Simulation of surface radiation balance on the Tibetan Plateau



[1] Tibetan Plateau is vulnerable to climate changes. At the same time, an altered surface radiation balance on the Plateau due to its rapid development would contribute to feedback and causes of global climate changes, particularly the Asian summer monsoon. In this study, we investigated surface radiation balance on the Tibetan Plateau by comparative analysis of two land surface models with the in-situ observation data. Our scrutiny reveals that radiative coupling, which has been neglected over low elevation region, is important in simulating surface energy balance on the Tibetan Plateau. For a proper assessment of surface radiation balance and hydrologic cycle on the Plateau, surface albedo and emissivity should be also correctly incorporated into land surface models and remote sensing algorithm.

1. Introduction

[2] Land-atmosphere interactions on the Tibetan Plateau, the roof-of-the-world, impact on energy and water cycles on both regional and global scales. In turn, environment in the Plateau is vulnerable to climate changes. Accordingly, the Plateau has been the climate research subject for several decades. The Tibetan Plateau is characterized by high elevation (>4000 m) and radiative fluxes show different magnitudes and patterns compared to those of the low elevation region [Smith and Shi, 1995; Choi et al., 2004]. Such unique environments gives us a opportunities to test structural adequacy in the model. To properly assess the surface radiation balance on this important area, the present study tested two representative land surface models. We then investigated the source of errors in modeling radiative fluxes.

2. Materials and Methods

2.1. Field Observation

[3] Through GEWEX Asian Monsoon Experiment (GAME)-Tibet, intensive field observation has been conducted to monitor and understand the energy and water cycles on regional scale in the central Tibetan Plateau since 1998. The present study used the data at BJ site in Naqu (31.37°N; 91.90°E, 4580 m above m.s.l.) from May 30 to September 14, 1998. The site is flat with fetch over 1 km depending on wind direction. Soil surface was sparsely covered with short grass with canopy height of <0.05 m and leaf area index (LAI) <0.5. For measuring surface radiative fluxes, a CNR1 (Kipp and Zonnen, Netherlands) and a Q7 (Campbell Scientific Inc., USA) net radiometers was also installed at 2.85 m above the ground with eddy-covariance system. More detailed information on the field observation can be found in Choi et al. [2004] and Hong et al. [2004].

2.2. Model Description

[4] We ran two land surface models (LSMs) (SiB2 and Noah LSM) for the simulation of water and energy cycles on the Plateau in this study. SiB2 incorporates mechanistic photosynthesis-conductance models. Albedos of soil and vegetation depends on soil and vegetation properties as well as on wavelength of incident radiation in SiB2. More details are given by Sellers et al. [1996].

[5] Noah LSM is a community model of NCEP (National Centers for Environmental Prediction), Oregon State University, the US-Air Force, and the Hydrologic Research Lab in the US. Albedo should be prescribed by users on a monthly basis before running Noah LSM. Detailed information for Noah LSM can be found in the website of To drive two models, the observed downward shortwave radiation was used with other observed meteorological variables (e.g., wind and humidity).

3. Observed Properties of Surface Radiation Balance

[6] Figure 1 shows temporal variation of the observed surface radiative fluxes. The surface radiative balance on the Tibetan Plateau has several unique properties: 1) Upward longwave radiations (Rlup) is larger compared to that of the sites over low elevation regions especially before monsoon. On the Plateau, the daytime surface temperature is much greater than the temperature of air column over the ground surface due to the strong downward solar radiation, high elevation and sparse canopy. Net longwave radiation sometimes exceeds 400 Wm−2, which is an order of magnitude larger than that of low altitude region; 2) instantaneous downward shortwave radiation (Rsdn) sometimes exceeds the solar constant because of strong diffusive radiation and reflection from towering clouds [Reiter et al., 1987; Smith and Shi, 1995; Choi et al., 2004; Gu et al., 2005]; and 3) surface albedo (α) decreased by 30% from pre-monsoon (0.2) to monsoon (0.15). Consequently, the net radiation (Rn) increased by 20%.

Figure 1.

Temporal variation of surface radiative fluxes. Dot is a half hourly value, and solid line is a 10-day moving average.

4. Results and Discussion

4.1. Emissivity

[7] For a given downward radiation, both LSMs substantially underestimated Rn mainly due to the positive bias of modeled upward longwave radiation (Figure 2). Like other land surface models, both LSMs calculate upward longwave radiation (Rlup) from the surface temperature (Ts) using the Stefan-Boltzman's law. Thus the modification of surface emissivity (equation image) directly influences Rlup. In model comparison, the uncertainty in emissivity is important [Chen et al., 1997] and it is worth noting that two models assume the black-body radiation (i.e., equation image = 1).

Figure 2.

The observed and simulated radiative fluxes.

[8] In arid regions with sparse vegetation like the Plateau, Schmugge et al. [2001] found that emissivity was less than 0.8 for the quartz rich soil in 8 ∼ 9.5 μm channels and about 0.96 in the longest wavelength channel. Over the Plateau the magnitude of radiative fluxes is large and thus small bias of surface emissivity produced the inaccurate simulation of the surface energy partitioning.

[9] Furthermore, we found that the two LSMs did not adopt the exact condition of a non-black body surface. The exact radiative balance equation is given to:

equation image

The last term in equation (1) explains the reflected longwave radiation [Molion, 1987]. Using the definition of albedo α and equation image and the Stefan-Boltzman law, we can rewrite the equation (1) to:

equation image

Where σ is the Stefan-Boltzmann constant (5.670 × 10−8 WK−4m−2. If we set equation image = 1, the last term on the right hand side of equation (2) disappears. Therefore, the bias due to the assumption of black body depends on the difference between the air and surface temperature as well as on the deviation of equation image from unity:

equation image

Equation (3) says that the bias due to the black-body assumption is proportional to 1 − equation image and temperature difference between the surface and the atmosphere. There was remarkably large net long wave radiation (σTs4Rldn ∼300 Wm−2) on the Plateau before monsoon season. In such a condition, small errors in emissivity would result in sizable bias in Rn.

[10] SiB2 does not contain the last term of the right hand side of equation (3) and Noah LSM also does not capture the non-black body surface. Default surface emissivity in the two models is constant as 1. Sensitivity analysis, reducing soil emissivity to 0.9, revealed that SiB2 showed better agreement of Rlup to the observed Rlup and therefore Rn showed about up to 20% increase in SiB2 only by the term of equation imageσTs4.

4.2. Surface Albedo

[11] In general, SiB2 credibly simulated upward (or reflected) shortwave radiation (Rsup) except in June (Figure 2). This exception in June resulted from the bias of pre-defined albedo in SiB2, underestimating upward shortwave radiation. By canceling out the overestimation of Rlup in June, the underestimation of (Rsup) in SiB2 produced Rn comparable to the observation.

[12] Such a bias due to the inappropriate assignment of albedo did not happen in Noah LSM because albedo was determined a priori with the observed values to match the monthly averaged reflected shortwave radiation with the corresponding observed values. The monthly albedo is interpolated into daily values linearly. However, the daily interpolated albedo causes bias in the diurnal variation of reflected shortwave radiation in spite of matching the monthly averaged albedo to the observed albedo.

[13] When the solar inclination angle is large (i.e., early morning and late afternoon), surface albedo is maximal, but the absolute magnitude of downward shortwave radiation is small. Noah LSM does not consider the diurnal variation of surface albedo, thereby overestimating the reflected shortwave radiation at low solar zenith angles (daytime) but underestimating at high solar zenith angles (near sunrise and sunset). Such biases of surface albedo can explain up to ∼10 Wm−2 difference of upward shortwave radiation (Rsup) between the model and observation. For better simulation, a radiation-weighted albedo should be applied to Noah LSM so that we can reduce the bias from the fact that albedo is large in the morning and evening.

4.3. Radiative Feedback

[14] Another possible reason of the overestimated Ts is substantial role of radiative feedback on the Tibetan Plateau. Radiative coupling involves the effect on Ts of the dependence of Rlup on Ts itself [Monteith and Unsworth, 1990; Raupach, 1998]. Radiative coupling has been commonly neglected in the land-atmosphere interactions because radiative conductance is relatively smaller than aerodynamic conductance for heat in general.

[15] However, in terms of radiative coupling, the Plateau is unique compared with other area at low altitudes. As Raupach [1998] summarized, radiative coupling is expressed by radiative conductance, gr = 4equation imageσTs3/ρcp. The relative contribution of gr can be quantified by the radiative decoupling factor, p, defined as p = gh/(gh + gr) [Raupach, 1998]. Here gh is aerodynamic conductance for heat. When p = 1, radiative coupling is absent in the surface energy balance, and the effect of radiative coupling increases as p decreases to 0. Raupach [1998] showed that only over smooth surfaces, p approaches to 0 (e.g., momentum roughness length <0.01 m or gh > 0.02 m s−1) and thus radiative feedback becomes negligible.

[16] On the Plateau, the maximum gr was about 0.008 ms−1 due to the low air density and high air temperature, and the maximum gh was about 0.01 ms−1 (Figure 3). The corresponding p was about 0.4 ∼ 0.6 in daytime. Radiative coupling accordingly is not negligible on the Plateau, and it is expected that modeled Ts decreases if LSMs consider the radiative coupling.

Figure 3.

Diurnal variation of aerodynamic conductance for heat and radiative conductance.

[17] We also should note that radiative feedback impacts on surface energy partitioning through the modification of sensible and latent heat fluxes. For example, McNaughton and Jarvis [1991] showed that radiative resistance (i.e., a reciprocal of conductance ≡ 1/gh reduces a negative feedback and so increased sensitivity to stomatal control, and Paw U and Gao [1988] analyzed the error due to the neglect of radiative resistance. Raupach [1998] showed that the combination equations for the fluxes of heat (H) and water vapor (LE) overestimated not only Ts, but also LE and H, if the radiative resistance is not considered.

5. Summary and Conclusions

[18] The observed surface radiative fluxes showed distinct magnitudes and patterns compared with those of low elevation regions. In particular, net longwave radiation was much larger than those at the low elevation area and thus the Plateau was a heating source to the atmosphere. The unique environment of the Plateau challenged the simulation of surface radiative balance using the land surface models and we noted some structural deficiencies of radiative transfer schemes in the two selected biosphere models. Our scrutiny reveals that over the Tibetan Plateau:

[19] 1. small error in surface albedo and emissivity can induce substantial bias in the modeled net radiation;

[20] 2. the black-body assumption is not appropriate in modeling surface radiation balance;

[21] 3. reflectance of downward longwave radiation plays an important role in radiation balance; and

[22] 4. radiative coupling, which has been commonly neglected at low elevation regions, is critical in modeling credible sensible and latent heat fluxes as well as surface temperature.

[23] Our results also imply that radiative coupling plays an important role in the surface radiation balance in extreme conditions like droughts. Therefore, radiative coupling should be implemented not only to satisfy the scalar conservation requirement of linear flux averaging for a proper scaling, but also to accurately model the surface energy balance including evapotranspiration using the land surface models and remote sensing such as MODIS [Raupach and Finnigan, 1995; Raupach, 2001; Cleugh et al., 2007; Hwang et al., 2008].


[24] This research was supported by a grant (code: 1-8-2) from Sustainable Water Resources Research Center for 21st Century Frontier Research Program and BK21 Program of the Ministry of Education and Human Resources Management of Korea. The first author was partially supported by the Korean Research Foundation Grant funded by the Korean Government (MOEHRD) (KRF-2006-214-C00095). We appreciate constructive comments of two anonymous reviewers on this manuscript. Our thanks go to Profs. H. P. Schmid, S. Y. Hong, Y. G. Noh, and J. I. Yun for their valuable insight.