Evaluation of Summer Monsoon Clouds over the Tibetan Plateau Simulated in the ACCESS Model Using Sat

HTML

--> --> -->

1. Introduction

The dynamic and thermodynamic impacts of the Tibetan Plateau (TP) on atmospheric circulation have been investigated by many scientists (Maussion et al., 2011; Luo et al., 2011; Duan et al., 2013; Wang et al., 2015; Wu et al., 2015). In these studies, the important role of the cloud systems over the TP on weather and climate around the TP has been extensively investigated in the past several decades, but the particular characteristics of these cloud systems are not well understood, due to the complex topography and lack of sufficient observations over the TP. Remote sensing data from satellites have greatly enriched our knowledge of clouds over the TP since the 1970s (Yan et al., 2016). Recent satellite-retrieved products from CloudSat and CALIPSO (Marchand et al., 2008) have provided information on the cloud microphysical properties and their vertical distributions, which can be used to validate model cloud properties (Bodas-Salcedo et al., 2008; Stein et al., 2015). Through analysis of these satellite data, (Luo et al., 2011) found that the deep convection on the TP is shallower, less frequent, and embedded in smaller-size convection systems than that along the southern slope of the TP and the regions of the southern Asia summer monsoon. They also found that the cloud-top height over the TP is lower than that of the southern slope region of the TP. A similar study was also conducted by (Yan et al., 2016). They found that the frequency of a single-layer cloud over the TP was higher than those over the southern slope of the TP and the tropical ocean. The thickness of the single-layer cloud over the TP was comparable with that over the tropical ocean, while that over the southern slope region was much thicker, due to abundant moisture supply and strong upward motion (Yan et al., 2016).
There have been many numerical modelling studies that have examined the effects of the TP on the Asian summer monsoon (e.g., Wu and Zhang, 1998; Liu et al., 2007; IPCC, 2007; Annamalai et al., 2007). (Duan et al., 2013) analyzed the TP summer monsoon using 15 CGCMs and eight AGCMs from CMIP5. They showed that GCMs are able to successfully simulate the climatological-mean monsoon over the TP, but large biases exist in the summer precipitation. Increasing the model spatial resolution and incorporating the indirect effects of sulfate aerosol helped to improve the model performance. Using ECHAM5 coupled with the ocean model developed by Max Planck Institute for Meteorology and atmosphere-only ECHAM5 models, (Fallah et al., 2016) showed that the large-scale patterns of the Asian summer monsoon change significantly after removing the TP. (Maussion et al., 2011) used the WRF model with different resolutions to evaluate precipitation over the TP and found that the simulated precipitation with higher spatial resolution was generally better than with lower resolution, but the precipitation bias due to the effects of the orography remains unsolved.
It can be seen from the literature that many studies have focused on the dynamic and thermodynamic effects of the TP on large-scale circulations, particularly those of the Asian monsoon. However, research activities comparing in detail the modeled and observed cloud properties over the TP seem to be insufficient. Such a study is important for improving the model physics and reducing the simulation biases in this region. For this reason, this study analyzes the cloud and radiative properties in the Australian Community Climate and Earth System Simulator (ACCESS) model and compares the simulation with the same properties derived from satellite observations. To ensure a consistent comparison between model output and satellite observation, the software of the Cloud Feedback Model Intercomparison Project Observation Simulator Package (COSP; Bodas-Salcedo et al., 2011) has been implemented in the ACCESS model (Franklin et al., 2013a). Thus, the modelled and satellite-retrieved cloud properties are determined in the same manner.
The paper is organized as follows: Section 2 introduces the data and model. Section 3 compares clouds over the TP between observations and simulations in boreal summer. Section 4 analyzes the dynamic and thermodynamic fields over the TP, as well as the surface energy budget and precipitation, to investigate the possible reasons for the differences between the model and observations. The last section discusses the key points of the study and draws conclusions.

2. Data and model

2.1. Cloud data from CloudSat and CALIPSO

-->

2.1. Cloud data from CloudSat and CALIPSO

Observational COSP datasets are used in this paper to validate the model-simulated cloud properties over the TP. In addition to the common cloud products retrieved from satellite observations, the CloudSat and CALIPSO products in COSP provide the vertical distributions of the cloud fractions and equivalent cloud microphysical properties. CloudSat provides multi-level cloud fractions and radar reflectivity, while CALIPSO produces multi-level lidar scattering ratio (SR) data. These datasets can be used to analyze the three-dimensional structures of clouds over the TP and validate the modeled results. The CALIPSO cloud fraction and SR are derived using the GCM-Oriented CALIPSO Cloud Product (GOCCP) (Chepfer et al., 2010), which diagnoses cloud properties from CALIPSO observations in the same way as in the model simulator. This ensures that discrepancies between the model and observations reveal biases in the model's cloudiness, rather than differences in the definition of clouds or diagnostics. The SR, which is proportional to the cloud optical depth, is defined as $$ {\rm SR}(z)=\dfrac{{\rm ATB}(z)}{{\rm ATB}_{\rm mol}(z)} , $$ where ATB is the attenuation backscattered profile from the 532-nm CALIOP lidar level-1 dataset, ATB_mol is the corresponding value in cloud-free conditions, representing the attenuated backscatter molecular signal profile, and z is the height of the ATB profiles. An atmospheric layer with SR >5 is classified as cloudy, while 0.01< SR <1.2 is clear, 1.2< SR <5 is unclassified, and SR <0.01 indicates the lidar signal is fully attenuated by a thick cloud above that layer. The monthly cloud fractions are determined at each vertical level by dividing the number of cloudy profiles identified during the month by the total number of SR profiles measured during that month.

2

2.2. ERA-Interim data

-->

2.2. ERA-Interim data

To evaluate the dynamic and thermodynamic fields in the ACCESS model, ERA-Interim (Dee et al., 2011) data are adopted. This dataset covers the period from January 1979 onwards and continues to be extended forward in near-real time. A large amount of observational data have been assimilated into the reanalysis, and these data are normally regarded as the best estimate of the state of the atmosphere. The temperature, moisture and wind fields for the same period as the CALIPSO/CloudSat observations are used to evaluate corresponding model variables. In addition, the cloud fraction and sensible and latent heat fluxes estimated by ERA-Interim are also used for model evaluation.

2

2.3. Precipitation data

-->

2.3. Precipitation data

For evaluating the model's precipitation fields, we collected three satellite-observed precipitation datasets. These precipitation data are: (1) Tropical Rainfall Measuring Mission (TRMM) (Huffman et al., 2010), designed to monitor tropical and subtropical precipitation and the associated energy balance——a joint space mission between NASA and the Japan Aerospace Exploration Agency. The mission delivered 17 years of precipitation data at a spatial resolution of 0.25^°, covering 50^°N to 50^°S for 1998-2015 (product 3B42; Kummerow et al., 2001); (2) Global Precipitation Climatology Project (GPCP) (Adler et al., 2012), which comprises monthly satellite-gauge precipitation computed from microwave and infrared sounder data observed by the International Constellation of Precipitation-related Satellites; and (3) CPC Merged Analysis of Precipitation (CMAP) (Xie and Arkin, 1997), which is derived from a combination of gauge data, satellite estimates, and NCEP-NCAR reanalysis, and comprises monthly mean data with a global coverage at a spatial resolution of 2.5^°× 2.5^°.

2

2.4. Radiation data

-->

2.4. Radiation data

The Surface Radiation Budget (SRB) project is a NASA/GEWEX project for the retrieval of surface radiative fluxes from satellite observations (Zhang et al., 2015). This dataset provides shortwave and longwave radiation on a global grid of 1^°× 1^° at three-hourly intervals using the algorithms of (Pinker and Laszlo, 1992) and (Fu et al., 1997). The data have a record length of 24 years from 1983 to 2007. These data are used in this work to evaluate the modeled radiation at the surface and the top of the atmosphere.

2

2.5. ACCESS model

-->

2.5. ACCESS model

The ACCESS model is used in this work to examine modelled cloud properties over the TP. A detailed description of the ACCESS model has been documented previously by (Bi et al., 2013). Briefly, however, the ACCESS model is a coupled ocean and atmosphere climate modeling system developed at the Collaboration for Australian Weather and Climate Research. It uses the UK Met Office Unified Model (MetUM; Walters et al., 2017) as its atmospheric component, and in this study we conducted an atmosphere-only model run. ACCESS MetUM, version 10.1, was run for 30 years with: September 1988 initial conditions; a horizontal resolution of 1.25^° latitude by 1.875^° longitude; and 85 vertical levels. Prescribed sea-surface temperature and sea-ice data with seasonal variations were used as forcing from the ocean. The simulated results for January 2006 to December 2012 period, with three-hourly model outputs, were used to generate the satellite-equivalent cloud products. These model outputs were then used to compare with observed cloud properties over the TP.
The scheme to simulate subgrid variability in the computation of the radar reflectivity is SCOPS (the Subgrid Cloud Overlap Profile Sampler), developed for the ISCCP (International Satellite Cloud Climatology Project) simulator (Webb et al., 2001). The subgrid distribution of clouds is generated within a grid box by dividing the grid box into a number of vertical columns using a pseudorandom sampling process, fully consistent with the maximum, random and maximum/random cloud overlap assumptions. Maximum overlap is used for the convective cloud, and maximum/random is specified to the large-scale cloud. The convective and large-scale cloud water contents are distributed evenly in the subcolumns that are occupied by convective and large-scale cloud, respectively. Thus, the subgrid distribution in this case only accounts for cloud overlap assumptions, but not for the subgrid distribution of cloud water itself.
The subgrid sampling data are aggregated to produce a final product at the model grid box resolution. In this study, model outputs are analyzed from the CloudSat and CALIPSO simulators. The monthly mean cloud products are produced from daily means of three-hourly calculations, which are further averaged over the seven years to form a short-period climatology to compare with the satellite observations (CloudSat and CALIPSO).

3. Comparison of modelled and observed cloud properties over the TP in boreal summer

4. Investigation of the ACCESS model cloud bias

4.1. Dynamic and thermodynamic fields

-->

4.1. Dynamic and thermodynamic fields

To find possible reasons that may explain the ACCESS model errors, dynamic and thermodynamic fields generated by ACCESS and ERA-Interim are analyzed. Figure 6 shows TP month-pressure cross sections of relative humidity (RH), temperature, and vertical velocity from ERA-Interim and ACCESS.

Figure6. Month-pressure cross section of RH (shaded, %), temperature (red contours, K), and vertical velocity (black contours) over the main TP in (a) ERA-Interim and (b) ACCESS. (the vertical velocity is multiplied by -1000 so that it has units of - 10^-3 Pa s^-1.)

ACCESS captures the main features of the seasonal variation of temperature; however, at levels above 250 hPa in JJA, the RH is larger and vertical velocity is stronger than in ERA-Interim. The strong vertical velocity in JJA brings more water vapor to upper levels, leading to the higher cloud fractions above 10 km (Fig. 5d). Note a large difference in RH between ACCESS and ERA-Interim occurs at the lower boundary, due to the boundary cutting through mountains for the ACCESS model, which is not applied in the ERA-Interim boundary condition.
Figures 7a and b show that the circulation and temperature around the TP are similar between ERA-Interim and ACCESS. The low-level southern air climbing over the TP southern slope, then turning back to tropical regions at high levels, is simulated successfully in ACCESS, but the vertical velocity at 850-200 hPa is overestimated over the TP southern slope, while it is underestimated over the Bay of Bengal (10^°-20^°N). Figure 7c shows the difference in wind and RH between ACCESS and ERA-Interim. Most of the vertical velocity over the TP is overestimated in ACCESS, especially over the southern slope, and the high-level RH is also overestimated over the TP.

Figure7. Mean latitude-pressure cross sections of wind (vectors), RH (color shading, %), and temperature (contours, K) over 85^°-101^°E in (a) ERA-Interim, (b) ACCESS, and (c) their differences. The black shade represents the topography.

Figures 6 and 7 suggest that the main reason for the ACCESS cloud simulation problem over the TP is that the lifting effect of the TP is stronger than in ERA-Interim, especially over its southern slope. The strong vertical velocity transports more water vapor to high levels, resulting in the higher frequency of ice cloud aloft (due to increased ice particle formation) and lower frequency of cumulus cloud at middle and low levels (due to reduced moisture content) over the TP.

2

4.2. Surface energy budget

-->

4.2. Surface energy budget

To see if the surface heating on the TP is too strong in ACCESS, which may be partly responsible for its positive bias in convection and vertical velocity, the surface energy budget is analyzed using radiation data retrieved from the NASA/GEWEX SRB project.
Figure 8a shows the surface net shortwave radiation difference in JJA between the model and SRB observations. It can be seen that the net shortwave radiation over the main TP is overestimated by the ACCESS model, and the cloud shortwave radiative effect (CSRE) in this region is less than observed in terms of absolute values, leading to positive biases as shown in Fig. 8c. These results are consistent with the cloud distribution presented in the previous sections: the high clouds are overestimated, while the middle clouds are underestimated. Such cloud distributions allow more solar radiation to reach the surface and reduce the CSRE, leading to a positive bias in CSRE. More shortwave heating at the surface facilitates convection and makes the strong convection over the TP even stronger.

Figure8. Radiation difference between ACCESS and SRB for (a) net shortwave radiation at the surface, (b) OLR at the TOA, (c) the cloud shortwave radiative effect at the surface, and (d) net upward longwave radiation at the surface. Units: W m^-2. The black contour represents the 3000 m topography. The dashed box (31^°-35^°N, 85^°-101^°E) shows the main region of the TP selected in this study.

Figure 8b shows the biases of the outgoing longwave radiation (OLR) at the top of the atmosphere (TOA). Negative values mean underestimations of OLR in the model, and these are seen in the eastern, southeastern, and northwestern TP regions. The lower OLR means that the cloud-top heights are too high. The model overestimates the OLR in the western TP region. Figure 8d also shows a net upward longwave flux difference at the surface, which results in the main TP having very similar OLR differences at the TOA. The negative bias suggests that the modeled downward longwave flux is larger than observed, which may reflect the effect of high-level clouds and thus indicate a consistent cloud effect.
Figure 9 compares the sensible and latent heat flux between ACCESS and ERA-Interim. The model overestimates the sensible and latent heat fluxes over the main TP, and the overestimation of the latent flux spreads over the whole TP too. The seasonal variations of surface energy fluxes over the main TP are also calculated, as plotted in Fig. 10. Compared with ERA-Interim, the ACCESS model underestimates the sensible heat fluxes from January to March and overestimates them from May to June and August to December, while the model's latent heat fluxes are systematically overestimated throughout the year. These results indicate that the overestimation of sensible and latent heat fluxes by ACCESS may also be responsible for the strong summer convection over the main TP.

Figure9. (a) Difference in surface sensible heat flux between ACCESS and ERA-Interim. (b) As in (a) except for latent heat flux. Units: W m^-2. The black contour represents the 3000 m topography. The dashed box (31^°-35^°N, 85^°-101^°E) shows the main region of the TP selected in this study.

Figure10. Seasonal variations of surface (a) sensible heat flux and (b) latent heat flux over the main TP. Units: W m^-2.

2

4.3. Precipitation

-->

4.3. Precipitation

Strong convection must result in strong precipitation, and so evaluation of the model's precipitation can be used as evidence confirming the excessive convection. For this purpose, the precipitation is compared with three observed precipitation climatologies, based on retrievals from satellites, gauge observations, and reanalysis, respectively. These precipitation data, corresponding to the period of the model simulations (2006-12), are averaged for the summer mean and compared with the model's results.
Figure 11 shows that the distribution pattern of precipitation from the three observed climatologies are very similar, with large precipitation occurring on the southern slope of the TP and extending to the Bay of Bengal. Over the main part of the TP, the precipitation decreases from southeast to northwest. The distribution pattern produced by the ACCESS model shows high values in the south and low values in the north, which is quite different from the observation. The precipitation along the TP southern slope in ACCESS is much larger than observed, and overestimation is also apparent in the western and northwestern parts of the TP. These comparisons confirm that the ACCESS model overestimates the precipitation along the southern slope and main part of the TP. An overestimation of precipitation over the TP was also identified by (Walters et al., 2017), using MetUM in both the GA4 and GA6 configurations, and the results presented here are consistent with their findings.

Figure11. Total precipitation (units: mm d^-1) in JJA determined by (a) GPCP, (b) CMAP, (c) TRMM, and (d) ACCESS. The black contour represents the 3000 m topography. The dashed box (31^°-35^°N, 85^°-101^°E) shows the main region of the TP selected in this study.

To analyze the precipitation more carefully, convective rain and large-scale rain need to be evaluated separately. Only TRMM data include convective and stratified components of precipitation, but these precipitation types over the TP region have been found to be unreliable (Fu and Liu, 2007), due to misidentified weak convective precipitation as large-scale precipitation, and therefore the results cannot be used for our purposes. Instead, the convective precipitation and large-scale precipitation from ERA-Interim are used in this study. Figures 12a and b compare the total precipitation between CMAP and ERA-Interim. It can be seen that the agreement between these two datasets is generally better than that between CMAP and the ACCESS model, albeit the results from ERA-Interim along the southern slope of the TP are less than those of CMAP. Figures 12c-f compare the convective and large-scale precipitation between the ACCESS model and ERA-Interim. It can be seen that ACCESS overestimates both large-scale and convective precipitation, but the overestimation of convective precipitation is much larger than that of the large-scale precipitation, especially along the southern slopes of the TP. This suggests that the total rainfall overestimation over the main TP in ACCESS is mainly due to the overestimation of convective precipitation.

Figure12. Total precipitation in JJA determined by (a) CMAP and (b) ERA-Interim. (c, d) JJA convective precipitation in (c) ACCESS and (d) ERA-Interim. (e, f) JJA large-scale precipitation in (e) ACCESS and (f) ERA-Interim. Units: mm d^-1. The black contour represents the 3000 m topography. The dashed box (31^°-35^°N, 85^°-101^°E) shows the main region of the TP selected in this study.

5. Conclusions and discussion

Using the radar reflectivity and lidar SR dataset from CloudSat and CALIPSO, the horizontal and vertical distributions and seasonal variations of cloud properties over the TP in boreal summer simulated by the ACCESS model are evaluated in this paper.
The results show that the ACCESS model generates a lower frequency of cloud at 7-11 km over the TP, but a higher frequency of cloud at 11-16 km and below 7 km. These results indicate that ACCESS produces less cumulus clouds at middle levels, but more small ice particles at high levels and drizzle-sized drops at low levels.
The ACCESS model also produces biases in the seasonal variation of cloud fraction over the TP. With the onset of the summer monsoon in May, observations show that the maximum of cloud fractions occurring at 7-11 km in May descends to lower levels, at 6-9 km, over the main TP; whereas, it shifts upward to a higher level, at 10-15 km, in ACCESS simulations.
The dynamic and thermodynamic fields simulated by ACCESS are then compared with ERA-Interim data, showing that the vertical velocity over the TP is stronger than in ERA-Interim, especially on the TP's southern slope. The strong vertical velocity brings more water vapor to higher levels, resulting in more cloud at high levels and less cumulus cloud at middle levels. This suggests that the effects of topography and the boundary layer scheme in the ACCESS model may need to be adjusted in the TP region to reduce the lifting effect.
The surface radiation fields generated by ACCESS are evaluated against the same variables estimated from satellite measurements, while the sensible and latent heat fluxes are compared with ERA-Interim. The results show that the model overestimates the surface net solar radiation, sensible and latent heat fluxes, which may be a possible cause for the strong convection over the TP. The modeled precipitation is evaluated using three observational datasets estimated from satellites, revealing that the ACCESS model also overestimates precipitation along the southern slope and western and northwestern parts of the TP, especially for convective precipitation.
The model biases over the TP region explored in this study raise an unclear issue of positive feedback, i.e., whether the surface forcing produces convection that is too strong and leads to the incorrect vertical cloud structure, or it is the vertical cloud distribution that results in the incorrect surface radiative forcing. To answer this question, we intend to conduct some sensitivity studies to test the effects of some critical physical processes, such as convection, topography, and boundary conditions. Due to the complex topography in this area, especially along the south slope of the TP, the effect of model resolution may also have an important impact. These sensitivity studies will be conducted in the near future using the ACCESS city-scale model, with a relatively high spatial resolution.

本站小编 Free考研考试/2022-01-02

Corresponding author: Difei DENG,ddfttkl@gmail.com;

HTML

2.1. Cloud data from CloudSat and CALIPSO

2.2. ERA-Interim data

2.3. Precipitation data

2.4. Radiation data

2.5. ACCESS model

4.1. Dynamic and thermodynamic fields

4.2. Surface energy budget

4.3. Precipitation

相关话题/Evaluation Summer Monsoon

领限时大额优惠券,享本站正版考研考试资料!

Variation in Principal Modes of Midsummer Precipitation over Northeast China and Its Associated Atmo

Factors Limiting the Forecast Skill of the Boreal Summer Intraseasonal Oscillation in a Subseasonal-

Predictable and Unpredictable Components of the Summer East Asia-Pacific Teleconnection Pattern

Changes in the Proportion of Precipitation Occurring as Rain in Northern Canada during Spring-Summer

Influence of Low-frequency Solar Forcing on the East Asian Winter Monsoon Based on HadCM3 and Observ

Seasonal Forecasts of the Summer 2016 Yangtze River Basin Rainfall

Skillful Seasonal Forecasts of Summer Surface Air Temperature in Western China by Global Seasonal Fo

Effect of Horizontal Resolution on the Representation of the Global Monsoon Annual Cycle in AGCMs

Different Asian Monsoon Rainfall Responses to Idealized Orography Sensitivity Experiments in the Had

Assessing Global Warming Induced Changes in Summer Rainfall Variability over Eastern China Using the