Local and Nonlocal Impacts of Soil Moisture Initialization on AGCM Seasonal Forecasts: A Model Sensitivity Study

H. Zhang Bureau of Meteorology Research Centre, Melbourne, Australia

Search for other papers by H. Zhang in
Current site
Google Scholar
PubMed
Close
and
C. S. Frederiksen Bureau of Meteorology Research Centre, Melbourne, Australia

Search for other papers by C. S. Frederiksen in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Using a version of the Australian Bureau of Meteorology Research Centre (BMRC) atmospheric general circulation model, this study investigates the model's sensitivity to different soil moisture initial conditions in its dynamically extended seasonal forecasts of June–August 1998 climate anomalies, with focus on the south and northeast China regions where severe floods occurred. The authors' primary aim is to understand the model's responses to different soil moisture initial conditions in terms of the physical and dynamical processes involved. Due to a lack of observed global soil moisture data, the efficacy of using soil moisture anomalies derived from the NCEP–NCAR reanalysis is assessed. Results show that by imposing soil moisture percentile anomalies derived from the reanalysis data into the BMRC model initial condition, the regional features of the model's simulation of seasonal precipitation and temperature anomalies are modulated. Further analyses reveal that the impacts of soil moisture conditions on the model's surface temperature forecasts are mainly from localized interactions between land surface and the overlying atmosphere. In contrast, the model's sensitivity in its forecasts of rainfall anomalies is mainly due to the nonlocal impacts of the soil moisture conditions. Over the monsoon-dominated east Asian region, the contribution from local water recycling, through surface evaporation, to the model simulation of precipitation is limited. Rather, it is the horizontal moisture transport by the regional atmospheric circulation that is the dominant factor in controlling the model rainfall. The influence of different soil moisture conditions on the model forecasts of rainfall anomalies is the result of the response of regional circulation to the anomalous soil moisture condition imposed. Results from the BMRC model sensitivity study support similar findings from other model studies that have appeared in recent years and emphasize the importance of improving the land surface data assimilation and soil hydrological processes in dynamically extended GCM seasonal forecasts.

Corresponding author address: Dr. Huqiang Zhang, Bureau of Meteorology Research Centre, GPO Box 1289K, VIC 3001, Australia. Email: h.zhang@bom.gov.au

Abstract

Using a version of the Australian Bureau of Meteorology Research Centre (BMRC) atmospheric general circulation model, this study investigates the model's sensitivity to different soil moisture initial conditions in its dynamically extended seasonal forecasts of June–August 1998 climate anomalies, with focus on the south and northeast China regions where severe floods occurred. The authors' primary aim is to understand the model's responses to different soil moisture initial conditions in terms of the physical and dynamical processes involved. Due to a lack of observed global soil moisture data, the efficacy of using soil moisture anomalies derived from the NCEP–NCAR reanalysis is assessed. Results show that by imposing soil moisture percentile anomalies derived from the reanalysis data into the BMRC model initial condition, the regional features of the model's simulation of seasonal precipitation and temperature anomalies are modulated. Further analyses reveal that the impacts of soil moisture conditions on the model's surface temperature forecasts are mainly from localized interactions between land surface and the overlying atmosphere. In contrast, the model's sensitivity in its forecasts of rainfall anomalies is mainly due to the nonlocal impacts of the soil moisture conditions. Over the monsoon-dominated east Asian region, the contribution from local water recycling, through surface evaporation, to the model simulation of precipitation is limited. Rather, it is the horizontal moisture transport by the regional atmospheric circulation that is the dominant factor in controlling the model rainfall. The influence of different soil moisture conditions on the model forecasts of rainfall anomalies is the result of the response of regional circulation to the anomalous soil moisture condition imposed. Results from the BMRC model sensitivity study support similar findings from other model studies that have appeared in recent years and emphasize the importance of improving the land surface data assimilation and soil hydrological processes in dynamically extended GCM seasonal forecasts.

Corresponding author address: Dr. Huqiang Zhang, Bureau of Meteorology Research Centre, GPO Box 1289K, VIC 3001, Australia. Email: h.zhang@bom.gov.au

1. Introduction

This paper presents complementary experimental results to the study of Frederiksen et al. (2001), which reported the results of dynamical seasonal forecasts using a version of the Australian Bureau of Meteorology Research Centre (BMRC) atmospheric general circulation model (AGCM). The study of Frederiksen et al. (2001) was focused on the role of sea surface temperature (SST) forcing the model seasonal integrations. This study aims to assess whether, and how, different land surface conditions can modulate the model forecasts.

Dynamically extended seasonal forecasts from AGCMs have been reported in many publications in recent years (e.g., Kumar et al. 1996; Fennessy and Shukla 1999; Brankovic and Palmer 2000; Shukla et al. 2000; Frederiksen et al. 2001). The scientific basis for conducting such model forecasts is that slow varying atmospheric boundary forcing could allow for predictability of climate that potentially extends for seasonal and longer timescales—excessively beyond the predictability determined by the atmospheric dynamics alone (Charney and Shukla 1981). In most experiments, oceanic boundary forcing is provided either by using statistically forecasted SST conditions or those predicted by fully coupled ocean–atmosphere general circulation models (OAGCMs). A detailed review of current approaches used in seasonal forecasts appears in Goddard et al. (2000).

Generally, dynamical models can replicate some of large-scale atmospheric circulation anomalies, but there is a lack of coherent model skill in the forecasts of surface temperature and precipitation anomalies (Goddard et al. 2000). Even when forced with observed SSTs, models are not yet skillful in reproducing the observed climate anomalies (Frederiksen et al. 2001). Efforts are continually being made to study the factors limiting model predictions of climate anomalies on seasonal and longer timescales (Goddard et al. 2000). A large part of such studies is dedicated to investigating the role of global or regional SST anomalies in determining the model skill when simulating the observed climate anomalies. In addition, in recent years we have seen model sensitivity studies to assess the impacts of land surface conditions, another part of the atmospheric boundary, on GCM seasonal forecasts of weather and climate anomalies.

It is well known that there are strong interactions between the land surface and the overlying atmosphere by mass and energy exchanges. Sensitivity studies (e.g., Shukla and Mintz 1982; Yeh et al. 1984; Sud and Fennessy 1984; Sud et al. 2001) have shown that model-simulated surface climate variables, such as precipitation and temperature, are strongly influenced by land surface conditions. Following these earlier studies, a number of model simulations (e.g., Oglesby and Erickson 1989; Oglesby 1991; Fennessy et al. 1994; Bonan and Stillwell-Soller 1998) have found that soil moisture conditions have significant impacts on the persistence and strength of a number of pronounced regional drought and wet events. In recent years, a number of studies have investigated the role of land surface conditions on AGCM seasonal forecasts. Fennessy and Shukla (1999) reported a study showing the impacts of initial soil moisture conditions on an atmospheric model's seasonal predictions. The study from Douville and Chauvin (2000) and Douville (2002) also demonstrated the importance of improving the initialization of land surface conditions in dynamical model seasonal predictions. As noted by Fennessy and Shukla (1999), the influence of land surface conditions on the model forecasts may be model dependent and different processes may be involved. This issue therefore needs further investigation. This has been the aim of this study, with the focus being not only on assessing the BMRC model sensitivities to different soil moisture initial conditions but also on trying to explore how land surface conditions affect the model behavior in terms of its physical and dynamical processes.

These physical and dynamical processes can be of high complexity. Fundamentally, soil moisture largely determines the partition of surface radiative energy into surface sensible and latent heat fluxes, therefore affecting water and energy exchanges between the land surface and the atmosphere. Keeping other factors unchanged, wet soil leads to more surface evapotranspiration and less sensible heat to the atmosphere. There are other factors that complicate the process, such as cloud radiative forcing and water vapor transport. For instance, more evaporation leads to more localized water recycling, which yields more moist air, more precipitation, and more evaporation. On the other hand, it is also possible that more evaporation and more localized water recycling can give more moist air, more clouds, less surface solar radiation and, therefore, less evaporation. In addition, more evaporation leads to less sensible heat (with the same surface radiation), which progressively yields near-surface cooling, atmospheric subsidence, horizontal divergence with high pressure (thermal induced), less incoming water vapor advection, less rainfall, and, finally, less evaporation. Therefore, the impacts of soil moisture on the model simulations may have different features at different locations in different seasons and in different models.

In this paper, we present a BMRC model case study to further demonstrate the importance of soil moisture initial conditions in AGCM seasonal forecasts. We have selected the severe Chinese flooding that occurred in June–July–August 1998 (JJA98) as our case study, for two primary reasons. One is that anomalous land surface conditions over the Tibetan plateau and southern China region were identified as one of the major causes of the flooding (Chinese National Climate Center 1998). The other is that this event happened in the mature phase of the strong 1997/98 El Niño period when the global atmospheric circulation response to the strong tropical SST anomalies was well established. Thus, it allows us to assess the relative importance of soil moisture process in the model predictions when significant SST anomalies coexist. In the summer of 1998, China experienced extremely severe flooding in the Yangtze River region and record-breaking flooding in the northeast China region (see Fig. 4d later). The total affected area from these two flooding events was about 21.2 million hectares, with 223 million people affected and 3004 fatalities (Chinese National Climate Center 1998).

This paper is structured as follows. Section 2 will briefly discuss the atmospheric model used in this study and the experiments conducted. Section 3 shows the model sensitivity to different soil moisture initial conditions. The relative roles of soil moisture anomalies and SST forcing are also discussed in this section. Section 4 explores physical and dynamical processes in analyzing the model results. Conclusions and discussions drawn from this study are presented in section 5.

2. Model description and experimental design

The global AGCM used in this study is a version of the BMRC climate model with rhomboidal R31 horizontal resolution and 17-sigma-level vertical representation. This is the same model employed in the trial AGCM seasonal forecasts reported by Frederiksen et al. (2001). The model reported in this sensitivity study uses a modified Kuo scheme (Kuo 1974) with shallow convection parameterized in terms of the model's vertical diffusion process. In the model, soil temperature is based on heat storage from three layers with a zero flux assumed in the calculation of soil temperature in the bottom layer. A single-layer “bucket” model (Manabe and Holloway 1975) is used to represent soil moisture storage. Although vegetation is not explicitly included in the bucket model, surface albedo and roughness length vary over land according to land use and the vegetation type of Wilson and Henderson-Sellers (1985). Snow fraction is calculated in the model as a function of snow depth and local roughness length of the vegetation. It affects surface albedo, roughness length, and evaporation. Both precipitation and snowmelt contribute to the soil moisture calculation. The evapotranspiration efficiency is a function of the ratio of soil moisture to the field capacity. Runoff occurs if this ratio exceeds unity. More details of the BMRC model dynamics and physics are given in Colman and McAvaney (1995), McAvaney and Hess (1996), and references therein.

Zhang and Frederiksen (2001) published the model results over the trial period (March 1997–March 1998), with the focus on the model performance over the east Asian region. Model forecasts showed some skill in each individual season over different regions. For instance, the model did successfully predict the severe flooding in part of the northeast China region in the summer of 1998 (see Fig. 4a later), but it did not capture the severe flooding in the Yangtze River region in south China. Overall, results from Zhang and Frederiksen (2001) showed that the model forecasts lacked systematic skill in predicting seasonal climate anomalies over the east Asian region. The details of the model performance over the east Asian region have also been demonstrated in the calculated skill scores of mean sea level pressure, surface temperature, and precipitation forecasts for the period of April 1997–May 1998 in Frederiksen et al. (2001).

Besides the unpredictable chaotic components in the weather and climate system, there are a number of possible factors that could affect the model skill. One is that the SST forcing in the model is not perfect. Even though Frederiksen et al. (2001) showed that persisting SST anomalies gives reasonably good estimation of SST variation in the Tropics three months ahead, midlatitude SST variations are less successfully estimated by simple SST persistence. The other is that land surface boundary conditions are not well represented in the model. In the experiments in Frederiksen et al. (2001) and Zhang and Frederiksen (2001), the soil moisture climatology of Mintz and Serafini (1989) was used to initialize the soil moisture in the model. It is unclear whether, and how, the soil moisture initial condition used in the experiments affects the model results. These two issues are investigated in this study.

As noted by Fennessy and Shukla (1999), there are no global measurements of soil moisture for use in model sensitivity studies, and considerable uncertainty in defining and measuring soil moisture. In this study, we superimpose the soil moisture anomalies (percentile) from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data (Kalnay et al. 1996) with the soil moisture climatology of Mintz and Serafini (1989) to generate different soil moisture initial conditions for the model sensitivity study. As soil moisture data from the reanalysis are products from the soil hydrological model used in the reanalysis system, they could well be different from that in the BMRC model and from reality.

Before utilizing the NCEP–NCAR reanalysis, it is worthwhile comparing the soil moisture results from the reanalysis data and the BMRC simulations, over part of the Chinese region, with the available, but limited, observational soil moisture datasets reported by Robock et al. (2000). The original dataset consists of 43 stations in China, for the period 1981–91. In this study, we choose four stations (numbers 32, 33, 34, 36) located in the central China region (Fig. 1a) for the comparison. These four stations show similar variations in the top 1-m plant available soil moisture for the period 1981–88 (Fig. 1b). Station 35 has been excluded from the study because it shows quite different characteristics when compared with the other four. The averaged soil moisture values from the four stations are then compared with the reanalysis and model simulations (Fig. 2).

The inadequacy of directly applying observational soil moisture data in numerical models is clearly demonstrated in Fig. 2a. There are large differences between observed soil moisture variations and the ones derived from the NCEP–NCAR reanalysis and the model (with results taken from a 10-yr simulation). The observed plant available soil moisture, with a range of 10–25 mm, is much smaller than the model and the NCEP–NCAR reanalysis. Such large differences are partially due to the different meaning of soil moisture in the observational data and each numerical model. In land surface models, soil moisture is more of an index used in the calculations of surface evapotranspiration and runoff rather than representative of the actual mass of moisture in the soil as in observations (Robock et al. 1998). There are also different seasonal variations of soil moisture, over this region, between the BMRC model and the reanalysis data. Figure 2b also shows the climatological seasonal cycle of soil moisture variations over this region. The soil moisture climatology from Mintz and Serafini (1989), which was used in the model initialization in the BMRC AGCM seasonal forecasting trial experiments (Frederiksen et al. 2001), is also included for comparison. Because of the large differences seen in Fig. 2a, and in order to make proper comparison, the climatological soil moisture seasonal cycles in these datasets are divided by their annual means. Thus, Fig. 2b shows the relative seasonal variations (percentile). In the observations, soil moisture starts to deplete from boreal spring and into the summer months. Then it begins to accumulate and remains stable during the winter season. In contrast, the reanalysis data show soil moisture peaking during the summer season and with a minimum in the wintertime. The soil moisture seasonal cycle, simulated in the BMRC model, shows a reasonable similarity with the observations in the first half of the year, but results during the autumn season are not satisfactory. The Mintz and Serafini (1989) soil moisture climatology has a much too weak seasonal cycle over this region, with soil being saturated during most of the year except for small decreases in the boreal summer season.

The relative soil moisture anomalies (expressed as a percentage and with respect to their corresponding climatologies over this region) are shown in Fig. 2c. The reanalysis data show similar relative variations as observed, but the BMRC model does not capture the features seen in the observations. The correlation of relative soil moisture anomalies between the observations and the reanalysis data is 0.45, while it is only about 0.11 between observations and the model simulations. Thus, even though soil moisture in the reanalysis data is quite different from the observations (as seen in Figs. 2a and 2b), its percentile anomalies could provide some useful information about the soil moisture anomalous conditions. This is the primary reason that we utilize the percentile soil moisture anomalies from the NCEP–NCAR reanalysis in this sensitivity study.

We do this by first combining two soil moisture fields from the reanalysis data. One is the volumetric soil moisture between 0 and 10 cm and the other is the volumetric soil moisture between 10 and 200 cm. These are combined using the representative soil depth as weight. Then, the soil moisture climatology for the period 1948–99 is calculated and the soil moisture anomalies of May 1998 are calculated. These soil moisture anomalies are then transformed into percentile values by dividing them by the corresponding soil moisture climatology from the reanalysis. Finally, we apply these percentile anomalies to the soil moisture climatology of Mintz and Serafini (1989) used in the BMRC model initialization, with the soil moisture range of 0–150 mm retained. Figure 3 shows the actual changes to the soil moisture initial condition after imposing the percentile anomalies from May 1998 (from the reanalysis data) in the model initialization. Note that there are only small changes in the soil moisture initial conditions over the east Asian region. Large changes are located over central Eurasia and the midlatitudes of North America (positive), South America, and southern Africa (negative). Small (positive) soil moisture anomalies, associated with the early spring rainfall anomalies of 1998, are imposed over the south China region. In addition, slightly dry conditions are imposed over the central and northern China region.

In addition to the soil moisture sensitivity experiment, we have also conducted an experiment where the model is forced with observed global SSTs, for the period of interest. This allows us to compare the relative roles of SST and soil moisture conditions in determining the mean state of the atmosphere on a seasonal timescale. Table 1 summarizes the experiments analyzed in this study. The control runs, where the model is forced with climatological SST and with soil moisture initialized with the climatology of Mintz and Serafini (1989), is hereafter named CTL. The original trial forecast runs as reported by Frederiksen et al. (2001) with the model forced with persistent SST anomalies is called ORG. The model integrations with soil moisture initializations using the percentile anomalies of May 1998 derived from the NCEP–NCAR reanalysis data is named SI, and the runs with the model forced with observed SSTs, in the course of the model integration, is called OS.

Each of the four sets of experiments comprised six ensemble runs of 120-day integrations starting with the initial conditions from 0500, 1100, 1700, 2300 UTC of 10 May 1998 and 0500, 1100 UTC of 11 May 1998. The dynamical fields in the initial conditions are taken from the Australian Bureau of Meteorology Global Assimilation and Prediction System (Seaman et al. 1995). All the experiments have the same SST and sea ice boundary conditions that are generated by superimposing the observed weekly global SST anomalies from 4 to 10 May 1998 into the SST climatology of Reynolds and Smith (1995) (Frederiksen et al. 2001).

It must be emphasized here that we have used a very simple approach for imposing reanalysis-based soil moisture anomalies into the BMRC model to conduct our sensitivity study. Our purpose is not to study how to initialize soil moisture in AGCM seasonal forecasts. Rather, it is to study whether different soil moisture initial conditions affect the skill of our seasonal forecasts, and to compare with results published recently from other AGCM experiments (e.g., Fennessy and Shukla 1999; Douville and Chauvin 2000). Our main focus is to explore the physical and dynamical processes involved so that we may better understand the model results. Recently, a number of international initiatives have considered the importance of soil moisture initialization, such as, for example, the Global Soil Wetness Project (Dirmeyer 2000) and Global Land Data Assimilation System projects (Houser 2001). Future experiments, using advanced soil moisture initialization approaches (e.g., Walker and Houser 2001), will answer the question of whether realistic soil moisture condition can improve the model seasonal integrations.

Besides the importance of proper land surface initialization, how well the land surface schemes represent the surface hydrological processes and the biosphere–atmosphere interactions can have equally significant impacts on the longer timescale integrations and predictability in atmospheric models. In the bucket-type soil hydrological model, as used in the version of BMRC AGCM, volumetric soil water is depleted directly through evaporation and replenished by precipitation up to maximum water storage capacity of 150 mm universally. This simplification of surface hydrological processes clearly affects the surface water and energy partitions (e.g., Shao and Henderson-Sellers 1996; Koster and Milly 1997). The memory from anomalous land surface conditions might be shorter in the bucket-type model than in the models with multilevel soil hydraulic process (Zhang 2002; Zhang et al. 2002a). On the other hand, Scott et al. (1997) found that in regions with dense canopy coverage, bucket models showed relative slow response of evapotranspiration to precipitation forcing as there is a lack of canopy interception component in such models to reflect the rapid canopy transpiration of intercepted water. Therefore, further understanding of the interactions between the land surface and the atmosphere is needed in the area of exploring the impacts of land surface processes on model predictions and predictability.

3. Impacts of soil moisture initial conditions on the model results

In this section, we compare surface climate anomalies simulated by the series of experiments as described in section 2. The model forecasts are defined as the difference between the forecasting experiments (ORG, SI, and OS) and CTL as defined in Table 1. The seasonal mean climate anomalies between the ORG and CTL experiments represent the AGCM model's response to the SST anomalies. The corresponding anomalies between ORG and SI reflect the influences of the soil moisture initial condition on the model integration. In addition, the differences between the OS and ORG experiments indicate whether forcing the model with realistic SSTs during the course of model integration will improve the model forecasts.

The observed climate anomalies, including rainfall anomalies derived from an updated version of observational data described by Xie and Arkin (1996) and surface temperature anomalies derived from an updated version of the observational dataset of Parker et al. (1994), are used as references for the purpose of comparison. Note that the primary aim of this study is to assess the model sensitivity to soil moisture conditions, and the relative importance of this compared with a better forecast of SST forcing. We therefore emphasize the differences between the experiments of SI − ORG and OS − ORG, rather than compare each set of forecasts with observed anomalies, even though it is the eventual goal.

Figure 4 contrasts three sets of model simulations of rainfall anomalies in JJA98 over the east Asian region. Figure 4a is the JJA forecast of the precipitation anomalies from the original experiments (Zhang and Frederiksen 2001; Frederiksen et al. 2001). Compared with the observed anomalies, the model forecast from the ORG experiment reasonably captured the rainfall anomalies over the northeast China region, where severe flooding occurred (Fig. 4d). However, it failed to simulate the devastating floods over the Yangtze River region in south China (roughly 20°–35°N, 95°–120°E). Here, the model forecasted dry anomalies.

Initializing the model with May 1998 soil moisture anomalies (in percentile) produces a number of distinct impacts, with some of them offering better agreement with observations. Figure 4b shows the differences between the simulation from the SI experiment (Fsi) and that of the original one (Forg). Contrasting the SI and ORG experiments, positive rainfall anomalies are simulated over the south China region, alleviating the incorrect dry forecasts seen in the original forecast (ORG) using a climatological soil moisture initial condition. Another region of improvement is over the lowland area south of the Himalayas, part of the northern India and Bangladesh regions, where above-normal rainfall was observed in JJA98 and the simulations from the SI experiment predict more rainfall than that in the ORG experiments. In the northeast China region, the SI experiments show an increase of rainfall over the observed flooding area. Overall the forecasts based on the ensemble SI runs offer better agreement with the observations than the original forecasts.

Both the ORG and SI experiments use persistent first-week SST anomalies from May 1998 in the course of the model integration. The question then arises whether the forecasts from ORG can be improved by forcing the model with observed weekly SST anomalies in the course of the model integration, and whether such improvements are greater or smaller than those we have just seen in the SI experiment. Figure 4c demonstrates the differences between the model simulations by forcing the model with observed global SST anomalies (OS experiment) and with persistent SST anomalies (ORG experiment). Note that the OS experiments show improvements over the Yangtze River flooding area with increased rainfall over the region. It also improves the original rainfall forecasts over the northeast China region. Such features are similar to the effects seen from the SI experiment and the magnitudes of the improvements are similar. Another area of large differences between the OS and ORG experiments in rainfall prediction is over the Indochina peninsula where the OS generates more rainfall than the ORG experiment.

In this study, we have only conducted six ensemble runs for each set of experiments. This makes it difficult to perform meaningful statistical tests on whether the mean difference between each set of ensemble runs, such as those seen in Fig. 4, is statistically significant. It should be noted again that our focus is on trying to identify physical and dynamical processes in the BMRC AGCM that relate to its sensitivity to different soil moisture conditions. This is more important than searching for appropriate statistical approaches to attach statistical significant levels to such differences (Nicholls 2001). Table 2 briefly summarizes and compares JJA three-month precipitation totals at the location of 25°N, 117.5°E simulated in the four sets of six-ensemble runs presented in Fig. 4. This is the place where the SI experiments show increased rainfall over the south China flooding region (Fig. 4b). Of the six ensemble runs in the SI experiment, five runs show increased rainfall compared with the average from the six ORG runs (Aorg). The mean difference between SI and ORG experiment is +178 mm. This difference is almost twice that between the ensemble averages of the ORG and CTL experiments (about 98 mm) that is due to anomalous SST forcing. Table 2 also suggests that the impacts due to the different soil moisture initial conditions are, at least, comparable with the differences that result when the model is forced with observed SST forcing. Such a conclusion is also reported in Entekhabi et al. (1999) and Douville (2002) in discussing the potential of soil moisture condition in affecting seasonal predictions. At this location, the mean difference due to different soil moisture initialization is even bigger than that by forcing the model with observed SST condition throughout the model integration. Overall, results from Table 2 suggest that the differences we see in Fig. 4b are largely due to the model's sensitivity to different soil moisture initial conditions, despite the fact that the model's internal variability is also high.

The impacts of the soil moisture conditions, on the surface temperature forecasts, are expected to be more discernible because of its direct impacts on the land surface energy partitions. Figure 5 illustrates the comparison of surface air temperature (at 1.5-m height) anomalies from three sets of forecasts in Table 1, as well as the observational anomalies derived from an updated version of Parker et al. (1994). Forecasts from the ORG experiment (Fig. 5a) show warming anomalies over the southern (south of 35°N) and northern (north of 50°N) parts of the east Asian continent, and cooling anomalies over the central region in between. In contrast, observations in Fig. 5d show warming anomalies over a large part of the east Asian continent, with only very small cooling anomalies in the northeast China region (Fig. 5d).

Initializing the model with the May 1998 soil moisture anomalies (SI experiments) generates higher surface air temperature over central and eastern Asia (Fig. 5b), compared with the original forecasts. Combining results from Figs. 5a and 5b suggests that surface air temperature forecasts are improved in SI over this region, offering better agreement with the observations (Fig. 5d). For instance, the large cooling anomalies seen in Forg (Fig. 5a) over the central east Asian continent, which is the opposite of the observed anomalies, are largely corrected in the SI experiment. In the western Siberian region, the SI experiment has cooler surface temperature than in the ORG experiment, thus model simulations over this region are affected (but not improved), when compared with observations. Figure 5c shows the corresponding impacts of forcing the model with observed SST anomalies. In this case, cooler surface temperatures are simulated over the Tibetan high region and eastern Siberian region than were seen in the ORG experiment. Despite the fact that surface temperature forecasts are modulated in the OS experiment, forcing the model with observed SST conditions does not significantly improve the forecasts.

It should be pointed out that in this study we only concentrate on the diagnosis of the model results using the Kuo convection scheme. In fact, the model had the option of using a version of the Tiedtke mass-flux convection scheme (Colman and McAvaney 1995) and Frederiksen et al. (2001) did report and compare the model trial seasonal forecasts in the 1997/98 ENSO event using both convection schemes. Despite the deficiencies known in the Kuo-type convection scheme (Emanuel 1994), the overall skill scores for the forecasts of the period of March 1997–March 1998 showed essentially the same features except for some differences in the extratropics. In this study, a complementary experiment using the Tiedtke mass-flux convection scheme was also conducted (not presented) to assess the relative importance of improved soil moisture initial condition, improved SST forcing, and improved model physics in developing skilful forecasts. It is noted that model forecasts can be significantly influenced by the physical parameterizations employed, highlighting the importance of improving the model physics in developing skillful forecasts, along with efforts in advancing the model land surface initialization and SST forecasting.

In addition, the sensitivity seen in this study can be affected by the land surface representations in the model itself. In the version of the BMRC AGCM, it has a very simple bucket soil hydrological model and lacks canopy representation in land surface energy and water balances. In a separate study, Zhang (2002) reported results from a version of the BMRC atmospheric model run with a bucket-type land surface scheme in the Atmospheric Model Intercomparison Project 2 (AMIP-2) context. Preliminary lag-correlation analysis was conducted and results showed that this model, with a simple bucket-type soil hydrology, performed in a very similar way as seen from other models with a bucket-type soil hydrology scheme (Zhang et al. 2002a): it had a more rapid decay rate in the retention of soil moisture anomalies, which may lead to soil moisture conditions having a weaker influence on forecasting surface climate anomalies. Thus, the initialization of soil moisture in GCMs and the parameterization of soil hydrological processes are two important issues in improving GCM seasonal integrations. Continual efforts are undertaken to improve the parameterization of soil hydrological processes in future BMRC model development (e.g., Zhang et al. 2002b).

4. Physical and dynamical processes

In section 3, we analyzed the impacts of soil moisture initialization on the model seasonal forecasts of precipitation and surface temperature anomalies. In this section, we concentrate on the analysis of the model physics and dynamics that are likely responsible for such influences. As discussed before, there are basically two ways in which different soil moisture conditions may affect the model forecasts. One is through its impacts on local water recycling, with more soil water content leading to more surface evaporation. The enhanced local moisture source favors the production of more precipitation. The other is through its impact on the surface energy partition with more soil water content leading to less surface sensible heat and cooler surface that could then alter the low-level boundary condition of the atmosphere and affect the model dynamical processes in simulating precipitation in the model.

The first question to be asked here is how long the soil moisture anomalies imposed in the model initial condition can be retained. Figure 6 compares soil moisture variations in the ORG and SI runs at the location of 52.5°N, 75°E where moderate positive soil moisture anomalies (about 10 mm) are seen in the central Eurasian region (Fig. 3). It shows the soil moisture variations in each of the six ensemble runs and their ensemble averages, starting from 2300 UTC 11 May 1998. Note that as the six ensembles runs start from 0600, 1100, 1700, and 2300 UTC 10 May 1998 and 0600 and 1100 UTC 11 May 1998, there are already differences between ensemble runs at 2300 UTC 11 May 1998. A large spread is seen between each of the six ensembles in the SI and ORG experiments due to the chaotic weather forcing components in simulating the model soil moisture variations. However, Fig. 6 does show, in the ensemble means, a clear separation between the runs from SI and those from ORG, despite the fact that one of the ORG runs behaves remarkably differently from the rest. Even though the spread between each ensemble run increases in the course of model integration, the averaged mean difference of soil moisture from the SI and ORG runs has a similar magnitude throughout the first month of the model integrations. It should also be pointed out that the imposed soil moisture anomalies in the east Asian flood region (see Figs. 3 and 4d) are small and soil moisture over this region is near its saturated value (150 mm) in the model. Consequently, the soil moisture difference between SI and ORG runs (not shown) is not as obvious as that seen in Fig. 6. As discussed later, we attribute the model sensitivity shown in Figs. 4 and 5 to the nonlocal impacts of soil moisture anomalies imposed and sustained elsewhere.

Next, we consider the impact of soil moisture conditions on the model surface energy partition. Figure 7a shows the differences in surface latent heat flux between the ORG and SI experiments. Averaged over the three months, the SI experiment produces more latent heat flux over the northern part of east Asia and the Siberian region, and over the land near the Bay of Bengal. The largest increases (10–20 W m−2) are located over northern China and western Siberia. A slight increase (10 W m−2) is seen over the south China coastal region where there is an increase of rainfall in Fig. 4b. In addition, surface latent heat flux decreases in most of the regions between 30° and 40°N of 10–20 W m−2, except for a weak increase in the central area. Results for surface sensible heat flux (Fig. 7b) show nearly opposite signs to those shown for the latent heat. Over the central region (roughly 30°–45°N belt) there are increases in surface sensible heat flux of 10–20 W m−2, offsetting the reduction in latent heat over the region to achieve surface energy balance. Similarly, over the northern part of the Eurasian and Siberian region, the large increases in surface latent heat fluxes seen in Fig. 7a are linked to reductions in surface sensible heat.

Comparing Figs. 4 and 7, the centers of the anomalies in surface fluxes (Fig. 7) over Siberia and northeast China roughly coincide with regions where of increased precipitation in the SI experiment (see Fig. 4b). In contrast, only weak changes of surface latent and sensible heat fluxes are seen in the south China flood region, where the model precipitation shows significant differences between the SI and ORG experiments. Compared with other regions, a different relationship between changes in surface flux anomalies and precipitation anomalies is seen over the region roughly covering the east Asian monsoon region (∼20°–40°N, 90°–120°E). For instance, Figs. 4b and 7 show that over large parts of the mid- and high latitudes of the east Asian continent (e.g., northeast China, Mongolia, and western Siberia), the model produces an increase in precipitation, an increase in surface latent heat, and a reduction in surface sensible heat flux. Over the northern and western China region, the decrease in precipitation corresponds to a reduction in latent heat. However, there are only weak and negative correlations over a large part of the east Asian monsoon region, where the increase in precipitation between SI and ORG corresponds to a reduction in surface evaporation.

To further understand the model results in terms of the interactions between surface fluxes and the model atmospheric physics, we have conducted a detailed analysis of the model results in the lower reaches of the Yangtze River where the SI experiment shows improvements in forecasting the rainfall anomalies in JJA98. Figure 8 shows the relationships between the model simulated 6-hourly precipitation and surface latent heat flux (25°–32.5°N, 107°–120°E) from six ensemble runs in the ORG experiments for the forecasts of JJA97. Unfortunately, due to a change in the BMRC supercomputing facility, we were unable to further process the model data for the JJA98 case. However, we believe that the model physics operates similarly during both periods and the results from JJA97 serve to illustrate our point. Over this region, the model shows no significant correlations between precipitation and surface latent heat flux. The separation seen in Fig. 8a is due to a strong diurnal cycle in surface latent heat flux. Only a weak positive correlation is seen at night when the surface latent heat flux is very low. In addition, the possibility that more latent heat flux leads to a cooling of the surface, more stable low-level air, and, consequently, a reduction in atmospheric convection and precipitation is rejected. This is because there is also a weak negative (rather than positive) correlation between surface sensible heat and precipitation (not shown). The negative correlations suggest that the changes in the surface radiative forcing, associated with the presence of cloudiness during rain events, control the changes in surface fluxes. A series of lag-correlation calculations (not shown) failed to show any significant lag correlations to demonstrate that precipitation, over this region, is sensitive to local surface flux conditions, at least on the timescale of 6-hourly and above. Therefore, in this region, local water vapor supply from the hydrological water recycling process may not be the main factor affecting the model rainfall forecasts. It is noted that, using a single column model, Sud et al. (2001) studied the evaporation–precipitation feedback under different conditions and they showed strong dependence of such relationship on background circulations, with only very weak correlations seen in dry circulation cases. Similar study will be pursued in further understanding the interactions between land surface flux and precipitation in future BMRC model studies (Zhang et al. 2002b).

Rather, as shown in Fig. 8b, there is a strong correlation between precipitation and local atmospheric dynamic conditions. Precipitation in the model is more dominated by the dynamics, such as low-level vorticity and convergence (not shown), than by surface fluxes in this model. In the Kuo convection scheme employed in this version of the BMRC model, the occurrence of penetrative cumulus convection requires (i) large-scale moisture convergence and (ii) a humid environmental air mass. Indeed, the model results show strong correlations between precipitation and low-level atmospheric humidity.

Theoretically, the extent to which water recycling, through surface evaporation, affects the model precipitation depends upon how much water from surface evaporation is used to increase the overlying atmospheric humidity. Two processes are involved in determining precipitable water in the model: one is the proportion of surface water recycling that becomes local precipitable water; the other is how much water is transported by the atmospheric circulation into the region. A number of studies (e.g., Oglesby and Erickson 1989; Entekhabi et al. 1992; Fan and Oglesby 1996) have shown that the effectiveness of surface water recycling in producing local precipitable water depends on the features of the regional atmospheric circulation.

To try to understand the two processes above, Fig. 9a shows the correlations between the 850-hPa humidity and surface evaporation from the six runs of the ORG experiments in JJA98. Over the geographic domain of this study, there are two distinct features. Over a large part of the east Eurasian continent, low-level atmospheric humidity is positively correlated to surface evaporation. The most significant correlations are over the inland region. In contrast, negative correlations are seen over the east Asian monsoon region, where more surface evaporation does not lead to more humid low-level atmosphere. Such a feature is broadly similar to the results seen in section 3, where changes of precipitation have higher correlations with changes of surface evaporation over most of the Eurasian continent except the east Asian monsoon region where such a correlation is not seen.

The correlation between 850-hPa atmospheric humidity and 850-hPa moisture convergence ∇·(qV), shown in Fig. 9b, demonstrates that over most of the east Asian monsoon region, low-level atmospheric humidity is strongly correlated to horizontal moisture transport. In other words, over the region controlled by the strong east Asian monsoon dynamics, it is the water vapor transport by the regional atmospheric circulation (i.e., the monsoon flow) that dominates the humidity condition of the overlying atmosphere, and surface water recycling plays a less important role. This is a prominent feature over southeast China and the middle and the lower reaches of the Yangtze River (Fig. 9b). Note that our results show different features over the east Asian monsoon and the Indian monsoon regions, underlining the complexity of the monsoon systems in this region (see, e.g., Webster et al. 1998).

The dynamical processes identified from Figs. 9a and 9b can be used to explain the connections between results seen in Figs. 4 and 7. In Fig. 4b, rainfall is increased in the southern China and Yangtze River region in the SI experiment. This region also shows a reduction in surface evaporation (see Fig. 7a). These results can be explained by the changes in water vapor transport seen in the three sets of experiments (Fig. 10).

Figure 10 compares horizontal water vapor transport at 10-m height in the model for the four sets of experiments (CTL, ORG, SI, and OS). Results serve as the representative of horizontal water vapor advection in the boundary layer as we are unable to calculate the vertically integrated water vapor transport from limited storage of the model data. With respect to the CTL runs, the ORG experiments simulate cyclonic circulation anomalies over the northwest Pacific Ocean. A large increase in moisture convergence is seen over the western and northwestern Pacific Ocean, away from the convergence belt over the south China flood region seen in the NCEP–NCAR reanalysis in Fig. 10d. Consequently, the model simulated above-normal rainfall anomalies over the western and northwestern Pacific Ocean. In the SI experiment, changes in the regional atmospheric circulation results in the enhanced moisture convergence over the south China region when compared with the ORG experiment (Fig. 10b). In Fig. 10b, there is an increase in moisture transport from the South China Sea warm water toward the continent. Even though surface evaporation is decreased over the region (when comparing the SI and ORG experiments), the increase of horizontal water vapor transport over this region and the convergent flow, results in an increase of JJA98 precipitation in the SI experiment over that in the ORG experiment. In the OS experiment, forcing the model with observed SST anomalies in the course of the model integration has also altered the simulation of regional circulation responses as shown in Fig. 10c. The increase in moisture transport from the western Pacific Ocean toward the east Asian continent and the increase of moisture convergence over the low reach of the Yangtze River are consistent with the rainfall anomalies shown in Fig. 4d.

The increase of airflow from warm ocean toward the east Asian continent in the SI experiment can be attributed to the drier and warmer land conditions simulated in the SI experiment. Results from Fig. 7 show the large reduction in surface latent heat flux and the increase in surface temperature over the east Asian continent in the SI experiment. Therefore, the land–sea temperature contrast increases and the airflow from the ocean toward the continent is enhanced. Figure 11 shows the difference in the number of troughs simulated over each model grid in the ORG and SI experiments from the model 6-h outputs. This is estimated by comparing the gridpoint sea level pressure with the nearby grids in the longitudinal direction. There are generally increases in the trough activities over the Yangtze River and south China region. Such increases correspond well with the enhanced precipitation in the SI experiment.

So far, we have discussed the model responses over the east Asian region to the changes in soil moisture initial conditions. Considering the close connections between the Asian monsoon and the Australian monsoon, we are also interested in whether the model simulations over the Australian region are affected by different soil moisture configurations in the model. A number of studies (e.g., Simmonds and Lynch 1991; Power et al. 1998; Viviand et al. 2000) have studied the impacts of soil moisture on the Australian climate, but there is a lack of studies with experiments particularly designed to assess its impacts on the AGCM seasonal forecasting in this region. Figure 12 shows the comparison of the model simulations of Australian winter rainfall anomalies in JJA98 from the four sets of experiments conducted in this study. Contrasting the model results with observed rainfall anomalies, it is noted that the anomaly patterns are, in general, improved using the soil moisture initialization. In the original forecasts, the Australian continent was dominated by dry anomalies except for the above-normal rainfall in the northeastern coastal region and the western corner of the continent. In contrast, anomalies derived from the Xie and Arkin (1996) dataset show overall above-normal rainfall anomalies over a large part of the continent, excluding the dry anomalies in some northern regions. The differences in rainfall simulations between the SI and ORG experiments (Fig. 12b) suggest that the results in the SI experiment offer better agreement with the observations by alleviating the dry anomalies seen in the ORG experiment in most regions. The magnitude, as well as the pattern of the differences of rainfall simulations from the OS experiment, is also similar to those seen in the SI experiment, except for the dry area in the western corner shown in Fig. 10c. Figure 13 shows changes in the regional atmospheric circulation between the four sets of experiments. The increases in precipitation in the SI and OS experiments over a large part of Australian continent are associated with the enhancement of water vapor convergence and the inflow of water vapor advection from the southwest Pacific Ocean and southern oceans.

It should be pointed out here that to assess how physical and dynamical processes identified here depend on the model configuration, in this study, we also analyzed the model results using a mass-flux convection scheme. We found (not shown) a similar relationship among precipitation, surface evaporation, and horizontal water vapor transports. The dominant role of water vapor transport in maintaining the atmospheric convection and rainfall process over the east Asian monsoon region was seen in both sets of model integrations using Kuo and mass-flux convection schemes. More detailed diagnosis of the interaction between land surface processes and the model atmospheric dynamics and physics (e.g., Sud et al. 2001) using the BMRC models will be pursued (Zhang et al. 2002b).

5. Discussion and conclusions

In this study, we have reported results from a series of experiments using the BMRC AGCM, aimed at assessing the model's sensitivity to different soil moisture initial conditions in its dynamically extended seasonal forecasts. This paper has focused on the June–July–August 1998 (JJA98) seasonal forecasting results from six ensemble runs with 120-day model integrations, with a detailed analysis of the model performance over south and northeast China. In the analysis, we have not only assessed the extent to which different soil moisture conditions can modify (not necessarily meaning improve) regional features in the model forecasts, but have also tried to identify plausible physical and dynamical processes that appear responsible for the changes in the model forecasts.

We initialized the model with soil moisture anomalies (percentile) derived from the NCEP–NCAR reanalysis data of Kalnay et al. (1996), and compared our results with those from the case where the soil moisture climatology of Mintz and Serafini (1989) has been used. As part of the study, we have assessed the soil moisture variations from the NCEP–NCAR analysis and 10-yr BMRC model runs against observational data (Robock et al. 2000) over an area in the central China region. The incompatibility of soil moisture from observations, reanalysis, and the model results has been clearly demonstrated. However, our analysis indicates that relative soil moisture anomalies, expressed as percentiles, from the NCEP–NCAR reanalysis data may contain useful information that can be utilized in conducting model sensitivity experiments.

A comparison of the model's simulations of JJA98 rainfall anomalies in the east Asian region has shown that regional features can be modulated by land surface conditions. Initializing the model with the percentile soil moisture anomalies of May 1998 (SI experiment) improves the model forecasts over the severe flooding region along the Yangtze River in the south China region. In addition, forecasts in the northeast China region offer better agreement with the observed anomalies. Using realistic SST forcing (OS experiments) also improved the model simulation over the south China region but its contribution to the simulation of rainfall anomalies in northeast China is weaker than seen from the SI runs. Initializing the model with soil moisture anomalies from the reanalysis also improved the model's forecasts of temperature anomalies over a large part of the central Eurasian continent, when compared with the observed anomalies. Results over western Siberia were also affected. The relative model response, in terms of surface temperature simulations, to the use of observed SST forcing in the course of the integration, is weaker in the OS experiment than seen in the SI runs.

In this paper, we found that the relationship between the precipitation anomalies and changes in surface fluxes depends on the dynamical features of the regional climate. In the region controlled by the strong Asian monsoon, the contribution of surface water recycling, through surface evaporation, to the precipitation process is limited. In the Yangtze River flooding region, the increase in precipitation in the SI experiment corresponds to a decrease in surface evaporation. In the inland east Asian continent and in the northern region, however, there is a close link between an increase (decrease) in precipitation and an increase (decrease) in surface evaporation. An analysis of 6-hourly model data has demonstrated that, in the Asian monsoon region, precipitation is primarily dominated by the dynamical conditions such as low-level atmospheric convergence and cyclonic vorticity. The main source of moisture in the model precipitation is from the horizontal water vapor transport through the monsoon flow, rather than surface water recycling. There are even negative correlations between surface evaporation and low-level atmospheric humidity in the east Asian monsoon region.

Our results also suggest that changes in the land–sea thermal contrast over the region, associated with the change in surface temperature, affect the regional atmospheric circulation. In the SI experiment, the monsoon airflow from warm ocean toward the continent is enhanced by the warming of the land. The enhanced airflow brings moister air mass from the oceanic region to the land area, contributing to the increase in the precipitation forecasts in the flooding region. Results from this study have clearly demonstrated that when assessing the impacts of land surface conditions in the model seasonal forecasts, both the local processes (direct interactions between land surface and the overlying atmosphere) and the nonlocal processes (indirect interactions through the impacts on the regional circulations) must be taken into account. Such processes have also been reported in Fan and Oglesby (1996) in a study of east Asian monsoon simulations. To further help us to explore the nonlocal impacts of soil moisture conditions on the model seasonal forecasts, we have also conducted a complementary experiment, not reported in this study. By saturating soil moisture over the east Asian continent (20°–70°N, 80°–125°E) in the model initial condition, the model shows nonlocal effects of the land surface conditions on the model seasonal forecasts over the Australian region. Thus, when anomalous land surface boundary conditions are observed in the Eurasian continent (e.g., excessive or less snow coverage and accumulation), one could expect some teleconnecting responses over the Australian region due to the interactions between the Asian monsoon and the Australian monsoon.

While the focus of this paper has been on assessing the model performance over the Chinese flooding regions in JJA98, we have also analyzed the model results over the Australian region. The model forecasts of precipitation and surface temperature (not shown) over the Australian region are affected, in a similar way, by the different soil moisture configurations. The improvements in the model forecasts of rainfall anomalies in JJA98 are consistent with the changes in regional atmospheric circulation in response to the different soil moisture conditions.

Our aim here has been to assess the BMRC AGCM's sensitivity to different soil moisture configurations when conducting seasonal forecasts. This study has successfully shown the extent to which, and how, the land surface conditions could affect the regional features of the model seasonal integrations. However, in this study we have used a very simple approach that imposes relative soil moisture anomalies from the NCEP–NCAR reanalysis in the model initialization of soil moisture. Recently, a number of studies have focused on the assimilation of soil moisture in GCMs (e.g., Bouttier et al. 1993; Giard and Bazile 2000; Walker and Houser 2001; Mahfouf and Viterbo 2001) and on the technical issues of transferring soil moisture from one model product to another (e.g., Fennessy et al. 1994). The current study is only the very first step (i.e., assessing the model sensitivity) in our plans for better soil moisture initialization and simulation in our model. Efforts are currently being made to improve the soil moisture simulation and land surface data assimilation in our model (e.g., Zhang et al. 2002b), along with efforts in improving other physical and dynamical process in the model. This will answer the question of whether a more realistic soil moisture condition improves the model forecasts.

This study has shown the impacts of different soil moisture initializations on the BMRC AGCM integrations. The sensitivity seen in this study could depend on the land surface representations in the model itself (Koster and Suarez 2001). The version of the BMRC AGCM used in this study has a very simple bucket soil hydrological model and lacks canopy interception and transpiration processes. In the bucket-type model, soil absorbs all rainfall with no occurrence of runoff till it reaches a prefixed threshold. This simplification of the runoff process clearly affects the surface water (and energy) partition(s), with the importance of surface runoff and drainage in determining surface water and energy balances being recognized in many studies (e.g., Shao and Henderson-Sellers 1996; Koster and Milly 1997). Also, bucket-type models present rapid processing of volumetric soil moisture in response to evaporation as the distribution of soil moisture within soil layer is ignored (Shao and Henderson-Sellers 1996). Thus, in the climate system, the memory from anomalous land surface conditions might be shorter in the bucket-type model than in the models with multilevel soil hydraulic process. A recent analysis of 16 AMIP-2 model results (Zhang et al. 2002a) has revealed that longer memory from the anomalous land surface conditions could be related to the complexities in the land surface representations, particularly the soil hydrological processes. Model forecasts also equally depend on how well some slow-varying land surface processes (e.g., soil moisture retention) and the interactions between land surface conditions and the overlying atmosphere are parameterized.

Acknowledgments

The authors appreciate the useful discussions with Dr. B. McAvaney. Comments from Drs. M. Harvey and X. Wang (NCC) are acknowledged. This study was partly funded by a grant from the Australian Land and Water Research and Development Corporation (LWRRDC). Comments and suggestions from two anonymous reviewers during the first submission of this study are appreciated.

REFERENCES

  • Bonan, G. B. , and L. M. Stillwell-Soller , 1998: Soil water and the persistence of floods and droughts in the Mississippi River Basin. Water Resour. Res., 34 , 26932701.

    • Search Google Scholar
    • Export Citation
  • Bouttier, F. , J-F. Mahfouf , and J. Noilhan , 1993: Sequential assimilation of soil moisture from atmospheric low-level parameters. Part I: Sensitivity and calibration studies. J. Appl. Meteor., 32 , 13351350.

    • Search Google Scholar
    • Export Citation
  • Brankovic, C. , and T. N. Palmer , 2000: Seasonal skill and predictability of ECMWF PROVOST ensembles. Quart. J. Roy. Meteor. Soc., 126 , 20352067.

    • Search Google Scholar
    • Export Citation
  • Charney, J. G. , and J. Shukla , 1981: Predictability of monsoons. Monsoon Dynamics, J. Lighthill and R. P. Pearce, Eds., Cambridge University Press, 99–109.

    • Search Google Scholar
    • Export Citation
  • Colman, R. A. , and B. J. McAvaney , 1995: Sensitivity of the climate response of an atmospheric general circulation model to changes in convective parameterization and horizontal resolution. J. Geophys. Res., 100 , 31553172.

    • Search Google Scholar
    • Export Citation
  • Dirmeyer, P. A. , 2000: Using a global soil wetness dataset to improve seasonal climate simulation. J. Climate, 13 , 29002922.

  • Douville, H. , 2002: Influence of soil moisture on the Asian and African monsoons. Part II: Interannual variability. J. Climate, 15 , 701720.

    • Search Google Scholar
    • Export Citation
  • Douville, H. , and F. Chauvin , 2000: Relevance of soil moisture for seasonal climate predictions: A preliminary study. Climate Dyn., 16 , 719736.

    • Search Google Scholar
    • Export Citation
  • Emanuel, K. A. , 1994: Atmospheric Convection. Oxford University Press, 580 pp.

  • Entekhabi, D. , I. Rodriguez-Iturbe , and R. L. Bras , 1992: Variability in large-scale water balance with land surface–atmosphere interaction. J. Climate, 5 , 798813.

    • Search Google Scholar
    • Export Citation
  • Entekhabi, D. , and Coauthors. 1999: An agenda for land surface hydrology research and a call for the second international hydrology decade. Bull. Amer. Meteor. Soc., 80 , 20432058.

    • Search Google Scholar
    • Export Citation
  • Fan, Z. , and R. J. Oglesby , 1996: A 100-yr CCM1 simulation of North China's hydrologic cycle. J. Climate, 9 , 189204.

  • Fennessy, M. J. , and J. Shukla , 1999: Impact of initial soil wetness on seasonal atmospheric prediction. J. Climate, 12 , 31673180.

    • Search Google Scholar
    • Export Citation
  • Fennessy, M. J. , J. L. Kinter III , L. Marx , E. Schneider , P. J. Sellers , and J. Shukla , 1994: GCM simulations of the life cycles of the 1988 US drought and heat wave. COLA Tech. Rep. 6, 53 pp. [Available from COLA, 4041 Powder Mill Road, Suite 302, Calverton, MD 20705.].

    • Search Google Scholar
    • Export Citation
  • Frederiksen, C. S. , H. Zhang , R. C. Balgovind , N. Nicholls , W. Drosdosky , and L. Chambers , 2001: Dynamical seasonal forecasts during the 1997/98 ENSO using persisted SST anomalies. J. Climate, 14 , 26752695.

    • Search Google Scholar
    • Export Citation
  • Giard, D. , and E. Bazile , 2000: Implementation of a new assimilation scheme for soil and surface variables in a global NWP model. Mon. Wea. Rev., 128 , 9971015.

    • Search Google Scholar
    • Export Citation
  • Goddard, L. , S. J. Mason , S. E. Zebiak , C. F. Ropelewski , R. Basher , and M. A. Cane , 2000: Current approaches to seasonal to interannual climate predictions. IRI Tech. Reports 00–01, 64 pp.

    • Search Google Scholar
    • Export Citation
  • Houser, P. , 2001: The Global Land Data Assimilation System. Abstracts, Fourth Int. Scientific Conf. on the Global Energy and Water Cycle, Paris, France, GEWEX, 211.

    • Search Google Scholar
    • Export Citation
  • Kalnay, E. , and Coauthors. 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437471.

  • Koster, R. D. , and P. C. D. Milly , 1997: The interplay between transpiration and runoff formulations in land surface schemes used with atmospheric models. J. Climate, 10 , 15781591.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D. , and M. J. Suarez , 2001: Soil moisture memory in climate models. J. Hydrometeor., 2 , 558570.

  • Kumar, A. , M. P. Hoerling , M. Ji , A. Leetmaa , and P. Sardeshmukh , 1996: Assessing a GCM's suitability for making seasonal predictions. J. Climate, 9 , 115129.

    • Search Google Scholar
    • Export Citation
  • Kuo, H. L. , 1974: Further studies of the parameterization of the influence of cumulus convection on large-scale flow. J. Atmos. Sci., 31 , 12321240.

    • Search Google Scholar
    • Export Citation
  • Mahfouf, J-F. , and P. Viterbo , 2001: Land surface assimilation. ECMWF Meteorological Training Course Note, ECMWF, Shinfield Park, Reading, United Kingdom, 23 pp.

    • Search Google Scholar
    • Export Citation
  • Manabe, S. , and J. L. Holloway , 1975: The seasonal variation of the hydrologic cycle as simulated by a global model of the atmosphere. J. Geophys. Res., 80 , 16171649.

    • Search Google Scholar
    • Export Citation
  • McAvaney, B. J. , and G. D. Hess , 1996: The revised surface fluxes parameterisation in the BMRC AGCM. BMRC Report 56, Bureau of Meteorology Research Centre, Melbourne, Australia, 27 pp.

    • Search Google Scholar
    • Export Citation
  • Mintz, Y. , and Y. V. Serafini , 1989: Global monthly climatology of soil moisture and water balance. Note Interne LMD 148, Laboratoire de Meteorologie Dynamique, Centre National de la Recherche Scientifique, Ecole Normale Superieure, Paris, France, 102 pp.

    • Search Google Scholar
    • Export Citation
  • National Climate Center , 1998: '98 Chinese Grand Flood and Climate Anomaly (in Chinese). Meteorology Press, 139 pp.

  • Nicholls, N. , 2001: The insignificance of significant testing. Bull. Amer. Meteor. Soc., 81 , 981986.

  • Oglesby, R. J. , 1991: Springtime soil moisture, natural climatic variability, and North American drought as simulated by the NCAR Community Climate Model 1. J. Climate, 4 , 890897.

    • Search Google Scholar
    • Export Citation
  • Oglesby, R. J. , and D. J. Erickson III , 1989: Soil moisture and the persistence of North American drought. J. Climate, 2 , 13621380.

    • Search Google Scholar
    • Export Citation
  • Parker, D. E. , P. D. Jones , A. Bevan , and C. K. Folland , 1994: Interdecadal changes of surface temperature since the late 19th century. J. Geophys. Res., 99 , 1437314399.

    • Search Google Scholar
    • Export Citation
  • Power, S. , F. Tseitkin , S. Torok , B. Lavery , R. Dahni , and B. McAvaney , 1998: Australian temperature, Australian rainfall and the Southern Oscillation, 1910–1992: Coherent variability and recent changes. Aust. Meteor. Mag., 47 , 85101.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W. , and T. M. Smith , 1995: A high-resolution global sea surface temperature climatology. J. Climate, 8 , 15711583.

  • Robock, A. , C. A. Schlosser , K. Y. Vinnikov , N. A. Spernskaya , J. K. Entin , and S. Qiu , 1998: Evaluation of the AMIP soil moisture simulations. Global Planet. Change, 19 , 181208.

    • Search Google Scholar
    • Export Citation
  • Robock, A. , K. Y. Vinnikov , G. Srinivasan , J. K. Entin , S. E. Hollinger , N. A. Speranskaya , S. Liu , and A. Namkhai , 2000: The Global Soil Moisture Data Bank. Bull. Amer. Meteor. Soc., 81 , 12811299.

    • Search Google Scholar
    • Export Citation
  • Scott, R. L. , D. Entekhabi , R. Koster , and M. Suarez , 1997: Timescales of land surface evapotranspiration response. J. Climate, 10 , 559566.

    • Search Google Scholar
    • Export Citation
  • Seaman, R. , W. Bourke , P. Steinle , T. Hart , G. Embery , M. Naughton , and L. Rikus , 1995: Evolution of the Bureau of Meteorology's Global Assimilation and Prediction System. Part 1: Analysis and initialisation. Aust. Meteor. Mag., 44 , 118.

    • Search Google Scholar
    • Export Citation
  • Shao, Y. , and A. Henderson-Sellers , 1996: Validation of soil moisture simulation in land surface parameterisation schemes with HAPEX data. Global Planet. Change, 13 , 1146.

    • Search Google Scholar
    • Export Citation
  • Shukla, J. , and Y. Mintz , 1982: Influence of land-surface evapotranspiration on the Earth's climate. Science, 215 , 14981501.

  • Shukla, J. , D. A. Paolino , D. M. Straus , D. De Witt , M. Fennessy , J. L. Kinter , L. Marx , and R. Mo , 2000: Dynamical seasonal predictions with the COLA atmospheric model. Quart. J. Roy. Meteor. Soc., 126 , 22652292.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I. , and A. H. Lynch , 1991: The influence of pre-existing soil moisture content on Australian winter climate. Int. J. Climatol., 12 , 3354.

    • Search Google Scholar
    • Export Citation
  • Sud, Y. C. , and M. J. Fennessy , 1984: Influence of evaporation in semi-arid regions on the July circulation: A numerical study. J. Climatol., 4 , 383398.

    • Search Google Scholar
    • Export Citation
  • Sud, Y. C. , D. M. Mocko , G. K. Walker , and R. D. Koster , 2001: Influence of land surface fluxes on precipitation: Inferences from simulations forced with four ARM–CART SCM datasets. J. Climate, 14 , 36663691.

    • Search Google Scholar
    • Export Citation
  • Viviand, J. , S. Lirola , B. Timbal , S. Power , and R. Colman , 2000: Impact of soil moisture on climate variability and predictability. BMRC Res. Rep. 81, BMRC, Melbourne, Australia, 49 pp.

    • Search Google Scholar
    • Export Citation
  • Walker, J. P. , and P. Houser , 2001: A methodology for initializing soil moisture in a global climate model: Assimilation of near-surface soil moisture observations. J. Geophys. Res., 106 , 1176111774.

    • Search Google Scholar
    • Export Citation
  • Webster, P. J. , T. Palmer , M. Yanai , V. Magaña , J. Shukla , and T. Yasunari , 1998: Monsoons: Processes and predictability and prospect for prediction. J. Geophys. Res., 103 , 1445114510.

    • Search Google Scholar
    • Export Citation
  • Wilson, M. F. , and A. Henderson-Sellers , 1985: A global archive of land cover and soils data for use in general circulation models. J. Climatol., 5 , 119143.

    • Search Google Scholar
    • Export Citation
  • Xie, P. , and P. A. Arkin , 1996: Analyses of global monthly precipitation using gauge observations, satellite estimates, and numerical model predictions. J. Climate, 9 , 840858.

    • Search Google Scholar
    • Export Citation
  • Yeh, T. C. , R. T. Wetherald , and S. Manabe , 1984: The effect of soil moisture on the short-term climate and hydrology change—A numerical experiment. Mon. Wea. Rev., 112 , 474490.

    • Search Google Scholar
    • Export Citation
  • Zhang, H. , 2002: A version of the BAM's AMIP2 simulation over the Australian region: Contrasting its performance with other AMIP2 models. Bureau of Meteorology Research Centre Rep., in press.

    • Search Google Scholar
    • Export Citation
  • Zhang, H. , and C. S. Frederiksen , 2001: Experimental dynamical seasonal forecasts: A model performance over East Asia during the 1997–98 El Niño event. Dynamics of Atmospheric and Oceanic Circulations and Climate, Chinese Academy of Science, Ed., China Meteorological Press, 827 pp.

    • Search Google Scholar
    • Export Citation
  • Zhang, H. , A. Henderson-Sellers , P. Irannejad , S. Sharmeen , T. Phillips , and K. McGuffie , 2002a: Land-surface modelling and climate simulations: Results over the Australian region from sixteen AMIP2 models. Bureau of Meteorology Research Centre Rep. 89, 18 pp.

    • Search Google Scholar
    • Export Citation
  • Zhang, H. , and Coauthors. 2002b: BMRC land-surface modelling and related studies: Current status and future plans—A report from BMRC climate forums and discussions. Bureau of Meteorology Research Centre Rep., in press.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

(a) The location of the observational soil moisture datasets used in this study. The observational data are from the Global Soil Moisture Data Bank of Robock et al. (2000). (b) Comparison of top 1-m plant available monthly soil moisture (mm) for the period of 1981–88 observed at locations 32, 33, 34, 36 as in (a). It shows the similarity of soil moisture variations observed in the four locations (differentiated by different line format)

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 2.
Fig. 2.

Comparison of soil moisture variations for the period of 1981–88 among observations, the NCEP–NCAR reanalysis, and the BMRC model 10-yr simulations over the region covered by stations 32, 33, 34, and 36 in Fig. 1. The observational values are the averages of these four stations. (a) Soil moisture variations (mm) for the period of 1981–88; (b) soil moisture seasonal cycle derived from the data covering the period of 1981–88. The Mintz and Serafini (1989) soil moisture climatology used in the BMRC model initialization is also included. Results are the percentage values referring to its climatological annual mean, respectively

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 3.
Fig. 3.

Differences of soil moisture initial conditions between the one used in the model soil moisture sensitivity experiment (SI) and the one in the model original experiments (ORG). Only positive values are shaded

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 4.
Fig. 4.

Forecasts of 1998 JJA rainfall anomalies (mm day−1) from five sets of experiments. Only positive values are shaded. (a) Forecasts from the original setup as in Frederiksen et al. (2001) in which the anomalous forecasts are the differences between the ORG and CTL experiments; (b) differences between the SI and ORG experiments due to different soil moisture initialization; (c) differences between the OS and ORG experiments due to forcing the model with observed global SST anomalies in the model integration; (d) precipitation anomalies derived from an updated version of the Xie and Arkin (1996) observational data

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 5.
Fig. 5.

As in Fig. 2 but for surface air temperature (K). Observational data are from an updated version of the Parker et al. (1994) dataset

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 6.
Fig. 6.

Variation of soil moisture (mm) simulated in the model at the location of 52.5°N, 75°E. Solid lines are from the six SI ensemble runs and dotted lines from the six ORG ensemble runs. Heavy solid and dotted lines are the ensemble averages of results from the SI and ORG experiments, accordingly

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 7.
Fig. 7.

(a) Differences between the SI and ORG experiments in surface latent heat flux (W m−2) in the seasonal averages of JJA98; (b) as in (a) but for surface sensible heat flux

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 8.
Fig. 8.

(a) Areal averages (25°–32.5°N, 105°–120°E) of precipitation (mm day−1) and latent heat flux (W m−2) from 6-hourly five-ensemble runs during the integration of JJA97 as in Frederiksen et al. (2001); (b) as in (a) but for 850-hPa horizontal vorticity (s−1)

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 9.
Fig. 9.

(a) Correlation between 850-hPa specific humidity (q) and surface evaporation (evp) during the ORG integration in JJA98; (b) correlation between 850-hPa specific humidity (q) and surface moisture convergence (conv) during the ORG integration in JJA98

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 10.
Fig. 10.

Horizontal moisture transport (vector: mg s−1 kg−1) at 10-m height and associated moisture convergence (shaded: ×10−6 g s−1 kg−1) between four sets of experiments. (a) Differences between the ORG and CTL experiments; (b) differences between SI and ORG; (c) differences between OS and ORG; (d) anomalies derived from the 40-yr NCEP–NCAR reanalysis

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 11.
Fig. 11.

Differences between the SI and ORG experiments in the numbers of trough positions derived from 6-h model output for the period of JJA98.

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 12.
Fig. 12.

As in Fig. 4 but for JJA98 precipitation (mm day−1) forecasts over the Australian region

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Fig. 13.
Fig. 13.

As in Fig. 10 but over the Australian region

Citation: Journal of Climate 16, 13; 10.1175/1520-0442(2003)16<2117:LANIOS>2.0.CO;2

Table 1.

Description of experiments conducted in this study. All the experiments are 120-day model integration with six ensemble runs

Table 1.
Table 2.

Comparison of JJA precipitation (mm) simulated in the model by each set of the ensemble runs as described in Table 1. Results are over the location of 25°N, 117.5°E

Table 2.
Save
  • Bonan, G. B. , and L. M. Stillwell-Soller , 1998: Soil water and the persistence of floods and droughts in the Mississippi River Basin. Water Resour. Res., 34 , 26932701.

    • Search Google Scholar
    • Export Citation
  • Bouttier, F. , J-F. Mahfouf , and J. Noilhan , 1993: Sequential assimilation of soil moisture from atmospheric low-level parameters. Part I: Sensitivity and calibration studies. J. Appl. Meteor., 32 , 13351350.

    • Search Google Scholar
    • Export Citation
  • Brankovic, C. , and T. N. Palmer , 2000: Seasonal skill and predictability of ECMWF PROVOST ensembles. Quart. J. Roy. Meteor. Soc., 126 , 20352067.

    • Search Google Scholar
    • Export Citation
  • Charney, J. G. , and J. Shukla , 1981: Predictability of monsoons. Monsoon Dynamics, J. Lighthill and R. P. Pearce, Eds., Cambridge University Press, 99–109.

    • Search Google Scholar
    • Export Citation
  • Colman, R. A. , and B. J. McAvaney , 1995: Sensitivity of the climate response of an atmospheric general circulation model to changes in convective parameterization and horizontal resolution. J. Geophys. Res., 100 , 31553172.

    • Search Google Scholar
    • Export Citation
  • Dirmeyer, P. A. , 2000: Using a global soil wetness dataset to improve seasonal climate simulation. J. Climate, 13 , 29002922.

  • Douville, H. , 2002: Influence of soil moisture on the Asian and African monsoons. Part II: Interannual variability. J. Climate, 15 , 701720.

    • Search Google Scholar
    • Export Citation
  • Douville, H. , and F. Chauvin , 2000: Relevance of soil moisture for seasonal climate predictions: A preliminary study. Climate Dyn., 16 , 719736.

    • Search Google Scholar
    • Export Citation
  • Emanuel, K. A. , 1994: Atmospheric Convection. Oxford University Press, 580 pp.

  • Entekhabi, D. , I. Rodriguez-Iturbe , and R. L. Bras , 1992: Variability in large-scale water balance with land surface–atmosphere interaction. J. Climate, 5 , 798813.

    • Search Google Scholar
    • Export Citation
  • Entekhabi, D. , and Coauthors. 1999: An agenda for land surface hydrology research and a call for the second international hydrology decade. Bull. Amer. Meteor. Soc., 80 , 20432058.

    • Search Google Scholar
    • Export Citation
  • Fan, Z. , and R. J. Oglesby , 1996: A 100-yr CCM1 simulation of North China's hydrologic cycle. J. Climate, 9 , 189204.

  • Fennessy, M. J. , and J. Shukla , 1999: Impact of initial soil wetness on seasonal atmospheric prediction. J. Climate, 12 , 31673180.

    • Search Google Scholar
    • Export Citation
  • Fennessy, M. J. , J. L. Kinter III , L. Marx , E. Schneider , P. J. Sellers , and J. Shukla , 1994: GCM simulations of the life cycles of the 1988 US drought and heat wave. COLA Tech. Rep. 6, 53 pp. [Available from COLA, 4041 Powder Mill Road, Suite 302, Calverton, MD 20705.].

    • Search Google Scholar
    • Export Citation
  • Frederiksen, C. S. , H. Zhang , R. C. Balgovind , N. Nicholls , W. Drosdosky , and L. Chambers , 2001: Dynamical seasonal forecasts during the 1997/98 ENSO using persisted SST anomalies. J. Climate, 14 , 26752695.

    • Search Google Scholar
    • Export Citation
  • Giard, D. , and E. Bazile , 2000: Implementation of a new assimilation scheme for soil and surface variables in a global NWP model. Mon. Wea. Rev., 128 , 9971015.

    • Search Google Scholar
    • Export Citation
  • Goddard, L. , S. J. Mason , S. E. Zebiak , C. F. Ropelewski , R. Basher , and M. A. Cane , 2000: Current approaches to seasonal to interannual climate predictions. IRI Tech. Reports 00–01, 64 pp.

    • Search Google Scholar
    • Export Citation
  • Houser, P. , 2001: The Global Land Data Assimilation System. Abstracts, Fourth Int. Scientific Conf. on the Global Energy and Water Cycle, Paris, France, GEWEX, 211.

    • Search Google Scholar
    • Export Citation
  • Kalnay, E. , and Coauthors. 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437471.

  • Koster, R. D. , and P. C. D. Milly , 1997: The interplay between transpiration and runoff formulations in land surface schemes used with atmospheric models. J. Climate, 10 , 15781591.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D. , and M. J. Suarez , 2001: Soil moisture memory in climate models. J. Hydrometeor., 2 , 558570.

  • Kumar, A. , M. P. Hoerling , M. Ji , A. Leetmaa , and P. Sardeshmukh , 1996: Assessing a GCM's suitability for making seasonal predictions. J. Climate, 9 , 115129.

    • Search Google Scholar
    • Export Citation
  • Kuo, H. L. , 1974: Further studies of the parameterization of the influence of cumulus convection on large-scale flow. J. Atmos. Sci., 31 , 12321240.

    • Search Google Scholar
    • Export Citation
  • Mahfouf, J-F. , and P. Viterbo , 2001: Land surface assimilation. ECMWF Meteorological Training Course Note, ECMWF, Shinfield Park, Reading, United Kingdom, 23 pp.

    • Search Google Scholar
    • Export Citation
  • Manabe, S. , and J. L. Holloway , 1975: The seasonal variation of the hydrologic cycle as simulated by a global model of the atmosphere. J. Geophys. Res., 80 , 16171649.

    • Search Google Scholar
    • Export Citation
  • McAvaney, B. J. , and G. D. Hess , 1996: The revised surface fluxes parameterisation in the BMRC AGCM. BMRC Report 56, Bureau of Meteorology Research Centre, Melbourne, Australia, 27 pp.

    • Search Google Scholar
    • Export Citation
  • Mintz, Y. , and Y. V. Serafini , 1989: Global monthly climatology of soil moisture and water balance. Note Interne LMD 148, Laboratoire de Meteorologie Dynamique, Centre National de la Recherche Scientifique, Ecole Normale Superieure, Paris, France, 102 pp.

    • Search Google Scholar
    • Export Citation
  • National Climate Center , 1998: '98 Chinese Grand Flood and Climate Anomaly (in Chinese). Meteorology Press, 139 pp.

  • Nicholls, N. , 2001: The insignificance of significant testing. Bull. Amer. Meteor. Soc., 81 , 981986.

  • Oglesby, R. J. , 1991: Springtime soil moisture, natural climatic variability, and North American drought as simulated by the NCAR Community Climate Model 1. J. Climate, 4 , 890897.

    • Search Google Scholar
    • Export Citation
  • Oglesby, R. J. , and D. J. Erickson III , 1989: Soil moisture and the persistence of North American drought. J. Climate, 2 , 13621380.

    • Search Google Scholar
    • Export Citation
  • Parker, D. E. , P. D. Jones , A. Bevan , and C. K. Folland , 1994: Interdecadal changes of surface temperature since the late 19th century. J. Geophys. Res., 99 , 1437314399.

    • Search Google Scholar
    • Export Citation
  • Power, S. , F. Tseitkin , S. Torok , B. Lavery , R. Dahni , and B. McAvaney , 1998: Australian temperature, Australian rainfall and the Southern Oscillation, 1910–1992: Coherent variability and recent changes. Aust. Meteor. Mag., 47 , 85101.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W. , and T. M. Smith , 1995: A high-resolution global sea surface temperature climatology. J. Climate, 8 , 15711583.

  • Robock, A. , C. A. Schlosser , K. Y. Vinnikov , N. A. Spernskaya , J. K. Entin , and S. Qiu , 1998: Evaluation of the AMIP soil moisture simulations. Global Planet. Change, 19 , 181208.

    • Search Google Scholar
    • Export Citation
  • Robock, A. , K. Y. Vinnikov , G. Srinivasan , J. K. Entin , S. E. Hollinger , N. A. Speranskaya , S. Liu , and A. Namkhai , 2000: The Global Soil Moisture Data Bank. Bull. Amer. Meteor. Soc., 81 , 12811299.

    • Search Google Scholar
    • Export Citation
  • Scott, R. L. , D. Entekhabi , R. Koster , and M. Suarez , 1997: Timescales of land surface evapotranspiration response. J. Climate, 10 , 559566.

    • Search Google Scholar
    • Export Citation
  • Seaman, R. , W. Bourke , P. Steinle , T. Hart , G. Embery , M. Naughton , and L. Rikus , 1995: Evolution of the Bureau of Meteorology's Global Assimilation and Prediction System. Part 1: Analysis and initialisation. Aust. Meteor. Mag., 44 , 118.

    • Search Google Scholar
    • Export Citation
  • Shao, Y. , and A. Henderson-Sellers , 1996: Validation of soil moisture simulation in land surface parameterisation schemes with HAPEX data. Global Planet. Change, 13 , 1146.

    • Search Google Scholar
    • Export Citation
  • Shukla, J. , and Y. Mintz , 1982: Influence of land-surface evapotranspiration on the Earth's climate. Science, 215 , 14981501.

  • Shukla, J. , D. A. Paolino , D. M. Straus , D. De Witt , M. Fennessy , J. L. Kinter , L. Marx , and R. Mo , 2000: Dynamical seasonal predictions with the COLA atmospheric model. Quart. J. Roy. Meteor. Soc., 126 , 22652292.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I. , and A. H. Lynch , 1991: The influence of pre-existing soil moisture content on Australian winter climate. Int. J. Climatol., 12 , 3354.

    • Search Google Scholar
    • Export Citation
  • Sud, Y. C. , and M. J. Fennessy , 1984: Influence of evaporation in semi-arid regions on the July circulation: A numerical study. J. Climatol., 4 , 383398.

    • Search Google Scholar
    • Export Citation
  • Sud, Y. C. , D. M. Mocko , G. K. Walker , and R. D. Koster , 2001: Influence of land surface fluxes on precipitation: Inferences from simulations forced with four ARM–CART SCM datasets. J. Climate, 14 , 36663691.

    • Search Google Scholar
    • Export Citation
  • Viviand, J. , S. Lirola , B. Timbal , S. Power , and R. Colman , 2000: Impact of soil moisture on climate variability and predictability. BMRC Res. Rep. 81, BMRC, Melbourne, Australia, 49 pp.

    • Search Google Scholar
    • Export Citation
  • Walker, J. P. , and P. Houser , 2001: A methodology for initializing soil moisture in a global climate model: Assimilation of near-surface soil moisture observations. J. Geophys. Res., 106 , 1176111774.

    • Search Google Scholar
    • Export Citation
  • Webster, P. J. , T. Palmer , M. Yanai , V. Magaña , J. Shukla , and T. Yasunari , 1998: Monsoons: Processes and predictability and prospect for prediction. J. Geophys. Res., 103 , 1445114510.

    • Search Google Scholar
    • Export Citation
  • Wilson, M. F. , and A. Henderson-Sellers , 1985: A global archive of land cover and soils data for use in general circulation models. J. Climatol., 5 , 119143.

    • Search Google Scholar
    • Export Citation
  • Xie, P. , and P. A. Arkin , 1996: Analyses of global monthly precipitation using gauge observations, satellite estimates, and numerical model predictions. J. Climate, 9 , 840858.