Wenjing Lin

Institute of Hydrogeology and Environmental Geology, Shijiazhuang, P.R.China

Z. Su, Rogier van der Velde, Xin Shan

International Institute for Geo-information Science and Earth Observation (ITC), Enschede, Netherlands

Abstract:  Evapotranspiration is one of the most significant components of the hydrologic budget. Conventional techniques that based on the point measurements are representative only of local scales and will fail for large scales. Satellite sensors that observe the earth from the space give a chance to estimate evapotranspiration in a big scale. The Surface Energy Balance System (SEBS) model was developed to estimate land surface fluxes using remotely sensed data and available meteorological observations. It has the most important advantage of its inclusion of the a physical model for the estimation of the roughness height for heat transfer which is the most critical parameter in the parameterization of the heat fluxes of land surface. In this paper, SEBS has been utilized to estimate the surface fluxes over Hebei Plain in Northeastern China by using MODIS/TERRA images, in combination of meteorological data collected in meteorological stations distributed over the study area. The estimated daily evapotranspiration by SEBS are first compared with measurements by large weighing lysimeter in Luancheng Agro-Ecosystem Station (LAES) located near Shijiazhuang city. The comparisons show that the estimated evapotranspiration from SEBS have a good agreement with the ground truth data. Based on the validation of the model, the spatial-temporal distributions of actual evapotranspiration were analyzed in combination of the up-to-date land cover map in Hebei Plain.

Keywords: Evapotranspiration, SEBS, MODIS/TERRA, Hebei Plain

 

1 Introduction

Apart from precipitation, the most significant component of the hydrologic budget is evapotranspiration. Evapotranspiration varies regionally and seasonally according to ambient environmental conditions, such as climate condition, land cover, land use, soil moisture, and available radiation etc. Because of this variability, research for integrate water resources modelling, dynamic crop-weather modelling and drought monitoring, a thorough understanding of the evapotranspiration process and knowledge about the spatial evapotranspiration is needed.

In the last few decades the theoretical and applied analysis of evapotranspiration and its components transpiration and evaporation have received much attentions. A physically based equation for potential evapotranspiration (ET₀) was derived by Penman by combining energy balance equation with the aerodynamic equation for vapour transfer(Penman 1947; Penman 1956). It was subsequently modified by Monteith to include a canopy resistance for vapour diffusion out of stomata. Apart from above mentioned principles, there are many other methods that have been proposed for estimating ET₀. By comparing 20 different methods of estimating ET0, Jensen et al. showed that the Penman-Monteith equation provide the best accurate estimate of evaporation from well-watered grass or alfalfa under varied climate conditions(Jensen, Burman et al., 1990). However, these conventional techniques are based on the point measurements and are representative only of local scales and will fail for large scales because of the dynamic nature and regional variation of ET. On the other hands, Penman-Monteith equation provided only an estimation of the potential ET, which limits its use in practical applications.

Evaporation of water requires relatively large amounts of energy, either in the form of sensible heat or radiant energy. Therefore the evapotranspiration process is governed by energy exchange at the land surface and is limited by the amount of energy available. Because of this limitation, it is possible to predict the regional actual evapotranspiration by applying the principle of energy conservation. Recently, remote sensingtechniques have developed rapidly. From satellite observation, people can obtain consistent and frequent spectral reflectance and emittance of radiation of the land surface in a basin scale, so it is possible to estimate the regional evapotranspiration rate by combining remotely sensing data with the solar radiation observation based on surface energy balance model. In the past decades, considerable efforts have been made to gaining experience and deriving appropriate models to counter this challenge(Norman, Kustas et al., 1995; Bastiaanssen, Menenti et al., 1998; Bastiaanssen, Pelgrum et al., 1998; Su 2002). Several algorithms were developed and they all have been applied and validated in some regions(Su and Jacobs 2001; Su, Mccabe et al., 2005).

The Surface Energy Balance System (SEBS) model was developed by Su (2002) to estimate land surface fluxes using remotely sensed data and available meteorological observations. It has the most important advantage of its inclusion of the a physical model for the estimation of the roughness height for heat transfer which is the most critical parameter in the parameterization of the heat fluxes of land surface. In this paper, SEBS has been utilized firstly to estimate the surface fluxes over Hebei Plain in Northeastern China by using MODIS/TERRA image. Thereafter, spatial distributions of the daily evapotranspiration in Hebei Plain in combining with the land cover maps were analyzed.

2. Study area

The study area, Hebei Plain, is located in the north-eastern China between the range of 114°15¢E-117°45¢E and 36°N-39°40¢N (figure 1). Hebei Plain is one part of the North China Plain, and it covers an area of 62004km², which is more than 33% of the Hebei Province. It is one of the largest agricultural areas in China and also one of the most densely populated regions in the world. Groundwater resource is one of the most important natural resources in this area, because it provides drinking water to urban and rural communities, supports irrigation and industry, sustains the flow of streams and rivers, and maintains the ecosystems. The amount of water for agricultural as well as industrial use has increased tremendously from 1970s(Zhang, Shi et al., 1997). The water shortage became one of the constraint prevents further development. Investigation has shown that more than 80% of groundwater resources abstracted from aquifer were used for irrigation in this area(Zhang, Shi et al., 2005). Due to the traditional irrigation pattern, a lot of water resources were wasted directly. With the over-exploitation for more than 30 years, a series of environmental problems have occurred, such as decline of regional groundwater level, change of flow field, decrease of water resources and downward movement of saline water body(Zhang, Shi et al., 1997). The hydrogeological environments have changed in the past few decades and the future sustainability of water resources in Hebei Plain is at risk.

In order to make better use of the groundwater resources in the Hebei Plain, many studies have been concentrated on the identification of groundwater net recharge and the identification of agriculture water use(Wang, Lin et al., 2005). Several approaches have been developed to quantify groundwater net recharge. In regional studies, the water balance method is commonly used to estimate areal net recharge, which is mainly controlled by three processes: precipitation, surface runoff and actual evapotranspiration. Due to the little difference of rainfall patterns in semi-arid area and commonly available real time runoff data, main difficulty comes from the estimation of areal evapotranspiration patterns, which have large differences because of land surface diversity. On the other hand, duo to the large proportion of agriculture water use in the Hebei Plain, it is also very important to determine the spatial and temporal evapotranspiration to guide the irrigation water use. Hence, the main problem goes into accurate regional evapotranspiration estimation.

Figure 1: Location map of the study area

 

3. Methodology

3.1. Surface Energy Balance System (SEBS)

The surface energy balance is commonly written as

  

Where R_n is the net radiation, G the soil heat flux, H the sensible heat flux and  is the latent heat flux, which can be expressed as height of water, i.e. evapotranspiration.

The equation to calculate the net radiation is given by

    

Where, R_swd, R_lwd is incoming shortwave and outgoing longwave radiation respectively,  is the surface albedo,  is the emissivity of the surface,  is the Stefan-Bolzmann constant, equals to 5.67e10^-8  and T₀ the surface radiative temperature.

The equation to calculate soil heat flux is parameterized as

where Γ_c and Γ_s are empirical coefficient. These values have been determined using experimental observations, but depend also on the soil and vegetation type. For most bare soil conditions a Γ_s value of 0.315 is valid, and for vegetation often Γ_c is assumed to be 0.05. An interpolation is then performed between these limiting cases using the fractional canopy coverage, f_c, which can be determined from remote sensing data.

The sensible heat flux is calculated by solving next three equations iteratively,

                  

 

                                                       

Where, ρ is density of air [kg m^-3], C_p is the heat capacity of dry air [-], k is the Von Karman constant [= 0.4], z is the height at which the meteorological observations are made [m], u^* is the friction velocity [m s^-1], θ₀ and θ_a are the potential temperature at height z_oh and at height z [K], d₀ is the displacement height [m], z_oh and z_om are the surface roughness heights for heat and momentum transport [m], ψ_h and ψ_m are stability correction function for heat and momentum transport, , g is the accerelation due to gravity [m s^-2] and θ_v is the virtual temperature [K].  

The surface energy balance computation with the SEBS algorithm is based on the determination of the relative evaporation fraction,

                                                                                              

Where, is the relative evaporation fraction [-], the subscript “wet” and “dry” denote the wet condition and dry condition respectively, detailed definition can be found in Su (2002).

The evaporative fraction is finally given by: 

     

By assuming that the daily value of evaporative fraction is approximately equal to the instantaneous value, the daily evaporation can be determined as,

where is the actual evaporation on daily basis, λ  the latent heat of vaporization (),  the density of water () and  is the daily net radiation flux.

3.2 Parameterization of Surface bio-physical characteristics

In order to run SEBS, surface bio-physical parameters should be prepared firstly as model inputs. Those inputs can be derived from remote sensing data by applying related empirical equations.

Normalized Difference Vegetation Index (NDVI)

The normalized difference vegetation index (NDVI) as defined by,

                                              

Where, _red and _nir are reflectance measurements in MODIS channels 1 (0.620-0.670μm) and 2 (0.841-0.876μm).

This is the most commonly used vegetation descriptor from satellite imagery. The difference in reflectance is divided by the sum of the two reflectance bands. This compensates for different amounts of incoming light and produces a number between 0 and 1. The typical range of actual values is about 0.1 for bare soils to 0.9 for dense vegetation.

Vegetation fractional cover

Fractional Vegetation Cover is an important parameter that have key role in the energy exchanges at the land surface. A simple procedure to determine fractional vegetation cover is proposed by Gutman et al.as(Gutman and Ignatov 1998),

Where, NDVI_min is the NDVI for bare soil and NDVI_max for full vegetation coverage.

Leaf area index (LAI)

Leaf Area Index (LAI) is the leaf area per unit ground area, which reflects the vertical vegetation amount. The relationship proposed by Su is used(Su and Jacobs 2001), this reads

This formula is strictly only good for low vegetation since NDVI saturates at higher LAI values. However, because of limited information for the study area to support more sophisticated formulations, this equation is adopted in this study.

Surface emissivity

By analysing the relation between surface emmisivity and NDVI, an experimental relationship was obtained by Van de Griend et al. to determine surface emissivity(Van de Griend and Owe 1993),

Aerodynamic roughness height

Aerodynamic roughness height is a very important parameter in surface energy balance model, which influence greatly the turbulent characteristics near the surface where the heat fluxes originate. Aerodynamic roughness height can be estimated by a simple relationship proposed by Su(Su and Jacobs 2001),

Vegetation height and displacement height

A conversion is performed according to Brutsaert to derive vegetation height and displacement height for a given aerodynamic roughness height(Brutsaert 1982),

           

4 Available Dataset

4.1 Satellite observations

Satellite image over the Hebei Plain from March 4, 2005 (DOY 63) were used. Use is made of the surface reflectance products and surface temperature products of MODIS images, MOD 09 and MOD 11, which can be ordered and downloaded from Earth Observing System Data Gateway. In order to get whole coverage of the study area, two set of images titled “h26v05” and “h27v05” were selected. The surface albedo were estimated based on the algorithm proposed by Liang(Liang 2001; Liang, Chad et al., 2003).

4.2. Meteorological observations

The Chinese National Meteorological Centre (NMC) operates several meteorological stations in Hebei Plain on a daily basis. The dataset available from these stations include relative humidity, wind speed, air temperature at 2m height, actual vapour pressure, rainfall, sunshine hours and open water evaporation, etc. All of those measurements have been collected from 6 meteorological stations equally distributed within the study area on March 4, 2005. Pre-processing of those measurements was made to derive the variables over satellite passing time as inputs of SEBS.

4.3. Land cover

Due to the limited knowledge about the ground truth land cover over Hebei Plain, the MOD12Q1 Land Cover Product was selected in this study. Totally, 12 kinds of land cover types were recognized in study area. Among them, the dominant land cover type is crop land, and more than 91.47 % of the area is for agricultural land use. Urban and built areas are the second large land cover class in Hebei Plain, which present more than 3 percent according to the map. Grass also play an important role in study area, most of which located in the seashore area of eastern Hebei Plain. 

4.4. Field Lysimeter Measurements

The lysimeter data measured at Luancheng Agro-Ecosystem Station (LAES) were selected as ground truth evapotranspiration to validate the remotely sensed actual evapotranspiration. Note that, to keep the represent, the same agronomic practices were carried out inside and outside of the lysimeter(Liu and Wang 1999).

5 Result and Discussions

5.1 Comparison SEBS results to field measurements

The accuracy of SEBS result daily actual evapotranspiration was analyzed by comparing with the ground truth daily actual evapotranspiration and crop evapotranspiration. Two empirical methods were applied to get the crop evapotranspiration based on the routine meteorological observations on the site.

‘Kc-ET0’ approach is introduced in FAO Irrigation and Drainage Paper No.56 to calculate the crop evapotranspiration under standard conditions(Allen, Luis et al., 1998). As the main crop product area in China, the farm land of Hebei plain is under a very good management and irrigation water supply, especially in Taihang Mountain foot plain where Luancheng Agro-Ecosystem Station locates, so ‘Kc-ET0’ approach can be applied, which is given as,

Where ET₀ is the reference crop evapotranspiration [mm d^-1] determined by Penman-Monteith equation and K_c is the crop coefficient [-]. Times series of Kc values can be referred to Liu et al(Liu, Zhang et al.,2002).

Another commonly used methods to estimate the crop evapotranspiration in China is based on the pan evaporation,

Where, E₀ is the pan evaporation [mm d^-1], the crop water consumption coefficient [-] and empirical values for main crops of different months in North China plain is summarized by Han et al(Han, Zhen et al., 2004).

Figure 2 shows the comparison between the daily evapotranspiration values obtained from SEBS and from other approaches mentioned above based on the point measurements. It shows a good agreement between the SEBS modeled ETa and ground truth observations. The little difference between them can be explained by surface diversity and pixel based remote sensing techniques. In conclusion, SEBS is a useful tool to estimate evapotranspiration.

5.2. Spatial Evapotranspiration

The estimated evaporative fraction and daily evapotranspiration over study area on 4 March, 2005 are shown as figures 3 and figure 4.  Figure 5 shows their histogramsThe range of evaporative fraction is between 0.37 and 0.53. The highest values that about 0.48 to 0.53 can be found in the central plain and other area have the low values ranging from 0.37 to 0.48. The lowest values only distributed sparsely in the south-west and north part of the area.

 

a: daily actual evapotranspiration measured by lysimeter

b: daily crop evapotranspiration estimated from ET₀

c: daily crop evapotranspiration estimated from Pan evaporation

d: daily actual evapotranspiration modelled by SEBS

Figure 2: Comparison of the daily evapotranspiration between estimated by SEBS and obtained from observations

        The spatial distributions of ETa have the similar patterns. Central Hebei Plain has the highest ETa ranging form 1.46 mm day^-1 to 1.79 mm day^-1, but some lower values that below 1.45 mm day^-1 can be seen clearly along the dry river bed to the north of Shijiazhuang City. The lower ETa values that below 1.3 mm day^-1 are sprinkled on the north part and south part the plain. The statistical characteristics and histogram of ETa over each land cover types are shown as table 1 and figure 5.

In early March, the highest ETa values that up to 2.34 mm day^-1 are found from the open water surface in the eastern seashore area and lakes in the plain. Although a high ETa values for croplands were found in the ETa maps, which mostly situated in the central plain, statistically lower mean values were obtained due to the large area it covers with different climatic conditions and variable actual evapotranspiration.High Eta values in barren orsparsely vegetated land in dicates the abundant soil moisture under giound in this season.

According to the distribution patterns in ETa histogram and daily ETa statistics over each land cover, the highest ETa were presented in the free surface water body, the mean of which is about 1.87 mm day^-1. Surface evapotranspiration is mainly controlled by available soil moisture and incoming solar radiation as well as other environmental aspects. When under the same environmental conditions, for example, same surface wind speeds, radiation, air temperature etc.highest ETa values is expected in open water surface with abundant water for evaporate. The lowest ETa values were found in crop land, which is only about 1.24 mm day^-1 as mean. This is contradicting to the well irrigation management status over the farm land of Hebei Plain. However, due to large coverage of the study area, high spatial variety is possible not only for the local climatic conditions but also the irrigation performance in different area, so a lower mean ETa values is reasonable and explainable.

Figure 3: Evaporative fraction [-] over Hebei Plain on March 4, 2005

 

 

Figure 4: Daily evapotranspiration over Hebei Plain on March 4, 2005

      

Figure 5: Histogram of evaporative fraction and evapotranspiration over Hebei Plain on March 4, 2005

Table 1: The statistics over each land cover classes in study area on 4 March, 2005 (mmday^-1)

Figure 6: Histogram of daily ETa over different land cover types in Hebei Plain on 4 March, 2004

        The statistics reflects that the urban and built-up area also have a certain amount of evapotranspiration. This is contrary to the knowledge that the residential areas should have lower ETa. This can be explained by the pixel based remote sensing techniques. The pixels of urban area includes not only the construction but also the water body, street trees and grass parcels, which all have very high evapotranspiration. Therefore, the estimated ETa in urban area represented the mixed effection of all these things.

6 Conclusion and recommendation

In this study, the SEBS model has been evaluated at local scales using in situ measurements and meteorological observations. Results indicate that daily evapotranspiration predictions form SEBS perform very well when assessed against in situ actual evapotranspiration derived from lysimeter measurements and crop evapotranspiration derived from empirical equations based on the meteorological observations. Without doubt, due to the pixel by pixel based remote sensing techniques, some error must occur, especially, when the coarse resolution images were used.

Based on the accuracy analysis of SEBS model, spatial distribution of daily actual evapotranspiration over Hebei Plain on March 4, 2004 were determined and discussed, in combination of the up-to-date land cover types in study area.  Results shows that Central Hebei Plain has the highest ETa ranging form 1.46 mm day^-1 to 1.79 mm day^-1, but some lower values that below 1.45 mm day^-1 can be seen clearly along the dry river bed. The lower ETa values that below 1.3 mm day^-1 are sprinkled on the north part and south part the plain. When taking land cover types into account, as expected, open water surface have the highest ETa over the study area, and due to high surface variety in Hebei Plain, the lowest values were found in the crop land with a mean values 1.24 mm day^-1.

This study could re-establish the fact that, the application of remote sensing brings a significant contribution to estimate the spatial evapotranspiration in regional scale for all types of land covers. However, the ground observation based methods are always very important, especially in verifying the results of different remote sensing based approaches. In this study, the lysimeter data in Luancheng Agro-Ecosystem Station is vital in this regard.

It is acknowledged that evapotranspiration is computed not for its own sake but for other purpose, regional water resources evaluation and management, irrigation performance assessment, as well as global climate change etc. for example. In this regard, further consideration should be addressed according to the fields it applied. Take the regional water balance research as an example, how to get regional scale actual evapotranspiration on cloudy days is challenging.

Reference

[1]      Allen, R. G., S. P. Luis, et al. (1998). FAO Irrigation and Drainage Paper No. 56: Crop evapotranspiration--guidelines for computing crop water requirements. Rome, FAO.

[2]      Bastiaanssen, W. G. M., M. Menenti, et al. (1998). "A remote sensing surface energy balance algorithm for land (SEBAL) 1. Formulation." Journal of Hydrology 212-213: pp.198-212.

[3]      Bastiaanssen, W. G. M., H. Pelgrum, et al. (1998). "A remote sensing surface energy balance algorithm for land (SEBAL) 2. Validation." Journal of Hydrology 212-213: pp.213-229.

[4]      Brutsaert, W. (1982). Evaporation into the atmosphere. Reidel, Dordrecht: 299.p.

[5]      Gutman, G. and A. Ignatov (1998). "The derivation of the green vegetation fraction from NOAA/AVHRR data for use in numerical weather prediction models." International Journal of Remote Sensing 19(8): pp.1533-1543.

[6]      Han, S. M., L. Zhen, et al. (2004). "Comparing Two Methods for Calculating Major Crop Evapotranspiration in the Pediment of Mt. Taihang." Journal of Irrigation and Drainage 23(6): pp.66-68 (in Chinese).

[7]      Jensen, M. E., R. D. Burman, et al. (1990). Evapotranspiration and Irrigation Water Requirements. ASCE Manuals and Reports on Engineering Practices. New York, American Society of Civil Engineers. 70:pp.332.

[8]      Liang, S. (2001). "Narrowband to broadband conversions of land surface albedo I: Algorithms." Remote Sensing of Environment 76(2): pp.213-238.

[9]      Liang, S., J. S. Chad, et al. (2003). "Narrowband to broadband conversions of land surface albedo: II. Validation." Remote Sensing of Environment 84(1): pp.25-41.

[10]  Liu, C. M. and H. Wang (1999). The Interface Processes of Water Movement in the Soil-Crop-Atmosphere System and Water-saving Regulation. Beijing, Science Press.

[11]  Liu, C. M., X. Y. Zhang, et al. (2002). "Determination of daily evaporation and evapotranspiration of winter wheat and maize by large-scale weighing lysimeter and micro-lysimeter." Agricultural and Forest Meteorology 111: pp.109-120.

[12]  Norman, J. M., W. P. Kustas, et al. (1995). "A two-source approach for estimating soil and vegetation energy fluxes from observations of directional radiometric surface temperature." Agricultural and Forest Meteorology 77: pp.263-293.

[13]  Penman, H. L. (1947). Natural evaporation from open water, bare soil and grass. Proceedings Royal Society of London, Series A. Mathamatical and Physical Sciences, London.

[14]  Penman, H. L. (1956). "Estimating Evaporation." Transactions of the American Geophysical Union 37(1): pp.43-50.

[15]  Su, H., M. F. Mccabe, et al. (2005). "Modeling evapotranspiration during SMACEX : comparing two approaches for local- and regional-scale prediction." Journal of hydrometeorology 6: pp.910-922.

[16]  Su, Z. (2002). "The Surface Energy Balance System (SEBS) for estimation of turbulent heat fluexs." Hydrology and Earth System Sciences 6: pp.85-99.

[17]  Su, Z. and C. Jacobs (2001). ENVISAT: actual evaporation. BCRS Report 2001 : USP-2 Report 2001  01-02. Delft, Beleidscommissie Remote Sensing (BCRS): 57.p.

[18]  Van de Griend, A. A. and M. Owe (1993). "On the relationship between thermal emissivity and the Normalized Difference Vegetation Index for natural surfaces." International Journal of remote sensing 14(6): pp.1119-1131.

[19]  Wang, G. L., W. J. Lin, et al. (2005). "Numerical modeling on groundwater table variation responding to water saving agriculture." Journal of Hydraulic Engineering 36(3): pp. 1-7  (in chinese with english abstract).

[20]  Zhang, Z. H., D. Shi, et al. (1997). "Evolution of Quaternary groundwater system in North China Plain." Science in China Series D (Earth Sciences) 40(3): pp.276-783 (in Chinese with English abstract).

[21]  Zhang, Z. H., D. Shi, et al. (1997). "Evolution and development of groundwater environment in North China Plain under human activities." Acta Geoscientica Sinica 18(4): pp.337-344 (in Chinese with English abstract).

[22]  Zhang, Z. H., J. S. Shi, et al. (2005). Water-saving agriculture and sustainable groundwater use in Hebei Plain. Science Report, Institute of Hydrogeology and Environmental Geology, CAGS: p.302 (in Chinese).