The applicability of a new soil hydraulic property of frozen soil scheme applied in Community Land Model 4.5 (CLM4.5), in conjunction with an impedance factor for the presence of soil ice, was validated through two offline numerical simulations conducted at Madoi (GS) and Zoige (ZS) on the Tibetan Plateau (TP). Sensitivity analysis was conducted via replacing the new soil hydraulic property scheme in CLM4.5 by the old one, using default CLM4.5 runs as reference. Results indicated that the new parameterization scheme ameliorated the surface dry biases at ZS but enlarged the wet biases which existed at GS site due to ignoring the gravel effect. The wetter surface condition in CLM4.5 also leads to a warmer surface soil temperature because of the greater heat capacity of liquid water. In addition, the combined impact of new soil hydraulic property schemes and the ice impedance function on the simulated soil moisture lead to the more reasonable simulation of the starting dates of freeze-thaw cycle, especially at the thawing stage. The improvements also lead to the more reasonable turbulent fluxes simulations. Meanwhile, the decreased snow cover fraction in CLM4.5 resulted in a lower albedo, which tended to increase net surface radiation compared to previous versions. Further optimizing is needed to take the gravel into account in the numerical description of thermal-hydrological interactions.
Land surfaces, by affecting water and energy flows between the ground and the atmosphere, have a significant impact on weather forecasting as well as climate change. At present, the importance of land surface processes is widely recognized [
The soil is the main body of land surface processes, connecting the earth spheres. The soil freeze-thaw process is an important parameter in the land surface and atmospheric modeling. The variations of soil temperature and soil moisture are indicators of frozen soil. Owing to the great gap of thermal conductivity and thermal capacity between liquid water and ice, the accuracy of simulated soil moisture can directly affect the calculation of thermal properties and thermal conductivities during freezing-thawing period. Meanwhile, the phase transition of soil moisture spends much of the energy exchanged between the atmosphere and the soil. The energy liberation and the energy absorption during the period change the distribution characteristics of ground temperature and also the pattern of energy. The soil temperature and soil moisture simulations should be considered prior to the assessment of land surface model performance.
The Tibetan Plateau (TP), which is characterized by flat topography and high elevation, plays an essential role in atmospheric circulation [
The Community Land Model (CLM) is one of the most widely used land surface models in the world. Many previous works have confirmed its simulation ability in different regions [
CLM simulates the partitioning of mass and energy from the atmosphere, redistributes the mass and energy of the land surface, and then exports the fresh water and heat to the oceans [
CLM uses Darcy’s law to describe the vertical flux of water, which depends on the hydraulic conductivity
However, the calculation of soil hydraulic properties of frozen soils in CLM4.0 depends on the total water content (
In CLM4.5, the calculation of the hydraulic properties of frozen soils was modified by replacing their dependence on total water content with liquid water content only (
Simulations were conducted at two typical sites on the TP (Figure
The Tibetan Plateau (TP), the locations of the Madoi site and the Zoige site.
The observation site, Grass Station (GS, 34°54′N, 97°33′E) at Madoi, is located on the alpine grass on the west side of Ngoring Lake with a distance of 1.7 km to lake. The Zoige Grassland Station (ZS, 33°89′N, 102°14′E) is located in an alpine meadow grassland.
At GS site, the sensible heat flux and latent heat flux were measured with the eddy covariance system (EC), consisting of a three-dimensional sonic anemometer (CSAT3, Campbell Scientific, Inc.) and an open-path CO2/H2O infrared gas analyzer (IRGA; LI 7500, LI-COR, Inc.). The eddy covariance system was mounted at a height of 3.2 m above the land surface, and the sensor signals were recorded by a data logger (CR3000, Campbell Scientific, Inc.) at a 10-Hz frequency. The incoming and outgoing shortwave and longwave radiation were measured with a net radiometer (CNR-1/CNR-4, Kipp and Zonen) at 1.5 m above the ground, respectively. Besides, the soil water content and temperature were measured at soil depths of 0.05, 0.10, 0.20, 0.40, and 0.80 m with CS616 and 109L, respectively (Campbell Scientific, Inc.). The air temperature and relative humidity sensors (HMP-45C) were also installed at a height of 3.2 m. Precipitation was measured with a weighing gauge (T200B, Geonor, Norway). Soil heat flux was measured using heat flux plates (HPF01) buried at 5 to 10 cm.
The EddyPro flux analysis software was used for data corrections in the postprocessing of 10 Hz EC data. Corrections such as time lag compensation spike and trend remove and coordinate rotation were also applied to the original data. Meanwhile, the raw virtual air temperature was also converted to air temperature. The water vapor fluxes were also amended using Webb-Pearman-Leuning (WPL) density correction [
At ZS site, the sensible heat flux and latent flux were also measured with the EC, consisting of a three-dimensional sonic anemometer (CSAT3, Campbell Scientific, Inc., Logan, UT, USA) and an open-path fast response infrared gas analyzer (LI-7500, LI-COR Biosciences Inc., Lincoln, NE, USA). Separation distance between the gas analyzer and sonic anemometer sensors was 0.15 m. Both sensors were mounted 3.15 m above the soil surface. The air temperature and relative humidity sensor (HMP-45C, Vaisala, Helsinki, Finland) was also installed at the same height, which was used for correction of flux measurements for density effects due to heat and water vapor transfer. Similar to the measurements conducted at GS, signals from EC instrumentation were also recorded at rates of 10 Hz, and the raw data were stored in a CR3000 data logger (Campbell Scientific, Inc.). Net radiation (
Similar to the GS site, the raw data from ZS site from EC were processed with EddyPro flux analysis software. Spike and trend remove and coordinate rotation were also applied to the raw data as well as the coordinate rotation. Turbulent heat fluxes (the sensible heat flux and latent heat flux) were calculated with 30 min average and also adjusted for fluctuations in air density due to water vapor. Sporadic missing data at these two sites were replaced through linear interpolation combined with other meteorological factors.
The data used in this study spanned the period from 6/1/2013, to 12/31/2014, at the GS, and 6/1/2009, to 12/31/2010, at the ZS (Figure
The main forcing data at Madoi (left panel, from 6/1/2013 to 12/31/2014) and Zoige site (right panel, from 6/1/2009 to 12/31/2014) on the Tibetan Plateau (TP). (a1), (a2) Temperature (°C); (b1), (b2) incident solar (W m−2); (c1), (c2) specific humidity (%); (d1), (d2) precipitation (mm day−1).
In this study, we use CLM4.5 in offline mode. In order to examine the capability of the new soil hydraulic parameterization scheme in CLM4.5, two numerical simulations were conducted. One is called control simulation (hereafter referred to as CTL), in which model run with the default new soil hydraulic parameterization scheme in CLM4.5; another is called sensitivity simulation (hereafter referred to as EXP), in which the new soil hydraulic properties formulas in CLM4.5 were replaced with the original ones in CLM4.0. The soil compositions parameters in the model at these two sites were settled based on field observations (Tables
Values of soil parameters for different soil layers at GS site.
Layer | Depth |
Sand (%) | Clay (%) | Organic (kg m−3) | Canopy height/m | Covered-area (alpine-meadow) |
---|---|---|---|---|---|---|
1 | 0.0175 | 38.64 | 26.96 | 85.00 | ||
2 | 0.0451 | 38.64 | 26.96 | 75.12 | ||
3 | 0.0906 | 68.60 | 14.21 | 40.14 | ||
4 | 0.1655 | 65.41 | 21.28 | 31.37 | ||
5 | 0.2891 | 65.41 | 21.28 | 18.14 | 0.05 | 55% |
6 | 0.4929 | 94.03 | 3.44 | 1.92 | ||
7 | 0.7289 | 93.42 | 2.69 | 1.18 | ||
8 | 1.3828 | 94.17 | 3.97 | 1.10 | ||
9 | 2.2961 | 94.17 | 3.97 | 0.00 | ||
10 | 3.8019 | 91.52 | 4.32 | 0.00 |
Same as Table
Layer | Depth |
Sand (%) | Clay (%) | Soil organic (kg m−3) | Canopy height/m | Covered-area (alpine-meadow) |
---|---|---|---|---|---|---|
1 | 0.0175 | 19.25 | 2.67 | 120.40 | ||
2 | 0.0451 | 19.25 | 2.67 | 120.40 | ||
3 | 0.0906 | 28.39 | 3.01 | 82.53 | ||
4 | 0.1655 | 28.39 | 3.77 | 82.53 | ||
5 | 0.2891 | 28.04 | 3.97 | 53.15 | 0.20 | 100% |
6 | 0.4929 | 32.66 | 3.65 | 28.91 | ||
7 | 0.7289 | 46.47 | 1.84 | 6.62 | ||
8 | 1.3828 | 68.35 | 1.82 | 1.67 | ||
9 | 2.2961 | 87.11 | 1.62 | 0.00 | ||
10 | 3.8019 | 92.01 | 1.54 | 0.00 |
The accuracy of soil liquid water content simulation is a significant indicator to assess the modeling capabilities of land surface models. The phase transition of soil moisture proceeds with energy absorption and release. Accurate soil liquid water content simulation has important influence on the energy redistribution between ground and atmosphere. Meanwhile, different volume partitions between ice and liquid water in the soil column could lead to different soil permeability. In addition, owing to the great huge gaps of thermal conductivity and thermal capacity between ice and liquid water, the soil liquid water content simulation could directly affect the accuracy of simulated soil temperature. In previous hydraulic properties formulas, the effect of ice on the flow of water through frozen soil is considered as a minor factor. Water in the surface layer would fail to contribute more subsurface runoff before it drains from the active layer into deeper frozen layers. And this leads to excessively dry soils in frozen soil regions [
Profile of soil moisture differences (DIFF, CTL minus EXP) at different sites ((a) GS, (b) ZS).
The relationships between precipitation and soil liquid water content at different sites were illustrated in Figure
Statistical results of simulated soil liquid water content from CTL and EXP at ZS site.
5 cm | 20 cm | 40 cm | 80 cm | Average | |
---|---|---|---|---|---|
|
|||||
CTL | 0.73 | 0.74 | 0.71 | 0.40 | 0.65 |
EXP | 0.64 | 0.68 | 0.63 | 0.26 | 0.55 |
|
|||||
Bias ( |
|||||
CTL | −0.05 | 0.04 | −0.02 | −0.02 | 0.03 |
EXP | −0.07 | 0.02 | −0.04 | −0.04 | 0.04 |
Daily average precipitation (mm day−1) (a) and soil moisture (m3 m−3) at (b) 5 cm, (c) 20 cm, (d) 40 cm, and (e) 80 cm at ZS site.
Same as Figure
However, as shown in Figure
Same as Table
5 cm | 20 cm | 40 cm | 80 cm | Average | |
---|---|---|---|---|---|
|
|||||
CTL | 0.49 | 0.56 | 0.47 | 0.86 | 0.60 |
EXP | 0.50 | 0.60 | 0.43 | 0.86 | 0.60 |
|
|||||
Bias ( |
|||||
CTL | 0.04 | −0.01 | −0.02 | 0.00 | 0.02 |
EXP | 0.04 | −0.04 | −0.02 | −0.01 | 0.03 |
The soil temperature simulation depends on the accuracy of simulated soil thermal conductivity and heat capacity. The thermal conductivity and the heat capacity of ice are 2.2 W m−1 K−1 and 1.9 MJ m−3 K−1, while those of the liquid water are 0.56 W m−1 K−1 and 4.2 MJ m−3 K−1, respectively. Based on these great gaps, the soil moisture simulation could influence the simulated soil temperature directly. Figure
Profile of soil temperature differences (CTL minus EXP) at (a) GS and (b) ZS site.
Comparisons of simulated soil temperature with observations at certain depths at different sites were shown in Figure
Statistical results of simulated soil temperature from CTL and EXP at ZS site.
5 cm | 20 cm | 40 cm | 80 cm | Average | |
---|---|---|---|---|---|
|
|||||
CTL | 0.94 | 0.98 | 0.98 | 0.99 | 0.97 |
EXP | 0.94 | 0.98 | 0.98 | 0.98 | 0.97 |
|
|||||
Bias (°C) | |||||
CTL | 0.79 | −0.13 | −1.17 | −0.27 | 0.59 |
EXP | 0.84 | 0.01 | −1.01 | −0.12 | 0.50 |
Same as Table
5 cm | 20 cm | 40 cm | 80 cm | Average | |
---|---|---|---|---|---|
|
|||||
CTL | 0.95 | 0.96 | 0.97 | 0.98 | 0.97 |
EXP | 0.95 | 0.96 | 0.97 | 0.98 | 0.97 |
|
|||||
Bias (°C) | |||||
CTL | 1.30 | 1.58 | 1.39 | 0.32 | 1.15 |
EXP | 1.27 | 1.57 | 1.38 | 0.32 | 1.14 |
Daily average soil temperature (°C) at (a) 5 cm, (b) 20 cm, (c) 40 cm, and (d) 80 cm at ZS site.
Daily average soil temperature (°C) at (a) 5 cm, (b) 20 cm, (c) 40 cm, and (d) 80 cm at GS site.
Four freeze/thaw stages were identified: the completely frozen stage (the maximum daily soil temperature below 0°C); the thawing stage (the soil column profile is experiencing thawing process); the completely thawed stage (the minimum daily soil temperature stay above 0°C); and the freezing stage (the soil column is experiencing freezing process). Meanwhile, ground diurnal freeze/thaw cycles were judged to have occurred when the daily maximum soil temperature was above 0°C and the minimum soil temperature was subzero. In order to avoid the potential impact of random weather processes on the judgement, the occurrence of three consecutive days meeting a chosen set of criteria was used as an indicator of the transition, and the first day was recorded as the starting date of the next freeze/thaw stage [
Four freeze-thaw stages determined the daily maximum and minimum soil temperatures at 10 cm under the surface (month-days).
Frozen | Thawing | Thawed | Freezing | |
---|---|---|---|---|
GS | ||||
OBS | 1/1/−3/27/2014 | 3/28/−5/2/2014 | 5/3/−10/20/2014 | 10/21/−12/31/2014 |
(86 days) | (36 days) | (171 days) | (72 days) | |
CTL | 1/1−3/4/2014 | 3/5/−4/29/2014 | 4/30/−9/26/2014 | 9/27/−12/31/2014 |
(63 days) | (56 days) | (150 days) | (96 days) | |
EXP | 1/1−3/4/2014 | 3/5/−4/29/2014 | 4/30/−9/26/2014 | 9/27/−12/31/2014 |
(63 days) | (56 days) | (150 days) | (96 days) | |
|
||||
ZS | ||||
OBS | 1/1/−3/19/2010 | 3/20/−4/3/2010 | 4/4/−11/11/2010 | 11/12/−12/31/2010 |
(78 days) | (15 days) | (222 days) | (50 days) | |
CTL | 1/1/−3/11/2010 | 3/12/−3/28/2010 | 3/29/−11/12/2010 | 11/13/−12/31/2010 |
(70 days) | (17 days) | (229 days) | (49 days) | |
EXP | 1/1/−2/17/2010 | 2/18/−4/13/2010 | 4/14/−11/11/2010 | 11/12/−12/31/2010 |
(48 days) | (55 days) | (212 days) | (50 days) |
Observed and simulated days in each freezing and thawing stage at different sites. (a) is for the GS site. (b) is for the ZS site.
Diurnal variations of soil temperature in the 10 cm soil layer in four freeze-thaw stages in the ZS site. (a) Completely frozen, (b) thawing, (c) completely thawed, and (d) freezing.
Same as Figure
Based on the observations at ZS (GS), the soil was completely frozen on 1/1/2010 to 3/19/2010 (1/1/2014 to 3/27/2014), thawing on 3/20 to 4/3 (3/28 to 5/2) until completely thawed on 4/4 to 11/11 (5/3 to 10/20), and refreezing on 11/12 (10/21). Compared to EXP, the CTL performed better at ZS site for simulating the starting dates of four stages, especially on the thawing stages. Because the hydraulic conductivity was calculated based on liquid water content in the CTL, the movement of water through frozen soil decreased greatly while porosities in the surface layer were filled with ice. As surface ice melted at progressively deeper depths, water infiltrates further into the deeper layer. However, in the old scheme in the EXP, the relatively warm water could infiltrate in the deeper frozen layer while the surface was icy, thus making the deeper layer begin to melt immediately. This led to the advanced melting date. In addition, due to the wetter surface during simulation at ZS site, an increased heat capacity in the surface ground induced the decrescent diurnal range of soil temperature during the freezing-thawing process and the simulated surface soil temperature agreed better with the observed in the completely frozen and thawed stages than that in the freezing-thawing periods. For the simulations of starting dates of the four soil freeze-thaw stages at GS site, the new soil hydraulic properties scheme did not show prominent advantages due to its undistinguished performance of soil temperature in simulations on the surface. Both of these two schemes tended to simulate longer thawing/freezing stage but shorter thawed/frozen stage compared to the observation. The wetter surface conditions were simulated at GS site releasing (absorbing) more energy in the freezing (thawing) stage, which expanded the daily amplitudes of surface soil temperatures compared to the observations (Figures
CLM follows the principle of surface energy balance:
Because energy components data at GS site were missing for a long time during the studied period, we have not analyzed that site. We also conducted the offline simulation using CLM4.0 with the same model parameter definitions used in CTL during 2010 at ZS site helping to make the assessment more completely. We defined the runs of CLM4.0 as EXP1, to distinguish them from the EXP. Sensitivity analysis of EXP has also been carried out. Figure
Differences of daily mean energy flux (W m−2) (a)
Furthermore, advanced thawing date simulated in EXP made the surface wetter compared to CTL, thus leading to the higher LE and lower
Statistical results of daily mean energy flux components between simulations and observation at ZS site.
|
LE |
|
| |
---|---|---|---|---|
|
||||
CTL | 0.23 | 0.78 | 0.73 | 0.47 |
EXP | 0.29 | 0.76 | 0.72 | 0.44 |
EXP1 | 0.22 | 0.82 | 0.73 | 0.48 |
|
||||
Bias (Wm−3) | ||||
CTL | 10.40 | −0.07 | 0.27 | 0.92 |
EXP | 8.80 | −0.97 | 0.40 | 0.54 |
EXP1 | 23.25 | −13.83 | 2.32 | 0.87 |
As shown in Figure
Same as Figure
This study conducted offline numerical simulations at two typical sites on the TP. The simulation ability of the newly released land surface model CLM4.5 was examined and the improvements of the new hydraulic properties scheme, which has direct effects on the simulation of freeze-thaw cycles and the hydraulic-thermal properties of seasonal frozen soil via the relative sensitivity analysis, were verified. The conclusions are as follows: The model can capture the variations of measured soil liquid water contents and the simulated peaks well consistent with the maximum rainfall. The new hydraulic properties scheme can effectively reduce the dry bias existing in surface soil at ZS. However, lacking consideration of the gravel effect made the modification not significant at GS site. The existence of gravel could cause the gravel soil to be dryer than fine soil. Failure to dispose the gravel in the model formulas made the simulated soil liquid water content overestimated compared to the observation. In contrast, modification in CTL conducted at GS site played a negative role. The wetter surface condition in CLM4.5 also led to the warmer surface soil temperature because of the greater heat capacity of liquid water. The accuracy of soil temperature simulation is closely related to the simulated soil moisture. Temperature difference between CTL and EXP is more obvious at ZS. The CTL performed better at ZS for simulating the starting dates of four stages compared to EXP, especially on the thawing stages. Under the influences of the ice impedance function and the new hydraulic properties schemes in CTL, the movement of water through frozen soil has been decreased greatly. As surface ice melts at progressively deeper depths, water infiltrates further into the deeper layer. This delayed the melting date effectively. In addition, the increased heat capacity in the surface ground at ZS also induced the decreased diurnal range of soil temperature during the freezing-thawing process which agreed better with the observation. However, at GS, CTL did not show prominent advantages due to its undistinguished performance of soil liquid water content simulations on the surface. CLM4.5 computed the surface fluxes separately for snow-covered and snow-free fractions of a grid cell, rather than the uniform snow cover assumption adopted in previous versions. Decreased snow covers fraction resulted in the lower albedo. This reduction in surface albedo tended to increase net surface radiation in CLM4.5 compared to previous versions. Furthermore, the improvement in freeze/thaw starting dates simulations also led to the more reasonable turbulent fluxes simulations.
In summary, CLM4.5 could capture the hydrological-thermal interactions of seasonal frozen soil on the TP. The land surface model could simulate a more realistic land surface climate and then positively impact on the climate prediction while coupled with the regional climate models. However, the optimization concerning the numerical description of the hydraulic and thermal properties during freeze/thaw cycle based on the endemic soil texture on the TP is still needed in future study.
The authors declare that they have no competing interests.
This work was supported by the National Natural Science Foundation of China, Grants 41375077, 41130961, 91537104, and 41405088. The observation data were provided by the Zoige Plateau Wetlands Ecosystem Research Station and the GS Grassland Station under sponsorship of the National Science Foundation. The authors acknowledge computing resources and time from the Supercomputing Center, Big Data Center of Cold and Arid Region Environment and Engineering Research Institute, Chinese Academy of Sciences.