Abstract

This study evaluated the effect of recent eastern Siberian land surface changes, such as water surface expansion, on water-energy fluxes and precipitation and focused on land surface parameters using a three-dimensional atmospheric model [the Japan Meteorological Agency Nonhydrostatic model (JMA-NHM)]. Five parameters were set (viz., surface albedo, evaporative efficiency, roughness length, heat capacity, and thermal conductivity), and a response of evaporation and precipitation was evaluated. Increased precipitation corresponded to 75% of the increased evaporation on interparameter average, indicating strong land–atmosphere coupling. Water-energy flux and precipitation responses to water surface expansion were evaluated by two methods: JMA-NHM and the parameter sensitivity method. The latter method used a linear combination of parameter sensitivity on the fluxes and precipitation and parameter changes with land surface change. JMA-NHM demonstrated an increase in evaporation and precipitation and a decrease in downward shortwave radiation with low-level cloud increases. The parameter sensitivity method gave the same order as JMA-NHM in the estimation. This method has minimal calculation cost; thus, water-energy flux and precipitation response with further water surface expansion and decreases in forest area were simulated, producing various land surface data. The enhancement of the precipitation response to evaporation was weak for further water surface expansion in the largely expanded water surface area; however, the ratio increased dramatically for the small water surface expanding area, indicating intense water cycle enhancement at the beginning of water surface expansion. Although grassland formation from forest has minimal impact, if incoming downward shortwave radiation were to increase because of the disappearance of the forest shading effect and the water surface formed by permafrost melting, the water cycle would be enhanced intensely.

1. Introduction

The land surface strongly affects the lower atmosphere in eastern Siberia in the summer. The relationship between precipitation originated from land surface evaporation Pm and total precipitation P at the local scale has been investigated on the basis of the recycling ratio (Pm/P). For example, Serreze et al. (2002) investigated the recycling ratio for the Ob, Yenisey, and Lena basins and demonstrated a significant effect of local evaporation to precipitation in summer. At the global scale, Dirmeyer and Brubaker (2007) demonstrated that eastern Siberia has a higher local recycling ratio than other regions in the summer. Moreover, van der Ent and Savenije (2011) defined local moisture feedback in terms of length and time scales. Eastern Siberia has a shorter time of precipitation depletion and evaporation replenishment. Furthermore, small-scale precipitation recycling and evaporation recycling are apparent in summer, indicating strong moisture feedback over eastern Siberia.

Land cover change was recently reported in eastern Siberia, and it was found that the forest area is decreasing because of increasingly frequent forest fires (Potapov et al. 2008; Achard et al. 2006; Lepers et al. 2005). Moreover, the water surface of alas has expanded (Iijima et al. 2010). Alas is one of the stages of a thermokarst, which is formed by repeated thawing and freezing of frozen soil (permafrost). Alas includes grassland and water surfaces at its center. Iijima et al. (2010) reported changes in the alas component in which the alas water surface expanded in eastern Siberia for several years (Fig. 4 in Iijima et al. 2010). They showed that increases in soil temperature and thawing depth caused by heavy snowfall were the reason for the expansion of the alas water surface area. This alas water surface expansion is summarized in Fig. 1. This type of land cover change affects the lower atmosphere in eastern Siberia. Little is known about land–atmosphere coupling in this region, but the high regional–local recycling found by Serreze et al. (2002), Dirmeyer and Brubaker (2007), and van der Ent and Savenije (2011) indicates that increases in evaporation due to land cover change may have important effects on precipitation in the region.

Fig. 1.

Schematic diagram of alas water surface expansion.

Fig. 1.

Schematic diagram of alas water surface expansion.

Lower atmosphere responses to land cover change have been investigated in the Indochina Peninsula (Sen et al. 2004; Mabuchi et al. 2005a,b), the Amazon (Shukla et al. 1990; Nobre et al. 1991; Hahmann and Dickinson 1997), and China (Fu 2003). However, the lower-atmospheric response to land cover change in eastern Siberia remains unclear. The aim of this study was to determine how the lower atmosphere will change with land cover changes in summer when evaporation is enhanced. We considered parameter changes through land cover changes, such as forest area decreases or water surface expansion, with the exception of permafrost change, because the surface soil layer melts in summer (e.g., Ohta et al. 2008).

2. Model and experimental design

a. Model

We used the Japan Meteorological Agency Nonhydrostatic model (JMA-NHM; Saito et al. 2007). The model includes atmospheric physical processes and land surface processes. Although Saito et al. (2007) described the structure of JMA-NHM in detail, we also describe the land surface processes because this study focused on the effects of land surface parameter changes caused by land cover change on the lower atmosphere.

The heat-balance equation for the temperature at the top of the soil is as follows:

 
formula

where G1 is the ground heat flux of the top soil layer, α is the surface albedo, Sd and Ld are downward shortwave–longwave radiation, H is the sensible heat flux, LE is the latent heat flux, ɛ is the emissivity of the land surface, σ is the Stefan–Boltzmann constant, and Ts is the surface temperature of the top soil layer.

A heat conduction equation is used to calculate the temperature of each soil layer as follows:

 
formula
 
formula

where Tg is the soil temperature of each layer and cρ and λ are the heat capacity and conductivity of the soil, respectively. The soil is separated into four layers with different depths: the top, second, third, and fourth layers have depths of 0.04, 0.15, 0.40, and 0.60 m, respectively.

For a surface with vegetation, LE is calculated with the stomatal resistance rs as follows:

 
formula
 
formula
 
formula

where ra is the aerodynamic resistance, β is the evaporative efficiency, qsat is the saturated specific humidity for the surface temperature, qa is the specific humidity at the bottom of the atmospheric layer, cq is the bulk transfer coefficient, u(z1) is the wind speed at the bottom of the atmospheric layer z1, rsd is the stomatal resistance in daytime (30 s m−1), rsn is the same as rsd but for nighttime (300 s−1 m), and S0 is a constant value (1 W m−2).

The wind speed in the boundary layer u(z) is calculated as follows:

 
formula

where u* is the friction velocity, k is the von Kármán constant, d0 is the zero-plane displacement, ζ is a dimensionless length (ζ = z/L), L is the Monin–Obukhov length, and φ is a universal function. Land surface parameters that decide the characteristics of the land surface, α, β, z0, cρ, and λ, were included in Eqs. (1)(7).

The calculation domain is shown in Fig. 2. The domain center was taken as 66°00′00″N, 126°30′00″E. We used Global Land Cover Characterization (GLCC) from the U.S. Geological Survey (USGS) (Loveland et al. 2000) for the land surface type distribution and a global digital elevation model with horizontal grid spacing of 30 arc s (GTOPO30) for the topography and water surface distribution (U.S. Geological Survey 1996). The dominant vegetation was deciduous coniferous trees in the southern area and dwarf trees and shrubs in the northern area (Fig. 2). Considering the alas width of 0.1–15 km (Lopez et al. 2008) and the available computer resources, the grid size for the nested domain was set to 5 km, which was nested in a 30-km grid size. This downscaling was conducted as one-way nesting. The domain size was 4200 km × 4200 km for the 30-km grid and 2000 km × 2000 km for the 5-km grid. Data from the Japanese 25-year Reanalysis (JRA-25; Onogi et al. 2007), were used for the initial and boundary conditions in the 30-km grid-size calculations. The results were used as the initial and boundary conditions for the 5-km grid-size calculation. The results for the 5-km grid size were used as data for analyses. The calculation period was from 0000 UTC 21 June 2000 to 0000 UTC 1 August 2000. This period was chosen using Baseline Meteorological Data in Siberia (BMDS) (Suzuki et al. 2006), which consists of observation data from a Siberian station. During the study period, the average amount of precipitation in the 17 BMDS sites contained in the domain of the 5-km grid run was most similar to that of the climatological mean for 1986–2004 from July to August. The length of the simulation was chosen by considering the available computer resources and the equilibrium state of the atmosphere after 10 days of spinup. Thus, the initial 10 days were excluded as spinup, and all of July 2000 was used for analysis. The other model settings, with the exception of the land surface parameter settings, are shown in Table 1.

Table 1.

Experiment design. NOAA is the National Oceanic and Atmospheric Administration.

Experiment design. NOAA is the National Oceanic and Atmospheric Administration.
Experiment design. NOAA is the National Oceanic and Atmospheric Administration.
Fig. 2.

Calculation domain and distribution of water surfaces and vegetation. Green shading indicates that the elevation is <250 m MSL. The black line and blue-hatched, purple-hatched, and red-hatched areas denote the water area, ground cover with dwarf trees and shrubs, broadleaf deciduous trees, and deciduous coniferous trees, respectively. The outer frame is the calculation domain of 141 × 141 grids with a grid size of 30 km, and the red inner frame indicates 401 × 401 grids with a grid size of 5 km.

Fig. 2.

Calculation domain and distribution of water surfaces and vegetation. Green shading indicates that the elevation is <250 m MSL. The black line and blue-hatched, purple-hatched, and red-hatched areas denote the water area, ground cover with dwarf trees and shrubs, broadleaf deciduous trees, and deciduous coniferous trees, respectively. The outer frame is the calculation domain of 141 × 141 grids with a grid size of 30 km, and the red inner frame indicates 401 × 401 grids with a grid size of 5 km.

b. Independent change in land surface parameters experiment

We designed this experiment to clarify which land surface parameters are significant for the lower atmosphere when they change independently. Land cover changes associated with an increase in forest fires occurred in the low elevation area (Potapov et al. 2008). Moreover, considering that water surfaces are primarily located below 250 m above mean sea level (MSL) (Fig. 2), the change in land surface parameters was determined for the area below 250 m MSL in the 5-km grid-size domain. Land surface parameters did not change for the 30-km grid-size domain. We examined five land surface parameters that describe the land surface state: α, β, z0, cρ, and λ. The lower-atmospheric response was examined with respect to the independent changes in each parameter. Five values were set for α, z0, cρ, and λ. Six values were set for β, as shown in Table 2. The upper and lower limits for each parameter were the maximum and minimum values among various land cover types derived from Kondo (1994), and other values were distributed equally between the maximum and minimum values. To investigate the seasonal and yearly dependence of parameter impacts, an additional sensitivity experiment was conducted. The first run was conducted for 21 June–31 July 2000; two additional periods were analyzed in that year, and two other periods were analyzed during other summer months. Following the first calculation period, the additional calculation periods were 1) 22 May–30 June 2000, 2) 22 July–31 August 2000, 3) 21 June–31 July 2001, and 4) 21 June–31 July 2002. For each duration, the first 10 days were excluded as spinup.

Table 2.

Parameter settings. The values of each parameter are changed for each sensitivity experiment. The parameter values changed only below 250 m MSL.

Parameter settings. The values of each parameter are changed for each sensitivity experiment. The parameter values changed only below 250 m MSL.
Parameter settings. The values of each parameter are changed for each sensitivity experiment. The parameter values changed only below 250 m MSL.

c. Water surface area expansion experiment

We evaluated the lower-atmospheric response associated with water surface expansion using two methods: JMA-NHM and parameter sensitivity. The former method was basically the same as that used in the previous experiment, but water surfaces were assigned random numbers. Considering the possibility that land cover changes to alas, including water-covered surfaces caused by permafrost melting, could occur anywhere in the lowlands of eastern Siberia, random numbers were used to determine the alas grid candidates. A random number (0.0–1.0) was generated for each grid in which the elevation was less than 250 m MSL in the 5-km grid-size domain. Almost 20% of lowland areas consist of alas (Lopez et al. 2008); thus, grids with random numbers between 0.0 and 0.2 were selected as alas. In this case, the grid ratio of lowlands was represented as (alas, original) = (0.2, 0.8). Moreover, alas consisted of grassland and water surfaces (Desyatkin et al. 2009); when grassland occupied 20% of the lowlands, the grid ratio was expressed as (grassland, water surface, original) = (0.2, 0.0, 0.8), which is written as (G, W, O) = (0.2, 0.0, 0.8) for simplicity. Using this abbreviated notation, the expansion of the water surface on alas (Iijima et al. 2010) is represented as (G, W, O) = (0.2, 0.0, 0.8) to (0.1, 0.1, 0.8) or (0.0, 0.2, 0.8). For (G, W, O) = (0.1, 0.1, 0.8), the grid to which a random number was assigned in the range of 0.0–0.1 was defined as grassland. When a random number between 0.1 and 0.2 was assigned, the grid was designated as a water surface. Original land surface information (GLCC, see Fig. 2) was used if the random number was greater than 0.2. These three grid-ratio patterns were used for the water surface area expansion experiments. The selection of water surface grids employed random numbers as a reference; thus, we conducted the experiment four times, each with a different random number set. Table 3 presents the respective values of the parameters for the land surface types. A lowland grid parameter was chosen according to Table 3. Figure 3 shows the distribution of evaporative efficiency for each grid ratio. The area-averaged evaporative efficiency was 0.224 when (G, W, O) = (0.2, 0.0, 0.8) and increased to 0.272 when (G, W, O) = (0.0, 0.2, 0.8). Other land surface parameters were also derived from each grid ratio.

Table 3.

Land surface parameters for forests, grasslands, and water surfaces.

Land surface parameters for forests, grasslands, and water surfaces.
Land surface parameters for forests, grasslands, and water surfaces.
Fig. 3.

Distribution of the evaporative efficiency for (left to right) three land surface distribution ratios: (grassland, water surface, original) = (0.2, 0.0, 0.8), (0.1, 0.1, 0.8), and (0.0, 0.2, 0.8).

Fig. 3.

Distribution of the evaporative efficiency for (left to right) three land surface distribution ratios: (grassland, water surface, original) = (0.2, 0.0, 0.8), (0.1, 0.1, 0.8), and (0.0, 0.2, 0.8).

3. Results

a. Reproducibility of the control run

We first verified the reproducibility of daily precipitation. Although a direct comparison of the latent heat flux is the best way to validate JMA-NHM, the flux was unavailable for the BMDS dataset. Thus, we verified the precipitation and the daily mean temperature, which are strongly related to the latent heat flux, instead of the latent heat flux itself.

Precipitation occurred in the mountain region (Fig. 4a). The overall lower-atmospheric response to changes in land surface parameters in the middle Lena basin was the main focus of our study, with an emphasis on the degree to which physical quantities changed as area-averaged values. The average precipitation at the 17 BMDS sites was 1.6 ± 3.1 mm day−1, whereas the calculated BMDS-interpolated precipitation was 1.1 ± 0.3 mm day−1. The bias of the calculated precipitation was −0.5 mm day−1, which fell within the standard deviation of observed BMDS precipitation, indicating that the calculated precipitation was almost equal to the BMDS precipitation.

Fig. 4.

(a) Distribution of daily precipitation (mm day−1). The shading indicates precipitation (mm day−1), and the black and blue contours show topography for 0 and 400 m MSL, respectively. The red points represent the BMDS sites. (b) Lines with circles: interdiurnal variation of daily mean temperature (°C); bars: daily precipitation (mm day−1) at the BMDS sites.

Fig. 4.

(a) Distribution of daily precipitation (mm day−1). The shading indicates precipitation (mm day−1), and the black and blue contours show topography for 0 and 400 m MSL, respectively. The red points represent the BMDS sites. (b) Lines with circles: interdiurnal variation of daily mean temperature (°C); bars: daily precipitation (mm day−1) at the BMDS sites.

Figure 4b shows the 31-day variability of the daily mean temperature and daily precipitation averaged over the BMDS sites. The variation of the daily mean temperature was captured well for JMA-NHM and was within a 2°C bias for most days. Although JMA-NHM failed to capture the variation in precipitation in the last half of the duration, the total averaged difference was within the standard deviation of BMDS, as shown in the previous paragraph. The 31-day site-averaged daily mean temperature and precipitation were 17.3°C and 1.6 mm day−1, respectively, for BMDS and 16.5°C and 1.1 mm day−1 for JMA-NHM.

b. Independent change in land surface parameters experiment

The simulated 31-day-averaged daily mean temperature and precipitation corresponded well to the BMDS values; thus, we examined the lower-atmospheric response caused by changes in the land surface parameters. Figure 5 shows the area-averaged meteorological quantities for each change in the land surface parameters below 250 m MSL. A similar trend was observed between the latent heat flux and precipitation (Figs. 5a,b), with a monotone increase or decrease corresponding to the changes in the land surface parameters under observation. Although parameters α and β appeared to have a strong impact on the latent heat flux and precipitation, direct assessment was impossible because each slope has dimension. We therefore introduced a normalized slope for each parameter with the following equation:

 
formula

where NPI is the normalized slope of a parameter and LEi and parmi (i = 1, 2) are the latent heat flux and the targeted surface parameter (the lowest parameter value and latent heat flux for i = 1, the highest parameter value and latent heat flux for i = 2). We designated the first fraction of Eq. (8) as the parameter impact (PI). The PI and NPI contained a land–atmosphere interaction effect because a three-dimensional atmospheric model was used to derive the PI.

Fig. 5.

The area-averaged quantities for each of the land surface parameter settings: (a) evaporation (left axis) and latent heat flux (right axis), (b) precipitation, (c) sensible heat flux, and (d) downward shortwave radiation. The horizontal axis shows the settings for each parameter: α signifies the surface albedo, β denotes the evaporative efficiency, z0 represents the roughness length, cρ indicates the heat capacity, and λ is the thermal conductivity.

Fig. 5.

The area-averaged quantities for each of the land surface parameter settings: (a) evaporation (left axis) and latent heat flux (right axis), (b) precipitation, (c) sensible heat flux, and (d) downward shortwave radiation. The horizontal axis shows the settings for each parameter: α signifies the surface albedo, β denotes the evaporative efficiency, z0 represents the roughness length, cρ indicates the heat capacity, and λ is the thermal conductivity.

The NPIs of the parameters α, β, z0, cρ, and λ were −0.20, 0.30, −0.07, −0.05, and −0.03, respectively. The evaporative efficiency β and surface albedo α exhibited larger absolute values of NPI than the other land surface parameters. Evaporative efficiency was the parameter with the greatest influence on latent heat flux, and surface albedo was the second most influential parameter with respect to the absolute values of NPI. The impacts of the other parameters—roughness length, heat capacity, and thermal conductivity—were one order of magnitude smaller than the impacts of the evaporative efficiency and surface albedo. Consequently, the latent heat flux was potentially defined by the evaporative efficiency and surface albedo.

A similar trend was observed for the precipitation response. By exchanging LE with precipitation in Eq. (8), the NPI for precipitation was derived as −0.11, 0.14, −0.03, −0.03, and −0.02 for α, β, z0, cρ, and λ, respectively. The NPI of precipitation in α and β were larger than other land surface parameters. To directly compare precipitation and latent heat flux, we converted latent heat flux to actual evaporation amount E and determined the response ratio of precipitation Pre to evaporation [ΔPre/ΔE = (Pre2 – Pre1)/(E2E1)]; the ratios were 0.78, 0.66, 0.55, 0.80, and 1.00 for α, β, z0, cρ, and λ, respectively. Because the average of the ratio was 0.75, 75% of the evaporation from the land surface was converted to precipitation. Direct comparison was difficult because this ratio simply takes precipitation/evaporation, which differs from the recycling ratio. This high evaporation/precipitation ratio would support a high recycling ratio in eastern Siberia, as found in previous studies (e.g., Serreze et al. 2002).

The sensible heat flux exhibited an opposite tendency of that of latent heat flux. From a heat-balance relationship, sensible heat flux decreased with increases in the latent heat flux (Fig. 5c). However, both fluxes decreased with increases in α. Incoming radiation, which is the difference between the sum of incoming radiation (downward shortwave and longwave radiation) and upward shortwave radiation, decreased with increase in α (485.9 and 436.4 W m−2 for α = 0.0 and 0.8, respectively), causing a reduction of the fluxes.

A negative tendency was calculated between downward shortwave radiation and β, whereas an unclear relationship was derived for z0, cρ, and λ (Fig. 5d). Downward shortwave radiation increased with increasing α, although it is not directly related to α. Reductions in evaporation and precipitation caused decreases in cloud amount and increases in downward shortwave radiation. These relationships are discussed in the water surface expansion experiment.

In general, the latent heat flux from the land surface is essentially attributed to land surface parameters and ambient conditions. From the perspective of ambient conditions, a previous study specifically examined which environment defines evaporation in eastern Siberia (Matsumoto et al. 2008). The results of the present study reveal which land surface parameters define the latent heat flux in eastern Siberia. Serreze et al. (2002), Dirmeyer and Brubaker (2007), and van der Ent and Savenije (2011) noted the importance of land surface evaporation of precipitation in summer eastern Siberia using reanalysis data. The results of the parameter sensitivity experiments as a modeling study support their work.

Table 4 also shows each NPI with additional calculation periods and variables. Although the meteorological conditions differed among the study periods and different variables, the NPI trends based on evaporative efficiency and surface albedo were dominant, and roughness length, heat capacity, and thermal conductivity had a small effect. These findings were confirmed for the July 2000 case and for the four other calculation periods. The order of magnitude of the NPI was the same among the four calculation durations for surface albedo and evaporative efficiency, which were the dominant components. A small seasonal–annual variation was observed for the NPI. Therefore, we used the July-2000-averaged PI, which is the prenormalized NPI, to calculate the latent heat flux of the water surface area expansion experiment because the value of the July-2000-averaged regional precipitation was closest to the climatological observations.

Table 4.

Summary of the NPI for the area-averaged latent heat flux (LE), precipitation (Pre), sensible heat flux (H), and downward shortwave radiation (Sd). Here Pr and Tm are the 17-BMDS-site-averaged daily precipitation and daily mean temperature, respectively. The duration of each calculation corresponds to the year–month. The calculation period ranges from the first to the 31st of each month.

Summary of the NPI for the area-averaged latent heat flux (LE), precipitation (Pre), sensible heat flux (H), and downward shortwave radiation (Sd). Here Pr and Tm are the 17-BMDS-site-averaged daily precipitation and daily mean temperature, respectively. The duration of each calculation corresponds to the year–month. The calculation period ranges from the first to the 31st of each month.
Summary of the NPI for the area-averaged latent heat flux (LE), precipitation (Pre), sensible heat flux (H), and downward shortwave radiation (Sd). Here Pr and Tm are the 17-BMDS-site-averaged daily precipitation and daily mean temperature, respectively. The duration of each calculation corresponds to the year–month. The calculation period ranges from the first to the 31st of each month.

c. Water surface area expansion experiment

1) JMA-NHM method

We show the results of the water surface expansion experiments with JMA-NHM, as described in section 2c. Figure 6 shows the latent heat flux distributions for (G, W, O) = (0.2, 0.0, 0.8), (0.0, 0.2, 0.8) and their differences. With water surface expansion, the increased latent heat flux exhibited a patchy geographical distribution. The increase in the latent heat flux spread over the lowlands. The latent heat flux reached more than 70 W m−2 in the central Lena basin. However, a low heat flux value was found in the center part of the middle Lena basin because of the small net radiation, corresponding to an area of precipitation.

Fig. 6.

Latent heat flux distribution for each land surface distribution ratio: (a) (grassland, water surface, original) = (0.2, 0.0, 0.8) and (b) (grassland, water surface, original) = (0.0, 0.2, 0.8). (c) Difference for (b) − (a). The unit for each of the latent heat fluxes is watts per meter squared.

Fig. 6.

Latent heat flux distribution for each land surface distribution ratio: (a) (grassland, water surface, original) = (0.2, 0.0, 0.8) and (b) (grassland, water surface, original) = (0.0, 0.2, 0.8). (c) Difference for (b) − (a). The unit for each of the latent heat fluxes is watts per meter squared.

Figure 7 shows the area-averaged meteorological quantity changes that occur with land cover changes. The latent heat flux or evaporation amount exhibits a positive trend with water surface expansion (Fig. 7a). The alas grid used in the calculations was chosen using random number groups. Few differences in the latent heat flux were observed among the four random number groups. Water surface expansion had a stronger effect than random number differences. The area-averaged latent heat flux was 44.7 ±0.04 W m−2 when (G, W, O) = (0.2, 0.0, 0.8) for the ensemble mean. It increased to 45.5 ±0.03 W m−2 when the water surface occupies 20% of lowland areas. The amount of latent heat flux change was 0.8 W m−2, which is larger than the standard deviation of the (0.2, 0.0, 0.8) and (0.0, 0.2, 0.8) experiment. This amount, 0.8 W m−2, was 1.8% of the latent heat flux of the 20% grassland experiment.

Fig. 7.

Area-averaged values for land cover changes from grassland to a water surface: (a) evaporation (latent heat flux) vs precipitation, (b) latent heat flux vs downward shortwave radiation, (c) latent heat flux vs cloud amount, and (d) low-level cloud amount vs precipitation. Each “Ex” signifies the difference between the random number groups. The four point groups in each figure correspond to the ratio for (grassland, water surface, original) = (0.2, 0.0, 0.8), (0.1, 0.1, 0.8), and (0.0, 0.2, 0.8) for the left, center, and right groups, respectively.

Fig. 7.

Area-averaged values for land cover changes from grassland to a water surface: (a) evaporation (latent heat flux) vs precipitation, (b) latent heat flux vs downward shortwave radiation, (c) latent heat flux vs cloud amount, and (d) low-level cloud amount vs precipitation. Each “Ex” signifies the difference between the random number groups. The four point groups in each figure correspond to the ratio for (grassland, water surface, original) = (0.2, 0.0, 0.8), (0.1, 0.1, 0.8), and (0.0, 0.2, 0.8) for the left, center, and right groups, respectively.

A positive relationship between latent heat flux and precipitation was also observed in the water surface expansion experiment (correlation coefficient = 0.95, Fig. 7a). By contrast, latent heat flux and downward shortwave radiation had a negative relationship (Fig. 7b); an increasing amount of low-level clouds (<850 hPa) led to a decrease in downward shortwave radiation (Fig. 7c). Thus, a positive correlation coefficient (0.91) was observed for low-level cloud amount and precipitation (Fig. 7d). These results demonstrated that land surface change affects the lower atmosphere intensely and that a three-dimensional atmospheric model such as JMA-NHM is desirable for evaluating the effects of land surface change on the lower atmosphere in eastern Siberia.

2) Parameter impact method

Presuming linearity between the latent heat flux and the changes in the five land surface parameters, the latent heat flux change can be written as the product of PI and the changes in the parameters with land cover change as follows:

 
formula

where ΔLE is the latent heat flux difference from the control run (G, W, O) = (0.2, 0.0, 0.8). The partial derivative coefficients, , where “para” denotes surface albedo, evaporative efficiency, roughness length, heat capacity, and thermal conductivity, are given based on the PIs. For latent heat flux and precipitation, the area-averaged PIs and land surface parameter values associated with the land surface distribution (G, W, O) = (0.2, 0.0, 0.8) and (0.0, 0.2, 0.8) are presented in Table 5.

Table 5.

Area-averaged land surface parameters when (G, W, O) = (0.2, 0.0, 0.8) and (0.0, 0.2, 0.8). The LE and Pre are the latent heat flux and precipitation, respectively. Here, a–f are column labels.

Area-averaged land surface parameters when (G, W, O) = (0.2, 0.0, 0.8) and (0.0, 0.2, 0.8). The LE and Pre are the latent heat flux and precipitation, respectively. Here, a–f are column labels.
Area-averaged land surface parameters when (G, W, O) = (0.2, 0.0, 0.8) and (0.0, 0.2, 0.8). The LE and Pre are the latent heat flux and precipitation, respectively. Here, a–f are column labels.

Table 5 shows the PI and NPI for latent heat flux and precipitation. There was a similar tendency among the five land surface parameters; evaporative efficiency had the largest impact on the Lena-basin-averaged latent heat flux with water surface expansion. Surface albedo and evaporative efficiency exhibited a larger absolute value of NPI and with respect to the latent heat flux in the land surface parameter sensitivity experiments. However, evaporative efficiency was the only dominant parameter in the water surface expansion experiment. Surface albedo was not selected as an influential parameter because the range of surface albedo change was smaller than the change in the evaporative efficiency with water surface expansion (−0.006). Large absolute values of NPI and and a small parameter change were observed for surface albedo.

The sum of all the effects on latent heat flux change were 1.2 W m−2, indicating a 0.4 W m−2 overestimation with the JMA-NHM method; this method was discussed in the previous section. Similarly, the calculated daily precipitation of 0.026 mm day−1 represented a 0.01 mm day−1 overestimation by the JMA-NHM method. Although we used a simple estimation method for the calculations, a similar order of magnitude was derived via the JMA-NHM method. The PI included the effect of the land–atmosphere interaction because it was derived with an online model of the JMA-NHM method. This result raises the possibility that the parameter impact method may be used to estimate the changes in the latent heat flux and other elements for a given land cover change. Taking deforestation and water surface expansion as examples, we discuss the change in the latent heat flux and other meteorological quantities associated with land cover change with the parameter impact method in the next section.

4. Discussion

On the basis of previous results, the area-averaged latent heat flux and precipitation can be calculated using the parameter impact method and various land surface data. Figure 8 presents a schematic diagram of the land surface distribution ratio below 250 m MSL. “Original” signifies forest in the southern area and bare soil in the northern area (see Fig. 2). The land cover changes from original to grassland or a water surface. For simplicity, the land cover change from original was designated as “deforestation” in this study. “WF” denotes the water fraction of the sum of the water surface area and the grassland area. WF = 0.0 indicates that all of the deforested area changes to grassland; WF = 1.0 denotes a change to a water surface. Land surface data were obtained for each land surface ratio. The water surface expansion experiment in the previous section corresponds to the red arrow in Fig. 8.

Fig. 8.

Schematic diagram of the land cover change. The land surface takes values of (grassland, water surface, original) = (0.0, 0.0, 1.0) if there is no land cover change. The WF signifies the water surface fraction of the sum of the water surface and grassland. The area changed from the original land surface becomes grassland only if WF = 0.0. The circled point at which (grassland, water surface, original) = (0.2, 0.0, 0.8) denotes the control run (see text for details).

Fig. 8.

Schematic diagram of the land cover change. The land surface takes values of (grassland, water surface, original) = (0.0, 0.0, 1.0) if there is no land cover change. The WF signifies the water surface fraction of the sum of the water surface and grassland. The area changed from the original land surface becomes grassland only if WF = 0.0. The circled point at which (grassland, water surface, original) = (0.2, 0.0, 0.8) denotes the control run (see text for details).

Considering that the alas area occupies 18% of the lowlands of central Yakutia (Lopez et al. 2008) and that most of the alas is normally covered with grassland, (G, W, O) = (0.2, 0.0, 0.8) was defined as the control run. In this case, the area-averaged evaporative efficiency was 0.23. We calculated the change in the latent heat flux for various land cover ratios using the parameter sensitivity method, as described in section 3c. For example, when (G, W, O) was (0.6, 0.2, 0.2), the area-averaged evaporative efficiency was 0.24. Consequently, the difference in evaporative efficiency between the (G, W, O) = (0.6, 0.2, 0.2) run and the control run was +0.01, which was equal to Δβ in Eq. (9). The PI of evaporative efficiency, , was found to be +25.9 W m−2 in the parameter sensitivity experiments (Table 5). Therefore, the total effect of evaporative efficiency was derived as (+0.01) × (+25.9) = +0.259 W m−2. Using this process, the latent heat flux change from the control run ΔLE was calculated for the other parameters and land surface distributions.

Figures 9a and 9b show the area-averaged latent heat flux and evaporative efficiency change from the control run for each land surface cover change. The area-averaged values for the control run were 44.7 W m−2 for the latent heat flux and 0.23 for the evaporative efficiency. Few changes in the latent heat flux and evaporative efficiency were observed when WF = 0.0; the latent heat flux slightly decreased with the increase in evaporative efficiency (Fig. 9b). This resulted from a negative effect of the other parameters. The total effects of each parameter for the original area ratio = 0.0 and WF = 0.0 were as follows: surface albedo, −0.33 W m−2; evaporative efficiency, +0.68 W m−2; roughness length, −0.24 W m−2; heat capacity, −0.19 W m−2; and thermal conductivity, −0.17 W m−2. The sum of these effects was −0.25 W m−2. No water surface existed for WF = 0.0. Therefore, the effect of evaporative efficiency was weak, and effect of other land surface parameter was significant. The decrease in the latent heat flux associated with deforestation has been studied in the Indochina Peninsula (Kanae et al. 2001), China (Fu 2003), and the Amazon (Shukla et al. 1990; Nobre et al. 1991; Hahmann and Dickinson 1997). These studies mainly focused on the surface albedo and roughness length changes associated with deforestation. However, other parameters, such as evaporative efficiency, were on the same order of magnitude as the latent heat flux change. In this study, the sum of the total impact derived from each land surface parameter was small for land cover changes not involving water surfaces (WF = 0.0). Consequently, when taiga forest in the southern part of the Lena basin and bare soil in the northern part changed to grassland, the Lena-basin-averaged latent heat flux did not change significantly.

Fig. 9.

(a) Area-averaged latent heat flux change from the control run (grassland, water surface, original) = (0.2, 0.0, 0.8). The red point indicates the control run. (b)–(f) As in (a), but for evaporative efficiency, precipitation, precipitation/evaporation ratio, downward shortwave radiation, and sensible heat flux, respectively. The ratio ΔPre/ΔE is undefined when WF = 0.0 and the original area ratio = 0.8 because ΔE is zero.

Fig. 9.

(a) Area-averaged latent heat flux change from the control run (grassland, water surface, original) = (0.2, 0.0, 0.8). The red point indicates the control run. (b)–(f) As in (a), but for evaporative efficiency, precipitation, precipitation/evaporation ratio, downward shortwave radiation, and sensible heat flux, respectively. The ratio ΔPre/ΔE is undefined when WF = 0.0 and the original area ratio = 0.8 because ΔE is zero.

However, the latent heat flux was significantly greater when the land surface includes a water surface. The latent heat flux decreased by 0.25 W m−2 from the control run to (G, W, O) = (1.0, 0.0, 0.0). The latent heat flux was 0.31 W m−2 from WF = 0.0 to 0.2 when the forest area ratio was 0.8. There was only a 4% change of grassland to a water surface, but the degree of latent heat flux change was greater than that for the land cover change in the control run, in which 100% of the lowland was occupied by grassland.

Other meteorological quantities could be calculated in the same manner as latent heat flux. Although the absolute value is smaller than that of evaporation, an increased tendency for precipitation with water surface expansion was also calculated (Fig. 9c). Taking a ratio of evaporation and precipitation response (ΔPre/ΔE) to evaluate the effect of land surface change on precipitation, the extent to which the ratio increased was mitigated by the WF increase (Fig. 9d). For the original area ratio = 0.0, 0.2, and 0.5, the size of the increase decreased from WF = 0.2 to 1.0, indicating that the water cycle in eastern Siberia is enhanced more intensely by water surface expansion of small expanding water areas than by large expanding areas. The value of the ratio was largest for WF = 0.0. However, the ΔE was close to 0 for WF = 0.0; thus, the value of the ratio was at its largest because ΔE was in the denominator.

The response of downward shortwave radiation and sensible heat flux was basically opposite of that of evaporation or latent heat flux and precipitation (Figs. 9e,f). Increased downward shortwave radiation (Fig. 9e) and decreased evaporation (Fig. 9a) were calculated for WF = 0.0, assuming that sensible heat flux increased from the heat-balance relationship; however, the sensible heat flux decreased with decreased original area. The fraction WF = 0.0 indicates that the decreased area (mainly forest in the southern area) converts to grassland only; thus, upward shortwave radiation increased, and incoming radiation decreased, because α increased after land surface change. Upward shortwave radiation increased 1.5 W m−2 and incoming radiation decreased 1.1 W m−2 for WF = 0.0 (α is 0.09 and 0.15 for forest and grassland, respectively).

Consequently, the land cover change to grassland had no significant impact on the area-averaged latent heat flux; however, land cover change involving a water surface enhances the latent heat flux intensely. In particular, the ratio ΔPre/ΔE increased strongly for small WF, indicating that water surface expansion in a small water area enhances the water cycle, in contrast to expansion in a large water surface area. Figure 9 shows the lower atmosphere response caused by land cover changes under the present climate.

Although we calculated the lower-atmospheric response on the basis of the land cover change for the present climate without permafrost because the analysis was conducted during the permafrost melting season, introducing permafrost and its change into the derivation of PI was essential to obtaining a deeper understanding of the water-energy flux and precipitation changes in eastern Siberia. Tchebakova et al. (2009) simulated vegetation change in a changing climate with their bioclimatic model, which explicitly considers permafrost. They demonstrated a decrease in Siberian forest and a shift northward, transforming original forest area to forest-steppe and steppe. According to their study, thawing permafrost caused a transition in vegetation, indicating that land cover change is strongly connected to permafrost change. Moreover, the land cover change affects the lower atmosphere through land surface parameters. Consequently, permafrost changes should be considered in future research.

5. Conclusions

The impact of land cover changes on the lower atmosphere was discussed in this study. The land surface parameter change experiment demonstrated that 75% of the evaporation change corresponded to precipitation changes on interparameter average, indicating strong land–atmosphere coupling in the eastern Siberian water cycle. The strong land–atmosphere coupling is responsible for the high water cycling ratio reported in previous studies (e.g., Serreze et al. 2002).

On the basis of the results of the parameter sensitivity experiment, the lower atmosphere response associated with land cover change was also evaluated. The increase in the rate of ΔPre/ΔE was weak for large expanding water surface areas. By contrast, the rate was intense for small expanding water surface areas in eastern Siberia. Land surface change such as grassland formation from forest through forest fire yielded little enhancement of evaporation and precipitation; however, if water surface formation was caused by an increase in incoming downward shortwave radiation to the land surface because of the disappearance of the forest shading effect and the melting of permafrost, the water cycle would be enhanced intensely. Moreover, the calculation of the latent heat flux with the parameter impact method yielded a similar value to that obtained when using the JMA-NHM method, implying that parameter impact method could be used to estimate the latent heat flux change caused by a given land cover change.

The potential lower-atmospheric response associated with land cover change is indicated by the results of our study. Agafonov et al. (2004) insisted that thermokarst expansion resulted from elevated precipitation, not temperature increases. Evaporation strongly affects precipitation in eastern Siberia. Thus, there appears to be positive feedback between evaporation and land cover change. The explicit treatment of thermokarst expansion will be essential for further studies. Moreover, these sensitivity experiments were conducted under present climatic conditions. Numerical calculations introducing a water surface expansion model, such as that presented by van Huissteden et al. (2011), and a dynamic vegetation model incorporating permafrost (e.g., Tchebakova et al. 2009) based on future prediction data are necessary to conduct more realistic numerical calculations.

Acknowledgments

This study was supported by Research Project C-07 of the Research Institute for Humanity and Nature (RIHN), entitled “Global Warming and the Human-Nature Dimension in Siberia: Social Adaptation to the Changes of the Terrestrial Ecosystem, with an Emphasis on Water Environments” (PI: Tetsuya Hiyama). The simulations were conducted using the Japan Meteorological Agency Nonhydrostatic Model developed by the Meteorological Research Institute and the Numerical Prediction of the Japan Meteorological Agency. Some of the experimentally obtained results in this research were obtained using the supercomputing resources of the Cyberscience Center at Tohoku University. We thank Professor Toshiki Iwasaki, Dr. Weming Sha, and Mr. Shota Ishii of Tohoku University for their useful and constructive comments. Our thanks are extended to the editor and three anonymous reviewers for their many informative and invaluable comments.

REFERENCES

REFERENCES
Achard
,
F.
,
D.
Mollicone
,
H.-J.
Stibig
,
D.
Aksenov
,
L.
Laestadius
,
Z.
Li
,
P.
Poporov
, and
A.
Yaroshenko
,
2006
:
Areas of rapid forest-cover change in boreal Eurasia
.
For. Ecol. Manage.
,
237
,
322
334
.
Agafonov
,
L.
,
H.
Strunk
, and
T.
Nuber
,
2004
:
Thermokarst dynamics in Western Siberia: Insights from dendrochronological research
.
Palaeogeogr., Palaeoclimatol., Palaeoecol.
,
209
,
183
196
.
Desyatkin
,
A. R.
,
F.
Takakai
,
P. P.
Fedorov
,
M. C.
Nikolaeva
,
R. V.
Desyatkin
, and
R.
Hatano
,
2009
:
CH4 emission from different stages of thermokarst formation in central Yakutia, east Siberia
.
Soil Sci. Plant Nutr.
,
55
,
558
570
.
Dirmeyer
,
P. A.
, and
K. L.
Brubaker
,
2007
:
Characterization of the global hydrologic cycle from a back-trajectory analysis of atmospheric water vapor
.
J. Hydrometeor.
,
8
,
20
37
.
Fu
,
C.
,
2003
:
Potential impacts of human-induced land cover change in East Asia monsoon
.
Global Planet. Change
,
37
,
219
229
.
Hahmann
,
A. N.
, and
R. E.
Dickinson
,
1997
:
RCCM2-BATS model over tropical South America: Applications to tropical deforestation
.
J. Climate
,
10
,
1944
1964
.
Iijima
,
Y.
,
A. N.
Fedorov
,
H.
Park
,
K.
Suzuki
,
H.
Yabuki
,
T. C.
Maximov
, and
T.
Ohata
,
2010
:
Abrupt increases in soil temperatures following increased precipitation in a permafrost region, central Lena River basin, Russia
.
Permafrost Periglacial Processes
,
21
,
30
41
.
Kain
,
J. S.
,
2004
:
The Kain–Fritch convective parameterization: An update
.
J. Appl. Meteor.
,
43
,
170
181
.
Kanae
,
S.
,
T.
Oki
, and
K.
Musiake
,
2001
:
Impact of deforestation on regional precipitation over the Indochina peninsula
.
J. Hydrometeor.
,
2
,
51
70
.
Kondo
,
J.
,
1994
: Meteorology of Water Environment. Asakura Publishing, 348 pp.
Lepers
,
E.
,
E. F.
Lambin
,
A. C.
Janetos
,
R.
Defries
,
F.
Achard
,
N.
Ramankutty
, and
R. J.
Scholes
,
2005
:
A synthesis of information on rapid land-cover change for the period 1981–2000
.
BioScience
,
55
,
115
124
.
Lopez
,
N. L.
,
C. E.
Gerasimov
,
T.
Machimura
,
F.
Takaaki
,
G.
Iwahana
,
A. N.
Fedorov
, and
M.
Fukuda
,
2008
:
Comparison of carbon and water vapor exchange of forest and grassland in permafrost regions, central Yakutia, Russia
.
Agric. For. Meteor.
,
148
,
1968
1977
.
Loveland
,
T. R.
,
B. C.
Reed
,
J. F.
Brown
,
D. O.
Ohlen
,
Z.
Zhu
,
L.
Yang
, and
J. W.
Merchant
,
2000
:
Development of a global land cover characteristics database and IGBP DISCover from 1 km AVHRR data
.
Int. J. Remote Sens.
,
21
,
1303
1330
.
Mabuchi
,
K.
,
Y.
Sato
, and
H.
Kida
,
2005a
:
Climatic impact of vegetation change in the Asia tropical region. Part I: Case of the Northern Hemisphere summer
.
J. Climate
,
18
,
410
428
.
Mabuchi
,
K.
,
Y.
Sato
, and
H.
Kida
,
2005b
:
Climatic impact of vegetation change in the Asia tropical region. Part II: Case of the Northern Hemisphere winter and impact on the extratropical circulation
.
J. Climate
,
18
,
429
446
.
Matsumoto
,
L.
, and
Coauthors
,
2008
:
Energy consumption and evapotranspiration at several boreal and temperature forests in the Far East
.
Agric. For. Meteor.
,
148
,
1978
1989
.
Nakanishi
,
M.
, and
H.
Niino
,
2004
:
An improved Mellor-Yamada level-3 model with condensation physics: Its design and verification
.
Bound.-Layer Meteor.
,
112
,
1
31
.
Nobre
,
C. A.
,
P. J.
Sellers
, and
J.
Shukla
,
1991
:
Amazonian deforestation and regional climate change
.
J. Climate
,
4
,
957
988
.
Ohta
,
T.
, and
Coauthors
,
2008
:
Interannual variation of water balance and summer evapotranspiration in an eastern Siberian larch forest over a 7-year period (1998–2006)
.
Agric. For. Meteor.
,
148
,
1941
1953
.
Onogi
,
K.
, and
Coauthors
,
2007
:
The JRA-25 Reanalysis
.
J. Meteor. Soc. Japan
,
85
,
369
432
.
Potapov
,
P.
,
M. C.
Hansen
,
S. V.
Stephan
,
T. R.
Loveland
, and
K.
Pittman
,
2008
:
Combining MODIS and Landsat imagery to estimate and map boreal forest cover loss
.
Remote Sens. Environ.
,
112
,
3708
3719
.
Saito
,
K.
,
J.
Ishida
,
K.
Aranami
,
T.
Hara
,
T.
Segawa
,
M.
Narita
, and
Y.
Honda
,
2007
:
Nonhydrostatic atmospheric models and operational development at JMA
.
J. Meteor. Soc. Japan
,
85B
,
271
304
.
Sen
,
O. L.
,
Y.
Wang
, and
B.
Wang
,
2004
:
Impact of Indochina deforestation on the East Asian summer monsoon
.
J. Climate
,
17
,
1366
1380
.
Serreze
,
M. C.
,
D. H.
Bromwich
,
M. P.
Clark
,
A. J.
Etringer
,
T.
Zhang
, and
R.
Lammers
,
2002
:
Large-scale hydro-climatology of the terrestrial Arctic drainage system
.
J. Geophys. Res.
,
107
,
8160
,
doi:10.1029/2001JD000919
.
Shukla
,
J.
,
C.
Nobre
, and
P.
Sellers
,
1990
:
Amazon deforestation and climate change
.
Science
,
247
,
1322
1325
.
Suzuki
,
R.
,
V. N.
Razuvaev
,
O. N.
Bulygina
, and
T.
Ohata
,
2006
: Baseline meteorological data in Siberia version 4. Institute of Observational Research for Global Change, Japan Agency for Marine-Earth Science and Technology. [Available online at http://www.jamstec.go.jp/iorgc/cgi-bin/database/v01/browse_summary.cgi?program=hcorp&group=CDC&cat=Russia&id=BMDS_41.]
Tchebakova
,
N. M.
,
E.
Parfenova
, and
A. J.
Soja
,
2009
:
The effects of climate, permafrost and fire on vegetation change in Siberia in a changing climate
.
Environ. Res. Lett.
,
4
,
045013
,
doi:10.1088/1748-9326/4/4/045013
.
U.S. Geological Survey
,
1996
: GTOPO30 global digital elevation model. EROS Data Center. [Available online at http://www1.gsi.go.jp/geowww/globalmap-gsi/gtopo30/gtopo30.html.]
van der Ent
,
R. J.
, and
H. H. G.
Savenije
,
2011
:
Length and time scales of atmospheric moisture recycling
.
Atmos. Chem. Phys.
,
11
,
1853
1863
.
van Huissteden
,
J.
,
C.
Berrittella
,
F. J. W.
Parmentier
,
Y.
Mi
,
T. C.
Maximov
, and
A. J.
Dolman
,
2011
:
Methane emissions from permafrost thaw lakes limited by lake drainage
.
Nat. Climate Change
,
1
,
119
123
.

Footnotes

*

Current affiliation: Department of Geophysics, Graduate School of Science, Tohoku University, Sendai, Japan.