accounting at the watershed to basin scale, and provide irrigation advisory to farmers. Here, we present a data-fusion approach that integrates satellite-based insolation product (8 km) from an Indian geostationary satellite (Kalpana-1) sensor (VHRR; Very High Resolution Radiometer) and high-resolution (~ 5 km) short-range weather forecast into an FAO56 model based on the classical Penman–Monteith (P-M) formulation. Five year (2009–2013) mean monthly estimates from the daily ET₀ product over the Indian landmass were found to vary between 10 and 350 mm. It increased from January to May (70–350 mm), followed by a decrease to reach the lowest in November (10–140 mm), thus typically showing unimodal distribution. The comparison of daily space-based and station-based estimates (at six ground stations) produced a root mean square deviation (RMSD) ranging from 21% to 38% for 977 paired data sets with the correlation coefficient (r) varying from 0.32 to 0.82. The error was reduced from 25% to 10% with an increase in ‘r’ from 0.43 to 0.98 for daily to 10 day summation period. Spatial grid-to-grid comparison of monthly ET₀ estimates with Global Data Assimilation System (GDAS) potential evapotranspiration (PET) showed RMSD within a range of 1.4–18.4% for most of the months, except for two. Further ET₀ analysis over normal and drought years showed that it could be used for comprehensive drought assessment with other existing indicators.     

Improvements to a MODIS global terrestrial evapotranspiration algorithm

MODIS global evapotranspiration (ET) products by Mu et al. [Mu, Q., Heinsch, F. A., Zhao, M., Running, S. W. (2007). Development of a global evapotranspiration algorithm based on MODIS and global meteorology data. Remote Sensing of Environment, 111, 519-536. doi: 10.1016/j.rse.2007.04.015] are the first regular 1-km² land surface ET dataset for the 109.03 Million km² global vegetated land areas at an 8-day interval. In this study, we have further improved the ET algorithm in Mu et al. (2007a, hereafter called old algorithm) by 1) simplifying the calculation of vegetation cover fraction; 2) calculating ET as the sum of daytime and nighttime components; 3) adding soil heat flux calculation; 4) improving estimates of stomatal conductance, aerodynamic resistance and boundary layer resistance; 5) separating dry canopy surface from the wet; and 6) dividing soil surface into saturated wet surface and moist surface. We compared the improved algorithm with the old one both globally and locally at 46 eddy flux towers. The global annual total ET over the vegetated land surface is 62.8 × 10³ km³, agrees very well with other reported estimates of 65.5 × 10³ km³ over the terrestrial land surface, which is much higher than 45.8 × 10³ km³ estimated with the old algorithm. For ET evaluation at eddy flux towers, the improved algorithm reduces mean absolute bias (MAE) of daily ET from 0.39 mm day⁻¹ to 0.33 mm day⁻¹ driven by tower meteorological data, and from 0.40 mm day⁻¹ to 0.31 mm day⁻¹ driven by GMAO data, a global meteorological reanalysis dataset. MAE values by the improved ET algorithm are 24.6% and 24.1% of the ET measured from towers, within the range (10-30%) of the reported uncertainties in ET measurements, implying an enhanced accuracy of the improved algorithm. Compared to the old algorithm, the improved algorithm increases the skill score with tower-driven ET estimates from 0.50 to 0.55, and from 0.46 to 0.53 with GMAO-driven ET. Based on these results, the improved ET algorithm has a better performance in generating global ET data products, providing critical information on global terrestrial water and energy cycles and environmental changes.     

Using ASTER satellite data to calculate riparian evapotranspiration in the Middle Rio Grande,New Mexico

Riparian evapotranspiration (ET) in the Rio Grande Basin in New Mexico, USA is a major component of the hydrological system. Over a period of several years, ET has been measured in selected locations of dense saltcedar and cottonwood vegetation. Riparian vegetation varies in density, species and soil moisture availability, and to obtain accurate measurements, multiple sampling points are needed, making the process costly and impractical. An alternative solution involves using remotely sensed data to estimate ET over large areas. In this study, daily ET values were measured using eddy covariance flux towers installed in areas of saltcedar and cottonwood vegetation. At these sites, remotely sensed satellite data from the National Aeronautics and Space Administration (NASA) Terra Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) were used to calculate the albedo, normalized difference vegetation index (NDVI) and surface temperature. A surface energy balance model was used to calculate ET values from the ASTER data, which were available for 7 days in the year. Comparison between the daily ET values of saltcedar and cottonwood measured from the flux towers and calculated from remote sensing resulted in a mean square error (MSE) of 0.16 and 0.37 mm day^?1, respectively. The regional map of ET generated from the remote sensing data demonstrated considerable variation in ET, ranging from 0 to 9.8 mm day^?1, with a mean of 5.5 mm day^?1 and standard deviation of 1.85 mm day^?1 (n = 427481 pixels) excluding open water. This was due to variations in plant variety and density, soil type and moisture availability, and the depth to water table.     

Synthetic retrieval of hourly net ecosystem exchange using the neural network model with combined MI and GOCI geostationary sensor datasets and ground-based measurements

Net ecosystem carbon dioxide (CO₂) exchange (NEE) is a key parameter for understanding the terrestrial plant ecosystems, but it is difficult to monitor or predict over large areas at fine temporal resolutions. In this research, we estimated the hourly NEE using a combination of the integrated neural network (NN) model with geostationary satellite imagery to overcome the limitations of existing daily polar orbiting satellite-derived carbon flux products. Two sets of satellite imageries (i.e. the meteorological imager (MI) and geostationary ocean colour imager (GOCI) aboard communication, ocean, and meteorological satellite (COMS)) and CO₂ flux data derived from eddy covariance measurements were used to verify the feasibility of applying hourly geostationary satellite imagery with an NN-based approach for estimating NEE at high temporal resolutions. For the NN model, the optimum neuronal architecture was established using an NN with one hidden layer that was trained using the Levenberg–Marquardt back propagation algorithm. The hourly NEE values estimated in test period from the NN model using the combined COMS MI and GOCI imagery and ground measurements as model inputs were compared with the eddy covariance NEE values from the measurement tower, which yielded reliable statistical agreement. The hourly NEE results from the NN model based on COMS MI and GOCI imagery and ground measurement data had the highest accuracy (RMSE = 2.026 μmol m^?2 s^?2, R = 0.975), while the root mean square error (RMSE) and the regression coefficient (R) generated by the NN model based on satellite imagery as the sole input variable were relatively lower (RMSE = 3.230 μmol m^?2 s^?2, R = 0.952). Although the simulations for the satellite-only NEE were showed as lower accuracy than the NN model that included all input variables, the hourly variations in NEE also appeared to describe its daily growth and development pattern well, indicating the possibility of deriving hourly-based products from the proposed NN model using geostationary satellite data as inputs.     

Impact and consequences of evapotranspiration changes on water resources availability in the arid Zhangye Basin,China

Evapotranspiration (ET) plays an important role in the hydrological cycle and it is essential to estimate ET accurately for the evaluation of available water resources. This is most important in arid and semi‐arid regions. In this paper, the long‐term changes in daily ET in the semi‐arid Zhangye Basin in northwest China and its impact factors were studied. The spatial distribution of ET was assessed by using the Surface Energy Balance System (SEBS). Cloud‐free National Oceanic and Atmospheric Administration Advanced (NOAA) Very High Resolution Radiometer (AVHRR) September images over the Zhangye Basin from 1990 to 2004 were used in combination with SEBS to estimate ET at a spatial resolution of 1.1 km. This daily ET was converted to a monthly ET (for September) using daily pan evaporation values from a meteorological station in the study area. Spatial aggregation of all pixels yielded the total monthly ET for the whole study area. Subsequently, the monthly ET was extrapolated to annual ET values using the pan evaporation data. The results were validated with ground‐based measurements on the water balance for the whole Zhangye Basin. The annual ET increased gradually from 23.7×10⁸ m³ in 1990 to 26.9×10⁸ m³ in 2004 for the Zhangye Basin. The main cause appeared to be change in vegetation.     

Development of a global evapotranspiration algorithm based on MODIS and global meteorology data

The objective of this research is to develop a global remote sensing evapotranspiration (ET) algorithm based on Cleugh et al.'s [Cleugh, H.A., R. Leuning, Q. Mu, S.W. Running (2007) Regional evaporation estimates from flux tower and MODIS satellite data. Remote Sensing of Environment 106, page 285-304- 2007 (doi: 10.1016/j.rse.2006.07.007).] Penman-Monteith based ET (RS-PM). Our algorithm considers both the surface energy partitioning process and environmental constraints on ET. We use ground-based meteorological observations and remote sensing data from the MODerate Resolution Imaging Spectroradiometer (MODIS) to estimate global ET by (1) adding vapor pressure deficit and minimum air temperature constraints on stomatal conductance; (2) using leaf area index as a scalar for estimating canopy conductance; (3) replacing the Normalized Difference Vegetation Index with the Enhanced Vegetation Index thereby also changing the equation for calculation of the vegetation cover fraction (F_C); and (4) adding a calculation of soil evaporation to the previously proposed RS-PM method.We evaluate our algorithm using ET observations at 19 AmeriFlux eddy covariance flux towers. We calculated ET with both our Revised RS-PM algorithm and the RS-PM algorithm using Global Modeling and Assimilation Office (GMAO v. 4.0.0) meteorological data and compared the resulting ET estimates with observations. Results indicate that our Revised RS-PM algorithm substantially reduces the root mean square error (RMSE) of the 8-day latent heat flux (LE) averaged over the 19 towers from 64.6 W/m² (RS-PM algorithm) to 27.3 W/m² (Revised RS-PM) with tower meteorological data, and from 71.9 W/m² to 29.5 W/m² with GMAO meteorological data. The average LE bias of the tower-driven LE estimates to the LE observations changed from 39.9 W/m² to − 5.8 W/m² and from 48.2 W/m² to − 1.3 W/m² driven by GMAO data. The correlation coefficients increased slightly from 0.70 to 0.76 with the use of tower meteorological data. We then apply our Revised RS-PM algorithm to the globe using 0.05° MODIS remote sensing data and reanalysis meteorological data to obtain the annual global ET (MODIS ET) for 2001. As expected, the spatial pattern of the MODIS ET agrees well with that of the MODIS global terrestrial gross and net primary production (MOD17 GPP/NPP), with the highest ET over tropical forests and the lowest ET values in dry areas with short growing seasons. This MODIS ET product provides critical information on the regional and global water cycle and resulting environment changes.     

Reliable estimation of evapotranspiration on agricultural fields predicted by the Priestley–Taylor model using soil moisture data from ground and remote sensing observations compared with the Common Land Model

Evapotranspiration (ET) is a crucial factor in understanding the hydrological cycle and is essential to many applications in hydrology, ecology and water resources management. However, reliable ET measurements and predictions for a range of temporal and spatial scales are difficult. This study focused on the comparison of ET estimates using a relatively simple model, the Priestley–Taylor (P-T) approach, and the physically based Common Land Model (CLM) using ground and remotely sensed soil moisture data as input. The results from both models were compared directly with hourly eddy covariance measurements at two agricultural field sites during the Soil Moisture–Atmosphere Coupling Experiment (SMACEX) in the corn soybean production region in the Upper Midwest, USA. The P-T model showed a significant overestimation of the potential ET compared to the measurements, with a root mean square error (RMSE) between 115 and 130 W m^–2. Actual ET was better predicted by the CLM, with the RMSE ranging between 50 and 75 W m^–2. However, actual ET from the P-T model constrained with a soil moisture dependency parameterization showed improved results when compared to the measurements, with a significantly reduced bias and RMSE values between 60 and 65 W m^–2. This study suggests that even with a simple semi-empirical ET model, similar performance in estimating actual ET for agricultural crops compared to more complex land surface–atmosphere models (i.e. the CLM) can be achieved when constrained with the soil moisture function. This suggests that remote sensing soil moisture estimates from the Advanced Microwave Scanning Radiometer – Earth Observing System (AMSR-E) and others such as the Soil Moisture and Ocean Salinity (SMOS) mission may be effective alternatives under certain environmental conditions for estimating actual ET of agricultural crops using a fairly simple algorithm.     

Evaluation of MODIS surrogates for meteorological humidity data in east Africa

Satellite remote-sensing technology has shown promising results in characterizing the environment in which plants and animals thrive. Scientists, biologists, and epidemiologists are adopting remotely sensed imagery to compensate for the paucity of weather information measured by weather stations. With measured humidity from three stations as baselines, our study reveals that normalized difference vegetation index (NDVI) and atmospheric saturation deficit at the 780 hPa pressure level (D_MODIS), both of which were derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor, were significantly correlated with station saturation deficits (D_stn) (|r| = 0.42–0.63, p < 0.001). These metrics have the potential to estimate saturation deficits over east Africa. Four to nine days of lags were found in the NDVI responding to D_stn. For the daily estimation of D_stn, D_MODIS yielded better performance than the NDVI. However, both of them poorly explained variation in daily D_stn using simple regression models (adj. R² = 0.17–0.39). When the estimation temporal scale was changed to 16 days, performance was similar, and both were better than daily estimations. For D_stn estimation at coarser geographic scales, given that many factors such as soil, vegetation, slope, aspect, and wind speed might complicate NDVI response lags and model construction, D_MODIS is preferable as a proxy to saturation deficit over ground due to its simple relationship with D_stn.     

Spatial–temporal dynamics of gross primary productivity,evapotranspiration, and water-use efficiency in the terrestrial ecosystems of the Yangtze River Delta region and their relations to climatic variables

Moderate Resolution Imaging Spectroradiometer (MODIS) products and climate data collected from meteorological stations were used to characterize the spatial–temporal dynamics of gross primary productivity (GPP), evapotranspiration (ET), and water-use efficiency (WUE) in the Yangtze River Delta (YRD) region and the response of these three variables to meteorological factors. The seasonal patterns of GPP and WUE showed a bimodal distribution, with their peak values occurring in May and August, and April and October, respectively. By contrast, the seasonal variation of ET presented a unimodal pattern with its maximum in July or August. The spatial distribution of ET and GPP was similar to higher values occurring in the south. From 2001 to 2012, GPP in the eastern YRD decreased, while GPP in the western part increased. In comparison, over the 12 years, ET in the northern part of YRD decreased, while ET in the southern part increased. The spatial distribution and spatial variation of WUE were both similar to those of GPP. This implies that the changes in WUE are primarily controlled by the variations in GPP. The annual average WUE over vegetation types followed the order of: evergreen broadleaf forest (1.95 g C kg^?1 H₂O) > deciduous broadleaf forest (1.87 g C kg^?1 H₂O) > evergreen needle leaf forest (1.70 g C kg^?1 H₂O) > deciduous needle leaf forest (1.68 g C kg^?1 H₂O) > grassland (1.66 g C kg^?1 H₂O) > cropland (1.61 g C kg^?1 H₂O). Both GPP and ET increased with increasing annual mean temperature (T_a) and annual mean precipitation across all of the plant function types. WUE decreased as vapour pressure deficit (VPD) increased in all of the biomes. Interestingly, the relationship between WUE and VPD was the most significant in broadleaf forest. Whether this phenomenon is universal should be investigated in future studies.     

Simplifying the revised three-temperature model for remotely estimating regional evapotranspiration and its application to a semi-arid steppe

Remote sensing is a feasible and economical way of mapping the spatial distribution of evapotranspiration (ET) at scales ranging from the regional to the global level. The revised three-temperature (3T-R) model, including land-surface temperature, reference temperature, and air temperature, is a promising approach for ET mapping, but aerodynamic resistance must be included in the model. The objectives of this study are to: (1) propose a simpler remote-sensing algorithm for estimating reference temperatures in the 3T-R model that would not depend on aerodynamic resistance; and (2) test the performance of the simplified 3T-R (3T-S) model. Assuming a pixel with the maximum surface temperature is a site without evaporation or transpiration, a method was proposed to replace the reference temperatures with paired maximum temperatures for soil and canopy, collected from a region with approximately equivalent solar radiation and terrain. A case study was conducted using Landsat Thematic Mapper (TM) images and synchronous ground observations performed in an Ecosystem Observation Station in northern China during the 2009 growing season. The results indicated that: (1) daily ET (ET_{3T_S}) estimated using the 3T-S model was close to the observational data, with a mean absolute error (MAE) of 0.24 mm d^?1 and a mean absolute percentage error (MAPE) of 9.78%; (2) the 3T-S model was much simpler than the 3T-R model with respect to its calculation procedures and data requirements, suggesting it might result in less error propagation, because the MAE and MAPE between ET_{3T_R} (daily ET estimated using the 3T-R model) and the observational data were 0.36 mm d^?1 and 14.71%, respectively. Therefore, it is concluded that the 3T-S model is a simpler method, but also an accurate way of estimating regional ET.     

On the net surface water exchange rate estimated from remote-sensing observation and reanalysis

This study compares the net surface water exchange rates, or surface precipitation (P) minus evapotranspiration (ET), and atmospheric water vapour sinks calculated from various observations and reanalyses, and investigates whether they are physically consistent. We use the observed precipitation from the Global Precipitation Climatology Project (GPCP) and the Tropical Rainfall Measuring Mission (TRMM) 3B43, ocean evaporation from Goddard Satellite-based Surface Turbulent Fluxes Version 2c (GSSTF2c), and land ET from the Moderate Resolution Imaging Spectroradiometer (MODIS) global ET project (MOD16) and PT-JPL products to calculate observed P minus observed ET. P–ET is also obtained from atmospheric water vapour sink calculated using Atmospheric Infrared Sounder (AIRS)/Advanced Microwave Sounding Unit observation specific humidity observation and wind fields from the Modern-Era Retrospective Analysis for Research and Applications (MERRA) and ERA-interim, denoted as AIRS_M and AIRS_E, respectively. MERRA and ERA-interim water vapour budgets are also calculated for cross-comparison and consistency check. The period of study is between 2003 and 2006 based on the availability of all of the data sets. Averaged water vapour sinks from AIRS and reanalysis are consistent over the global ocean and are close to zero (range: 0.02–0.06 mm day^?1), but range between 0.14 and 0.23 mm day^?1 when land is included. Over ocean within 50^oS--50^oN, averaged observed P minus observed evaporation shows a much larger negative number than that obtained from AIRS and reanalysis. The differences mainly occur over subtropical oceans, especially in the southern hemisphere in summer and the northern hemisphere in winter. Over land, generally higher agreement between observed P minus observed ET and atmospheric water vapour sinks (calculated from AIRS and reanalysis) is found. However, large regional differences, often with strong seasonal dependence, are also observed over land. Estimates of atmospheric water vapour sinks are influenced by both winds and biases in water vapour data, especially over tropics and subtropical oceans, thereby calling for the need for further investigations and consistency checks of satellite-based and reanalysis water vapour, reanalysis winds, P observations, and surface evaporation estimates. In higher latitudes, atmospheric water vapour sinks calculated from AIRS_M, AIRS_E, MERRA, and ERA-interim are more consistent with each other.     

Satellite-derived UV irradiance for a region with complex morphology and meteorology: comparison against ground measurements in Santiago de Chile

Ground-based measurements of ultraviolet (UV) irradiance, carried out by a four-channel UV radiometer in Santiago de Chile from October 2004 to December 2011, have been used to estimate daily values of the UV index (UVI). These ground-based data have been compared with UVI estimates retrieved from the Ozone Measurement Instrument (OMI) on board the Aura spacecraft. Since the widely used OMI-gridded UVI data may not be suitable for the complex local morphology and meteorology, a careful screening of overpass OMI data was applied.

Nevertheless, we found that OMI-derived UVI data overestimate ground-based values; depending on cloud-cover conditions, the mean bias (MB) and the root mean square error (RMSE) range from 34.53% to 30.29% and from 35.22% to 43.50%, respectively, with the lowest MB (and the highest RMSE) values occurring under overcast conditions. Moreover, the difference between satellite-derived and ground-based UVI data exhibits a limited seasonality with somewhat larger differences in the fall season. The detected overestimation seems to be linked with the boundary layer aerosol absorption that is not accounted for by the OMI algorithm. Indeed, we found that the difference in UVI increases with the aerosol concentration (which in Santiago shows seasonal variations). Ceilometer profiles of backscatter intensities, directly related to aerosol concentrations, and PM10 concentrations correlate with UVI differences (correlation coefficient r of approximately 0.6 and 0.4, respectively) under cloud-free conditions for time scales ranging from months to years.

Additional comparisons were performed between UVI estimates retrieved from our ground-based measurements in Santiago and from the Tropospheric Emission Monitoring Internet Service (TEMIS) Scanning Imaging Absorption Spectrometer for Atmospheric Cartography (SCIAMACHY). Under cloudless conditions, also TEMIS-derived data overestimate ground-based UVI estimations (by about 31%) and exhibit a small seasonality.     

Estimating daily maximum air temperature from MODIS in British Columbia,Canada

Air temperature (T_a) is an important climatological variable for forest research and management. Due to the low density and uneven distribution of weather stations, traditional ground-based observations cannot accurately capture the spatial distribution of T_a, especially in mountainous areas with complex terrain and high local variability. In this paper, the daily maximum T_a in British Columbia, Canada was estimated by satellite remote sensing. Aqua MODIS (Moderate Resolution Imaging Spectroradiometer) data and meteorological data for the summer period (June to August) from 2003 to 2012 were collected to estimate T_a. Nine environmental variables (land surface temperature (LST), normalized difference vegetation index (NDVI), modified normalized difference water index (MNDWI), latitude, longitude, distance to ocean, altitude, albedo, and solar radiation) were selected as predictors. Analysis of the relationship between observed T_a and spatially averaged remotely sensed LST indicated that 7 × 7 pixel size was the optimal window size for statistical models estimating T_a from MODIS data. Two statistical methods (linear regression and random forest) were used to estimate maximum T_a, and their performances were validated with station-by-station cross-validation. Results indicated that the random forest model achieved better accuracy (mean absolute error, MAE = 2.02°C, R² = 0.74) than the linear regression model (MAE = 2.41°C, R² = 0.64). Based on the random forest model at 7 × 7 pixel size, daily maximum T_a at a resolution of 1 km in British Columbia in the summer of 2003–2012 was derived, and the spatial distribution of summer T_a in this area was discussed. The satisfactory results suggest that this modelling approach is appropriate for estimating air temperature in mountainous regions with complex terrain.