The turbulent flux parameterization schemes in the surface layer are crucial
for air pollution modeling. There have been some deficiencies in the
prediction of air pollutants by atmosphere chemical models, which is closely
related to the uncertainties of the momentum and sensible heat fluxes
calculated in the surface layer. The differences between two surface layer
schemes (Li and MM5 schemes) were discussed, and the performances of two
schemes were mainly evaluated based on the observed momentum and sensible
heat fluxes during a heavy haze episode in Jing-Jin-Ji in eastern China. The
results showed that the aerodynamic roughness length
Adequate air quality modeling relies on the accurate simulation of meteorological conditions, especially in the planetary boundary layer (PBL) (Hu et al., 2010; Cheng et al., 2012; Xie et al., 2012). The PBL is tightly coupled with the earth's surface through turbulent exchange processes. As the bottom layer of the PBL, the surface layer (SL) reflects the surface state by calculating momentum, heat, water vapor, and other fluxes, and influences the atmospheric structure through a turbulent transport process. Many studies have illustrated the important roles of meteorological factors in the SL during air pollution formation. It has been demonstrated that weak wind speed, high relative humidity (RH), and strong temperature inversion are favorable for the concentration of haze (Zhang et al., 2014; Yang et al., 2015; Liu et al., 2017; Zhong et al., 2018). Strong stable stratification and weak turbulence are mainly responsible for many haze events. The relationship between the flux and the atmospheric profile in the atmospheric surface layer is a critical factor for air pollution diffusion, especially under stable stratification conditions (Li et al., 2017). However, there are still some uncertainties in the study of the stable boundary layer due to the poor description of surface turbulent motion. The simulation study on a severe haze in eastern China by the Weather Research and Forecasting with Chemistry (WRF-Chem) model concluded that current PBL schemes have a weak ability to distinguish between haze days under stable conditions and clean days under unstable conditions (T. Li et al., 2016). Another study (Vautard et al., 2012) of mesoscale meteorological models also pointed out there was a systematic overestimation of near-surface wind speed in the stable boundary layer which should contribute to the underestimation of surface concentrations of primary pollution. In addition, atmospheric conditions in both the PBL and upper layers are highly dependent on turbulent fluxes which are computed in the SL (Ban et al., 2010). Flux parameterization in the SL plays an important role in studies of the hydrological cycle and weather prediction (Yang et al., 2001; Li, 2014). An adequate SL scheme is crucial in providing an accurate atmospheric evolution by numerical models (Jiménez et al., 2012) and hence it may introduce significant impacts on air pollution simulation.
The bulk aerodynamic formulation based on Monin–Obukhov similarity theory (hereinafter MOST; Monin and Obukhov, 1954) is usually employed to calculate surface fluxes in numerical models. Turbulent fluxes are parameterized by wind, temperature, humidity in the lowest layer in the model, and temperature and humidity at the surface. Many international scholars verified MOST using field experiments and then proposed the universal functions, the most commonly used of which is the Businger–Dyer (BD) equation (Businger, 1966; Dyer, 1967). With the development of observation technology, the coefficients in the BD equation have been further modified (Paulson, 1970; Webb, 1970; Businger et al., 1971; Dyer, 1974; Högström, 1996). In addition to the BD equation, some other schemes have been put forward and they performed better, especially for strongly stable stratification (Holtslag and De Bruin, 1988; Beljaars and Holtslag, 1991; Cheng and Brutsaert, 2005). The schemes can be divided into two types according to the computing characteristics. One type is called an iterative algorithm (Paulson, 1970; Businger et al., 1971; Dyer, 1974; Högström, 1996; Beljaars and Holtslag, 1991), and it keeps MOST completely with less approximation so that the results can be more precise. However, many more steps are necessary for it to converge, and hence the CPU time is longer, which reduces the computational efficiency of modeling (Louis, 1979; Li et al., 2014). The other one is called a non-iterative algorithm (Louis et al., 1982; Launiainen, 1995; Wang et al., 2002; Wouters et al., 2012). There is no requirement for loop iteration in the calculation due to the approximate treatment. This algorithm is much simpler and less CPU-time-consuming, but the results are based on the loss of the calculation accuracy.
A new non-iterative scheme proposed by Li et al. (2014, 2015; Li
hereinafter) speeds up effectively with a higher accuracy compared with
classic iterative computation. It is remarkable that this new scheme
has just been theoretically evaluated and it has never been applied in any
models. Haze pollution has occurred frequently in recent years in eastern China.
The concentration of PM
The definitions of momentum and sensible heat fluxes as well as the detailed algorithms of the Li and MM5 schemes are introduced in this section.
The turbulent fluxes from the ground surface are defined as follows:
Both the Li and MM5 schemes are based on bulk flux parameterization. As an
important dimensionless parameter related to the stability, the bulk
Richardson number
This new scheme employs a non-iterative algorithm to compute the surface
fluxes. Its basic idea is to parameterize the stability parameter The determination of The universal function is also a key factor in flux calculation. The form
of universal function here is adopted from Cheng and Brutsaert (2005) under
stable conditions (Eq. 8a, b) and it is adopted from Paulson (1970)
under unstable conditions (Eq. 9a, b):
In addition, the RSL effect is taken into account in the Li scheme. The
definition and influence of the RSL will also be discussed in Sect. 4.1. De
Ridder (2010) proposed the expression of
The MM5 scheme is a classic one which is widely applied in modeling
investigations (Hu et al., 2010; Wang et al., 2015a, b; Tymvios et al.,
2017). This scheme does not distinguish Strongly stable condition ( Weakly stable condition ( Neutral condition ( Unstable condition (
This scheme calculates turbulent fluxes of the momentum and sensible heat
with
Location
The observational fluxes used in this study were measured at Gucheng station
from 1 December 2016 to 9 January 2017. Gucheng station (115.40
Wind rose map at Gucheng station from 1 December 2016 to 9 January 2017.
To obtain accurate flux data, quality control has been performed for the
observational data, including (1) the elimination of the outliers and the data on
rainy days, (2) double rotation and WPL correction (Webb et al., 1980), and
(3) omission of the dataset when the wind speed is less than 0.5
The surface emissivity
The surface skin temperature at Gucheng station is calculated from the
radiation data by the following formula:
Using the observed momentum and sensible heat fluxes and the meteorological
variables including wind speed, temperature, humidity, and pressure after
quality control at Gucheng station,
The definitions and influences of the RSL on the calculation of the turbulent flux are discussed in detail in this section. The Li and MM5 schemes are tested offline and evaluated during the haze pollution from 13 to 23 December 2016.
Typical values of
The RSL is usually defined as the region where the flow is influenced by the
individual roughness elements as reflected by the spatial inhomogeneity of
the mean flow (Florens et al., 2013). In the RSL, turbulence is strongly
affected by individual roughness elements, and standard MOST is no
longer valid (Simpson et al., 1998). Therefore, it is necessary to consider
the RSL effect in the calculation of the turbulent flux, especially for rough terrain such as forest or large cities.
The relationships between
Considering the lowest level in mesoscale models is usually about 10
The relationships between
Secondly, the effects of difference between
Comparison of calculated and observed fluxes at Gucheng station
from 1 December 2016 to 9 January 2017.
Using the obtained roughness lengths and the observations, the momentum and
sensible heat fluxes were calculated by the Li and MM5 schemes. Firstly,
Variations of hourly turbulent fluxes and observed PM
There were two obvious pollution processes during this observation period,
and one occurred during 13 to 23 December 2016. Figure 7 shows the
variations of hourly observed PM
Probability distribution functions (PDFs) of the differences
between calculated fluxes (momentum fluxes: left; sensible heat fluxes:
right) using two schemes (the Li scheme: red bars; the MM5 scheme: green
bars) and for observations in different stages (
Figure 8 shows the probability distribution functions (PDFs) of the
difference between calculated fluxes (using the Li and MM5 schemes) and
observations in different stages at Gucheng station. In the whole pollution
process, for the momentum fluxes (Fig. 8a), the PDF from Li tends to cluster
in a narrower range centered by 0, and the probability within
Statistics between the Li and MM5 schemes of the calculated turbulent flux at Gucheng station.
As in Fig. 7 but for Beijing station.
Mean bias (MB), normalized mean bias (NMB), normalized mean error (NME), and
root mean square error (RMSE) were calculated to test the results of the two
schemes. Table 2 shows that the Li scheme generally estimates better than
the MM5 scheme. In the whole haze process, the Li scheme underestimates the
momentum fluxes by 3.63 % relative to the observations, while the MM5
scheme overestimates the momentum fluxes by 34.03 %. The Li and MM5 schemes underestimate the
sensible heat fluxes by 15.69 % and 50.22 %, respectively. In the
three stages, the Li scheme performs much better than the MM5 scheme in stage 1 and stage 2; especially in stage 2 when atmospheric stratification
transforms from unstable to stable conditions, the difference between the Li
and MM5 schemes is particularly significant. That is, the Li and MM5 schemes
overestimate the momentum fluxes by 7.68% and 45.56 %, respectively,
and they underestimate the sensible heat fluxes by 33.84 % and 76.88 %.
The error of Li is much less than that of MM5. In view of the
important role of atmospheric stratification in the generation and
accumulation of PM
The mean momentum and sensible heat fluxes calculated using
two schemes (
Based on the good behavior of the Li scheme in Gucheng, the same experiment
was performed at Beijing station to discuss the effect of different
land-cover types on flux calculation. For Beijing station, the assumption
To quantify the difference between the two schemes, a relative difference is
defined as a percentage:
We further estimated the surface fluxes in the whole Jing-Jin-Ji region using
the two schemes. Figure 10 shows the mean momentum and sensible heat fluxes
calculated by the Li and MM5 schemes and their differences in Jing-Jin-Ji during
the pollution episode. The assumption (
Using the observed momentum and sensible heat fluxes, together with
conventional meteorological data including pressure, temperature, humidity, and wind speed from 1 December 2016 to 9 January 2017, including a severe
pollution episode from 13 to 23 December 2016, the differences between the
Li and MM5 schemes and the specific performances of the two were discussed
and evaluated in this paper. The evolution process of atmospheric
stratification from unstable to stable conditions corresponding to PM The effect of The Li scheme generally performed better than the MM5 scheme in the
calculation of both the momentum flux and the sensible heat flux at Gucheng
station. The Li scheme described atmospheric
stratification, which is closely related to haze pollution, better compared
with the MM5 scheme. This advantage was the most prominent in the transition
stage from unstable to stable atmospheric stratification, corresponding to
the PM
The offline study of the two SL schemes in this paper showed the superiority
of the Li scheme for surface flux calculation corresponding to the
PM
Data used in this paper can be provided by Yue Peng (nuist_py@163.com) upon request.
HW and YP conducted the study design. YL and CL provided the Li scheme and the flux data. CL helped with data processing. YP wrote the manuscript with the help of HW and TZ. XZ, ZG, TJ, HC, and MZ were involved in the scientific interpretation and discussion. All the authors commented on the paper.
The authors declare that they have no conflict of interest.
This article is part of the special issue “Regional transport and transformation of air pollution in eastern China”. It is not associated with a conference.
The study was supported by the National Key Project (2016YFC0203306, 2016YFC0203304), the National (Key) Basic Research and Development (973) Program of China (2014CB441201), the National Natural Science Foundation of China (41505004, 41675009), and Jiangsu Provincial Natural Science Fund Project (BK20150910). Edited by: Zhanqing Li Reviewed by: three anonymous referees