1. Introduction
It is well recognized that part of the land surface’s memory lies in soil moisture (Koster and Suarez 2001; Liu and Avissar 1999), which plays an important role in enhancing the predictability of climate and hydrological models (Dirmeyer 2003). Soil moisture is an important factor that influences the climate by modulating sensible and latent heat fluxes, and hence energy and water exchanges or interactions between atmosphere and land surface. Land–atmosphere interactions can trigger mesoscale circulation (Weaver and Avissar 2001), change the local planetary boundary layer (Betts and Ball 1998), and regulate regional water recycling (Dirmeyer and Brubaker 1999). Soil moisture is the most significant boundary condition that controls summer precipitation over large midlatitude continental regions and holds essential initial information for seasonal predictions (Koster et al. 2000, 2003, 2004; Koster and Suarez 2004; Salvucci et al. 2002).
Soil moisture observations at large scales are critical for a variety of applications, including assimilation into weather forecasting, monitoring of crop growth and drought, assessing initial conditions in flood forecasting, and quantification of the earth’s water budget. An accurate assessment of spatial and temporal variation of soil moisture may therefore be useful for improving the predictive capability of runoff models, and for improving and validating hydrologic process representations. Unfortunately, there is no global in situ observational network for soil wetness (Fennessy and Shukla 1999), although approaches to set up such a network exist (Robock et al. 2000). The problem remains that very little of the earth’s land surface area lies within a reasonable radius of an in situ soil moisture measurement site (Dirmeyer et al. 2004). Useful in situ observations are rarely available as area representative measurements and they are expensive and tedious to collect (Hollinger and Isard 1994). Direct in situ measurements are either based on gravimetric sampling or time domain reflectometry. However, they are invariably point measurements. It is very difficult to estimate catchment-average soil moisture from such point estimates because of the immense spatial soil moisture variability at small scales (Western et al. 2002). Also, because of logistic constraints, the spatial coverage of in situ measurements is usually rather limited.
Therefore, model output for soil moisture is usually used to perform research at large scale (Robock et al. 2000). The Land Data Assimilation System (LDAS) is forced with real-time output from numerical prediction models, satellite data, and radar precipitation measurements, and is therefore not affected by numerical weather prediction (NWP) forcing biases (Mitchell et al. 2004).
The difficulty of measuring soil moisture on the ground has motivated considerable research in the field of remote sensing (Engman and Chauhan 1995; Kustas et al. 2005; Pellarin et al. 2003). The thickness of the soil layer directly accessible to microwaves generally decreases with increasing frequency and soil moisture content. Microwave sensors operating at longer wavelengths are the most suitable for collecting soil moisture information (Wagner et al. 2007a, b). Even though C-band data may be contaminated with radio frequency interference (RFI)—especially at 6.8–6.9 GHz—the active scatterometer employed in this study working at 5.3 GHz is not affected by this phenomenon. Microwave remote sensing offers the possibility to retrieve soil moisture information at various scales. The main appeal of remote sensing methods is that they provide average estimates over areas (or footprints) that may range from a few square meters to thousands of square kilometers, depending on the method. Hence, there is no need to infer areal averages from point data as the remotely sensed data directly come as areal averages. These traits have motivated much research in the field of remote sensing to retrieve soil moisture (Engman and Chauhan 1995). Microwave sensors offer a relatively direct means of assessing soil moisture since they exploit, like many in situ observation techniques, the strong relationship between the moisture content and dielectric constant of the soil. They can acquire imagery unimpeded by cloud cover during day and night but cannot provide soil moisture information when the soil is frozen or snow covered, or if the soil is completely vegetation covered as in rain forest areas.
The first multiyear global soil moisture dataset derived from European Remote Sensing Satellite (ERS) scatterometer data for the years 1992–2000 (available online at http://www.ipf.tuwien.ac.at/radar/ers-scat/home.htm) was presented by Wagner et al. (2003). This dataset comprises retrieved surface soil moisture data and a so-called soil water index (SWI; Wagner et al. 1999b). The SWI is a measure of the moisture data content obtained by filtering the surface moisture time series with an exponential function (Wagner et al. 1999a). So far, few studies have checked the accuracy of surface soil moisture data retrieved from the ERS scatterometer at regional scales. Studies have focused on either one test area (one scatterometer pixel) or global patterns (Scipal 2002; Wagner et al. 2007b). Dirmeyer et al. (2004) compared the SWI with seven other global wetness products, three produced by land surface model calculations, three from coupled land–atmosphere reanalysis, and furthermore with a so-called soil wetness index dataset derived from Special Sensor Microwave Imager (SSM/I) data (Basist et al. 1998). They found that while the SSM/I data have clearly a different character from all the other datasets, ERS scatterometer data revealed many similarities with the simulated wetness products. A study conducted by Drusch et al. (2004) compared the ERS-derived surface soil moisture data to in situ volumetric soil moisture data at 10-cm depth collected during the Southern Great Plains Hydrology Experiment (SGP99) and obtained a coefficient of determination of R2 = 0.43. Ceballos et al. (2005) compared the SWI to 0–100-cm soil moisture data and obtained a coefficient of determination of R2 = 0.75 and a root-mean-square error of 0.022 m3 m−3. With ERS-derived SWI, Zhao et al. (2006) found that this data can improve the simulation accuracy of heavy rainfall events in East Asia. ERS scatterometer-derived soil moisture along with other three published satellite datasets were compared to high-quality soil moisture data from a test site located in the Duero basin in Spain (Wagner et al. 2007b).
Compared to synthetic aperture radar (SAR) systems scatterometers offer multiple viewing direction capability, which enables one to better account for the effects of vegetation and surface roughness. Furthermore, contrary to SAR, lower-resolution active scatterometer sensors allow one to map the earth’s surface within less than 3 days. Here a coarse spatial resolution of 50 km (ERS Scat) or 25 km [Meteorological Operation satellite (MetOp) Advanced Scatterometer (ASCAT)] is accepted, since an excellent temporal resolution can be achieved. Many authors agree that soil moisture can be investigated at two different spatial scales. The first is the spatial scale below 100 m, where spatial and temporal soil moisture variability are mainly driven by vegetation, soil type, and topography (Scipal et al. 2005; Vachaud et al. 1985). The second scale at several kilometers represents soil moisture variability induced by atmospheric forcing effects, thus mainly being influenced by climatic conditions and large-scale precipitation events and differences in evaporation (and thus temperature) (Vinnikov et al. 1999; Ceballos et al. 2002; Entin et al. 2000).
In this paper in situ soil moisture and precipitation data are compared with ERS scatterometer-derived SWI data. It is unique to this study that soil moisture patterns are investigated at the local, regional, and countrywide scale. The main focus of this paper is to evaluate the use of the SWI as an index to monitor water availability including water stress at the macroscale in China for the 9-yr time period 1992–2000. Section 2 describes the data used. Comparisons between area-averaged soil water (SWI and in situ soil moisture) and precipitation data in one local area are presented in section 3. Comparisons between area-averaged SWI and precipitation in east (EC), southwest (SWC), and northern China (NC) will be presented in section 4. Section 5 depicts the relationship between SWI and precipitation patterns in China. Thus, the individual sections reflect the three different scales: the local scale in section 3, the regional scale in section 4, and the countrywide scale in section 5. Conclusions are given in section 6.
2. Data description
a. SWI data
The ERS scatterometer is part of the payload of the ERS-1 (1991–96) and ERS-2 (1995–present). It is the first spaceborne active scatterometer to provide global coverage for several years. The recorded backscattering coefficient is sensitive to vegetation and, for low vegetation types, like grass and agricultural crops, to the surface soil moisture content (Wagner et al. 1999a). This duality has led to two different types of studies: those that employ the ERS scatterometer data for vegetation applications (Frison and Mougin 1996) and those that monitor soil moisture (Magagi and Kerr 1997).
The ERS scatterometer is a vertically polarized radar operating at C band (5.3-GHz domain) (Wagner et al. 1999b). The sensor acquires imagery independent of cloud cover and independent of solar illumination. The instrument measures the backscattering coefficient from three different viewing directions using three sideways-looking antennas. One antenna is normal to the satellite track, one is pointing 45° forward, and one is pointing 45° backward with respect to the satellite flight track. Because of the multiple viewing capabilities of the instrument, it is possible to differentiate temporal vegetation and soil moisture effects on the signal. The spatial resolution of the ERS scatterometer data is about 50 km and its temporal resolution is about 2–3 days. From this data SWI is calculated by the Vienna University of Technology every 10 days. The data are interpolated to 0.25° × 0.25°, which is about a 28-km resolution at middle latitude in China.
The basic assumptions for SWI calculation are as follows: electromagnetic waves transmitted by scatterometers penetrate only a few centimeters into the soil surface. Therefore the signals scattered back to the sensor only provide information about the moisture content in the soil surface layer, the so-called topsoil moisture. However, the topsoil moisture may change significantly within a few hours. Its magnitude furthermore depends on the amount of rainfall, the evaporation rate, and the time that has passed since the rainfall event.
b. In situ relative soil moisture data
In situ soil moisture data, provided by the National Climatic Data Center of the Chinese Meteorological Administration, include relative soil moisture (θf), field capacity (FC), and bulk density of soil (ρ). Soil moisture is recorded at depths of 10, 20, 50, 70, and 100 cm and measured at agricultural meteorological gauge stations in China on the 8th, 18th, and 28th of each month for the years 1992–2000. The term “in situ soil moisture” will be adopted in the discussion.
The distribution of the agricultural meteorological gauge stations shows that most measurements were collected north of 32°N in China. Because of the scarcity of in situ soil moisture in most regions of China, it is of great importance to use remote sensing data to derive soil moisture distribution in areas less densely covered with in situ measuring networks.
c. Precipitation data
Global Precipitation Climatology Centre (GPCC) 50-yr precipitation data from 1951 to 2000 (available online at http://www.dwd.de/vasclimo), based on quality-controlled and homogenized time series from 9343 stations, are employed in this study (Fuchs et al. 2007). GPCC precipitation data cover the global land areas with a spatial resolution of 0.5° × 0.5°. They are developed on the basis of the most comprehensive database of monthly observed precipitation data worldwide that resides with the GPCC within the framework of the German Climate Research Program (DEKLIM)-funded research project Variability Analysis of Surface Climate Observations (VASClimO). Precipitation data are represented by the precipitation rate.
For this study, monthly averaged soil moisture data (SWI and in situ) were adopted. Using the area-averaged data instead of site data can significantly reduce the measurement uncertainties (Wu et al. 2001). Furthermore, since all datasets (SWI, GPCC, in situ soil moisture) are available as long-term time series, anomalies for each month (with respect to the typical monthly mean) can be calculated. These anomalies indicate deviations from the mean. Strong anomalies in the data can indicate extraordinarily dry or wet periods.
3. Local scale: Comparison of area-averaged SWI, precipitation, and in situ soil moisture
Soil moisture is sensitive to precipitation. From a hydrologic viewpoint, soil moisture controls the partitioning of rainfall into runoff and infiltration and therefore has an important effect on the runoff behavior of catchments (Aubert et al. 2003). Since in situ soil moisture is scarce, it is difficult to be employed for analysis at a regional scale. However, in situ soil moisture data collected at eight stations (32.9°–34.88°N, 112.5°–116°E) can be used here to perform comparisons with SWI and GPCC precipitation data as shown in Fig. 1.
Correlation coefficients between area-averaged SWI and precipitation are 0.479 comparing the two time series, and 0.433 for the anomalies (the limit of the 0.01 level of confidence for correlation coefficient is 0.25, the same in the below). This shows that SWI and SWI anomaly series can very well express precipitation variations.
Correlation coefficients between precipitation and in situ soil moisture at different vertical layers (10, 20, and 50 cm) are 0.367, 0.313, and 0.186, which shows that there is also significant correlation between precipitation and in situ soil moisture at the first two layers according to the 0.01 level confidence tests. The lower correlation coefficient between precipitation and soil moisture at 50 cm expresses that precipitation influence is less accentuated at deeper layers. Correlation coefficients of in situ soil moisture at 10-, 20-, and 50-cm layers with the SWI are 0.453, 0.517, and 0.444, respectively, and 0.518, 0.492, and 0.351 for the anomalies, which are all significant. Since in situ soil moisture is measured on 8th, 18th, and 28th of each month and usually slightly lagging behind precipitation, the correlation between SWI and precipitation is higher than that between in situ soil moisture and precipitation.
Furthermore, it should be emphasized that the SWI correlated better to precipitation than the in situ soil moisture data to precipitation. One reason for this can be the fact that GPCC data and SWI data are of equal spatial resolution, while the in situ data represent point measurements. However, another possibility is that the SWI data simply represent the precipitation events better than the in situ data. Errors in the in situ data could account for that. In Fig. 1 it is obvious that the course of SWI seems to follow precipitation better than the in situ data, especially when precipitation decreases around the end of the year and in situ soil moisture still remains high. It should furthermore be mentioned that the years 1998 and 2000 were El Niño years with above-average precipitation in many areas worldwide. Above-average precipitation for these years can also be observed in Fig. 1b.
Annual and interannual variations of soil water content and precipitation are shown in Fig. 2. Annual variations of soil water content (SWI and in situ) and precipitation show that the precipitation focuses on summertime. The SWI depicts similar change tendencies as in situ soil moisture, which shows lower values in spring and winter, and greater values in summer. Interannual variations of SWI anomaly distribution can demonstrate its relationship with in situ soil moisture and precipitation. June–August (JJA) interannual variations further show soil water contents’ interannual variations due to the effect of summer precipitation. Soil water contents and precipitation variations match well (Figs. 2c,d).
4. Regional scale: Comparison of area-averaged SWI and precipitation
In larger regions in situ soil moisture data are usually not available. Here, the relationship between SWI and precipitation can be investigated to evaluate if SWI reflects precipitations patterns well and if it can thus act as an indicator for water availability.
a. Comparisons of area-averaged SWI and precipitation
Three regions covering 26°–32°N, 108°–116°E; 35°–41°N, 106°–116°E; and 26°–32°N, 104°–108°E, which are located in EC, NC, and SWC, representing different precipitation and soil water distribution regimes, are analyzed. Area-averaged SWI and precipitation in EC, NC, and SWC for the years 1992–2000 are shown in Fig. 3. All three regions have precipitation peaks in summer and reach lowest values in winter. However, SWI distributions in the three local areas show great differences.
In EC and SWC, the SWI shows higher values in summertime than in wintertime. This is in accordance with precipitation distribution. The SWI maximum appears in summer due to the stronger precipitation. However, the behavior is different for SWI distributions in NC (Fig. 3b), where ground is frozen or snow covered in wintertime. As the frozen ground or/and snowpack melt in early spring, a rapid increase of the SWI can be noted. Therefore, two maxima can be found for SWI distributions, the former benefiting from frozen ground or snowpack melting in early spring, and the latter—higher one—resulting from summer precipitation. Correlation coefficients between SWI and precipitation (shown in Table 1) are 0.695 and 0.694 in EC and SWC. With only 0.539 it is lower in NC (they are all significant according to the 0.01 level of confidence test). The lower correlation coefficient in NC may result from the rapid increase of the SWI in early spring, where precipitation itself is weak.
Meanwhile, there are differences in SWI variations for the three regions. In EC the SWI maintains high values, due to a positive precipitation anomaly in the winters of 1994/95 and 1997/98. The anomalies for SWI and precipitation depict similar change tendencies (Fig. 4). Correlation coefficients of the anomalies in EC, SWC, and NC are 0.458, 0.397, and 0.453 (Table 1), which all are significant.
Distributions of averaged precipitation rate and standard deviation (SD) variations in three regions show that the precipitation rate is strongest in EC and lowest in NC. Precipitation decreases accordingly from EC to NC. However, SD ratios between SWI and precipitation are higher in NC and weaker in EC. This indicates that SWI variations in NC are greater than in EC (snowmelt effect).
b. Interannual and annual variations
Interannual (yearly) variations between SWI and precipitation for the years 1992–2000 are presented in Figs. 5a,b,c. Figures 5a,b,c show that SWI and precipitation match well. The SWI reflects the majority of extreme positive or negative anomalies of precipitation, such as the extreme drought in 1992, floods in 1993 and 1998 in EC (Fig. 5a), the extreme drought in 1997 and floods in 1993 and 1999 in SWC (Fig. 5c), and the extreme drought in 1997 in NC (Fig. 5b).
However, there is no coincidence between SWI and precipitation anomalies in some years, such as 1996 in EC and 1995 in SWC. The precipitation rate in 1996 is greater than in 1995 and 1997 in EC. However, the SWI shows the opposite change tendency. Interannual variations (Figs. 5a,b,c) represent the distributions during the whole year. However, JJA SWI interannual variations can better express precipitation variations and disclose extreme climate events in summer (Figs. 5d,e,f).
SWI and precipitation annual variations (Fig. 6) in EC and SWC both show maxima in summertime and match well. However, there also exists another SWI maximum in October in SWC and SWI varies little from July to October. The behavior is different for SWI distributions in NC. Two maxima for SWI variations exist in NC, in which the first benefits from the melting of frozen ground or snow in early spring, while the latter results from stronger summer precipitation.
5. Countrywide scale: SWI and precipitation distributions in China
Here precipitation and SWI distributions for the countrywide scale (whole of China) are analyzed. First, SWI and precipitation are compared (presented for selected areas of China). Second, SWI seasonal variations are discussed.
a. Comparisons between SWI and precipitation
Annual variations of monthly averaged SWI and GPCC precipitation of 9 yr (1992–2000) are shown in Figs. 7 and 8 (frozen ground or snowpack, and data missing when SWI is above 100). The majority of precipitation occurs during summertime. There are significant differences for the start and end time of the rainy season in different regions (Figs. 7d–i). Significant regional distribution characteristics for the rain belt can be found in the east. The rainy period results from the migration of the large-scale rain belt, which is connected with the subtropical ridge, the 100-hPa Tibet high, the subtropical westerly jet, and the East Asia monsoon (Zhu et al. 2000). Meanwhile, the transformation between dry and wet seasons in the west is more obvious. For 9-yr averaged precipitation, generally speaking, precipitation decreases from the southeast to the northwest. In the following, more detailed explanations of some relevant areas in China are presented.
As the rainy belts are mainly located in eastern China, only the comparisons between the moving of rainy belts and SWI variations in eastern China will be discussed here.
1) The spring rainy season to the south of the Yangtze River valley
The rain belt moves to the south of the Yangtze River valley with weak rainfall from March to May (shown by Figs. 7c–e). The SWI increases noticeably from March and reaches a maximum in April (shown by Figs. 8c–e).
2) The preflood season in South China
The preflood season in south China lasts from April to June (Figs. 7d–f). The precipitation mainly occurs in the westerly belt to the north of the subtropical high. The rainfall increases significantly from April and is strongest in May. There are two strong rainfall centers, which are located in the northern and the southern part of the rainfall area, respectively. The former one benefits from the frontal rainfall due to cold air from the north. The latter one results from the influence of the East Asia monsoon.
The corresponding SWI distributions show that SWI values are about 60%–70% and can reach 70%–80% in certain areas (see Figs. 8d–f). The SWI increases obviously from April with an enhanced 60%–70% SWI covered area. The SWI high-value area, which results from the rainfall center in the north, can extend to the Yangtze River valley. The SWI increases significantly on the seashore of south China in May and June due to East Asia monsoon precipitation.
3) The Mei-yu in the Yangtze River valley
The rain belt is located in the Yangtze River valley from the middle of June to early July, and related rains are referred to as the Mei-yu (shown by Figs. 7f–g). Precipitation occurs as continuous overcast rain, accompanied by frequent heavy rainfall with the maximum rainfall intensity through the year. The corresponding SWI distributions (Figs. 8f–g) show that the SWI high-value area is connected with areas influenced by the prior flood season. The SWI decreases in the region in July as a result of the end of the Mei-yu and increasing temperature.
4) The rainy season in north and northeast China
The rain belt moves to the north and the northeast of China from the middle of July to late August and induces the rainy season in this region (Figs. 7g,h). SWI distributions show that soil moisture increases from 30%–50% in June to 60%–70% in July and August (Figs. 8g,h).
With the migration of the rain belt in the rainy season in the east, a relatively dry period exists in certain areas: in the Yangtze River valley, the relatively dry period lasts from the middle of July to the middle of August; however, it starts in the middle of June and ends in late July in the south. No obvious relatively dry period exists in the north and northeast (Zhu et al. 2000). Severe relative dry periods can induce drought in the region. In south China and the Yangtze River valley, there are two concentrated rainy seasons, which lead to SWI regional variations.
b. SWI seasonal variations
Overall, in spring, the SWI is high in the northeast and the south, where high values result from ice or/and snow melting while the south benefits from the spring rainy season to the south of the Yangtze River valley and preflood season precipitation. The SWI is low in the Yellow River valley and in the southwest. From April on, low SWI values appear in Inner Mongolia and in the northwest, and extend greatly, partially remaining till September.
In summer, the SWI in June is also high in the northeast and south; however, SWI high-value areas are more strongly accentuated in the south and the southwest due to the rainy season in the Yangtze River valley (the Mei-yu), in the southwest and at the Qinghai–Tibetan Plateau. In July and August, with the rain belt moving to the north and northeast, the SWI in the north and northeast increases significantly. The SWI is high in the east, except for low values in the Yangtze River valley due to high temperatures there. In August, there are two SWI high-value centers in the north and the northeast respectively. In summer low SWI areas occur in Inner Mongolia and the northwest.
In autumn, with the rainy season ending in the north and northeast, it is similar for SWI distributions in the region between August and September. However, high SWI areas in the north and the northeast are weaker and low SWI areas in the Yangtze River valley are accentuated more strongly with greater covered area in September. In October and November, the SWI distribution in the north shows relatively high values due to less evaporation according to lower temperatures. The low SWI values in the Yangtze River extend southward and westward due to less rainfall in the south and the southwest. The SWI is still low in the northwest in September and October.
In winter, during frozen or snow-covered conditions, the SWI can only be displayed in the south. The SWI is higher in the Yangtze and Huaihe River valley, and lower in most of the south because of less precipitation.
6. Conclusions
The ERS scatterometer, with a spatial resolution of 50 km and temporal resolution of about 3 days, has been collecting data from its satellite platforms ERS-1 (1991–96) and ERS-2 (1995–present). Its parallel successor, with similar technical characteristics, is the Advanced Scatterometer (ASCAT), mounted on the Meteorological Operational satellite (MetOp). It has collected data since December 2006, at improved spatial (25 km) and temporal resolution (1–2 days). A near-real-time soil moisture processor for MetOp ASCAT has been developed by the Vienna University of Technology (TU Vienna) (Bartalis et al. 2005). This grants the extension of TU Vienna’s global soil moisture archive, with the time series now already extending for 15 yr from 1992 on. Each 10 days the soil water index (SWI) is retrieved from this series.
In this paper, SWI has been used as an index to evaluate water availability and water stress. The analysis was performed at three scales, opposite to most published studies, which were done for only a small test site or for the complete global dataset. First, on a local scale, the correlation between SWI, precipitation data, and in situ soil moisture from eight stations has been analyzed. Significant correlations between all three parameters could be retrieved. In the 9-yr time series the monthly average SWI reflected precipitation distribution very well. Also, general seasonal patterns between SWI and in situ moisture coincide well. However, a noteworthy finding is that the correlation between SWI and precipitation is higher than the correlation between precipitation and in situ moisture. SWI, precipitation, and in situ moisture anomalies show the same deviations over the 9-yr period. For the local area investigated SWI strongly reflects precipitation patterns and can thus be employed for models in case other data are not available.
Second, on a regional scale, three regions in north China, east China, and southwest China were analyzed. For such large regions in situ data are usually not available. For all three regions the SWI strongly correlates with precipitation data. In north China the SWI furthermore enables one to visualize the effect of spring snowmelt, which could not be derived from precipitation data. The SWI data furthermore clearly depict extremes in water availability, such as the droughts in 1992 and floods in 1993 and 1998 in east China, droughts in 1997 and floods in 1999 in southwest China, and extreme droughts in 1997 in North China.
At the third—countrywide—scale precipitation as well as SWI patterns similarly reflect the annual movement of the Chinese rain belt. Large-scale phenomena such as the spring rain season south of the Yangtze River valley and the preflood season in south China, as well as the rainy season in north and northeast China, are clearly represented in the data.
For all three scales correlation coefficients between SWI, precipitation, and (at local scale) in situ soil moisture were good. Furthermore, anomalies (extreme wet and dry conditions) of SWI and precipitation showed similar trends in almost all cases; thus, the SWI is a suitable index to monitor water availability and water stress situations at different scales. Also, the SWI has not been studied at such temporal and spatial detail in China. Especially in countries like China, where the availability of good data is often still limited, the constantly extended and easily available SWI time series is a valuable asset. However, further analysis on SWI data at different locations need to be performed because of the vast territory of China. Meanwhile, SWI-retrieved soil moisture will be used as a driving force to simulate climate and climate changes in parts of China with regional climate models. For the future, it is furthermore planned to extend studies up to 2008.
Acknowledgments
This work was supported by the National Key Program for Developing Basic Sciences (Grant 2006CB400500) and China Postdoctoral Science Foundation (Grant 20060400492). We thank the reviewers for their numerous valuable comments to improve the manuscript.
REFERENCES
Aubert, D., Loumagne C. , and Oudin L. , 2003: Sequential assimilation of soil moisture and streamflow data in a conceptual rainfall runoff model. J. Hydrol., 280 , 145–161.
Bartalis, Z., Scipal K. , and Wagner W. , 2005: Soil moisture products from C-band scatterometers: From ERS-1/2 to METOP. Proc. ENVISAT and ERS Symp., Salzburg, Austria, European Space Agency, ESA SP-572, 1417–1423. [Available online at http://earth.esa.int/salzburg04/.].
Basist, A., Grody N. C. , Peterson T. C. , and Williams C. N. , 1998: Using the Special Sensor Microwave Imager to monitor land surface temperature, wetness, and snowcover. J. Appl. Meteor., 37 , 888–911.
Betts, A. K., and Ball J. H. , 1998: FIFE surface climate and site-average dataset 1987–89. J. Atmos. Sci., 55 , 1091–1108.
Ceballos, A., Martinéz-Fernández J. , Santos F. , and Alonso P. , 2002: Soil water behavior of sandy soils under semi-arid conditions in the Duero Basin (Spain). J. Arid Environ., 51 , 501–519.
Ceballos, A., Scipal K. , Wagner W. , and Martínez-Fernández J. , 2005: Validation of ERS scatterometer-derived soil moisture data in the central part of the Duero Basin, Spain. Hydrol. Processes, 19 , 1549–1566.
Dirmeyer, P. A., 2003: The role of the land surface background state in climate predictability. J. Hydrometeor., 4 , 599–610.
Dirmeyer, P. A., and Brubaker K. L. , 1999: Contrasting evaporative moisture sources during the drought of 1988 and the flood of 1993. J. Geophys. Res., 104 , D16. 19383–19398.
Dirmeyer, P. A., Guo Z. C. , and Gao X. , 2004: Comparison, validation, and transferability of eight multiyear global soil wetness products. J. Hydrometeor., 5 , 1011–1033.
Drusch, M., Wood E. F. , Gao H. , and Thiele A. , 2004: Soil moisture retrieval during the Southern Great Plains Hydrologic Experiment 1999: A comparison between experimental remote sensing data and operational products. Water Resour. Res., 40 .W02504, doi:10.1029/2003WR002441.
Engman, E. T., and Chauhan N. , 1995: Status of microwave soil moisture measurements with remote sensing. Remote Sens. Environ., 51 , 189–198.
Entin, J. K., Robock A. , Vinnikov K. Y. , Hollonger S. E. , Liu S. , and Namkhai A. , 2000: Temporal and spatial scales of observed soil moisture variations in the extratropics. J. Geophys. Res., 105 , D9. 11865–11877.
Fennessy, M. J., and Shukla J. , 1999: Impact of initial soil wetness on seasonal atmospheric prediction. J. Climate, 12 , 3167–3180.
Frison, P. L., and Mougin E. , 1996: Monitoring global vegetation dynamics with ERS-1 wind scatterometer data. Int. J. Remote Sens., 17 , 3201–3218.
Fuchs, T., Schneider U. , and Rudolph B. , 2007: Global precipitation analysis products of the GPCC. Global Precipitation Climatology Centre (GPCC), DWD. 1–10. [Available online at http://dwd.de.].
Hollinger, S. E., and Isard S. A. , 1994: A soil moisture climatology of Illinois. J. Climate, 7 , 822–833.
Kim, G., and Barros A. P. , 2002a: Space–time characterization of soil moisture from passive microwave remotely sensed imagery and ancillary data. Remote Sens. Environ., 81 , 393–403.
Kim, G., and Barros A. P. , 2002b: Downscaling of remotely sensed soil moisture with a modified fractal interpolation method using contraction mapping and ancillary data. Remote Sens. Environ., 83 , 400–413.
Koster, R. D., and Suarez M. J. , 2001: Soil moisture memory in climate models. J. Hydrometeor., 2 , 558–570.
Koster, R. D., and Suarez M. J. , 2004: Suggestions in the observational record of land–atmosphere feedback operating at seasonal time scales. J. Hydrometeor., 5 , 567–572.
Koster, R. D., Suarez M. J. , and Heiser M. , 2000: Variance and predictability of precipitation at seasonal-to-interannual timescales. J. Hydrometeor., 1 , 26–46.
Koster, R. D., Suarez M. J. , Higgins R. W. , and Van den Dool H. M. , 2003: Observational evidence that soil moisture variations affect precipitation. Geophys. Res. Lett., 30 .1241, doi:10.1029/2002GL016571.
Koster, R. D., and Coauthors, 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305 , 1138–1140.
Kustas, W. P., Hatfiled J. L. , and Prueger J. H. , 2005: The soil moisture–atmosphere coupling experiment (SMACEX): Background, hydrometeorological conditions, and preliminary findings. J. Hydrometeor., 6 , 791–804.
Liu, Y., and Avissar R. , 1999: A study of persistence in the land– atmosphere system using a general circulation model and observations. J. Climate, 12 , 2139–2153.
Magagi, R. D., and Kerr Y. H. , 1997: Retrieval of soil moisture and vegetation characteristics by use of ERS-1 wind scatterometer over arid and semi-arid areas. J. Hydrol., 188/189 , 361–384.
Mitchell, K. E., and Coauthors, 2004: The multi-institution North American Land Data Assimilation System (NLDAS): Utilizing multiple GCIP products and partners in a continental distributed hydrological modeling system. J. Geophys. Res., 109 .D07S90, doi:10.1029/2003JD003823.
Pellarin, T., Wigneron J-P. , Calvet J-C. , and Waldteufel P. , 2003: Global soil moisture retrieval from a synthetic L-band brightness temperature data set. J. Geophys. Res., 108 .4364, doi:10.1029/2002JD003086.
Robock, A., Vinnikov K. Y. , Srinivasan G. , Entin J. K. , Hollinger S. E. , Speranskaya N. A. , Liu S. , and Namkhai A. , 2000: The Global Soil Moisture Data Bank. Bull. Amer. Meteor. Soc., 81 , 1281–1299.
Salvucci, G. D., Saleem J. A. , and Kaufmann R. , 2002: Investigating soil moisture feedbacks on precipitation with tests of Granger causality. Adv. Water Resour., 25 , 1305–1312.
Scipal, K., 2002: Global observation of soil moisture with the ERS Scatterometer. Ph.D thesis, Institute of Photogrammetry and Remote Sensing, Vienna University of Technology, 128 pp.
Scipal, K., Scheffler C. , and Wagner W. , 2005: Soil moisture-runoff relation at the catchment scale as observed with coarse resolution microwave remote sensing. Hydrol. Earth Syst. Sci., 9 , 173–183.
Vachaud, G., Passerat de Silans A. , Balabanis P. , and Vauclin M. , 1985: Temporal stability of spatially measured soil water probability density functions. Soil Sci. Soc., 45 , 1040–1048.
Vinnikov, K. Y., Robock A. , Qiu S. , Entin J. K. , Owe M. , Choudhsury B. J. , Hollinger S. E. , and Njoku E. G. , 1999: Satellite remote sensing of soil moisture in Illinois, United States. J. Geophys. Res., 105 , 4145–4168.
Wagner, W., 1998: Soil moisture retrieval from ERS scatterometer data. Ph.D. thesis, Vienna University of Technology, 101 pp.
Wagner, W., Lemoine G. , Borgeaud M. , and Rott H. , 1999a: A study of vegetation cover effects on ERS scatterometer data. IEEE Trans. Geosci. Remote Sens., 37 , 938–948.
Wagner, W., Lemoine G. , and Rott H. , 1999b: A method for estimating soil moisture from ERS scatterometer and soil data. Remote Sens. Environ., 70 , 191–207.
Wagner, W., Scipal K. , Pathe C. , Gerten D. , Lucht W. , and Rudolf B. , 2003: Evaluation of the agreement between the first global remotely sensed soil moisture data with model and precipitation data. J. Geophys. Res., 108 .4611, doi:10.1029/2003JD003663.
Wagner, W., Blöschl G. , Pampaloni P. , Calvet J-C. , Bizzarri B. , Wigneron J-P. , and Kerr Y. , 2007a: Operational readiness of microwave remote sensing of soil moisture for hydrologic applications. Nordic Hydrol., 38 , 1–20.
Wagner, W., Naeimi V. , Scipal K. , Dejeu R. , and Martinez-Fernandez J. , 2007b: Soil moisture from operational meteorological satellites. Hydrogeol. J., 15 , 121–131. doi:10.1007/s10040-0060104-6.
Weaver, C. P., and Avissar R. , 2001: Atmospheric disturbances caused by human modification of the landscape. Bull. Amer. Meteor. Soc., 82 , 269–281.
Western, A., Grayson R. , and Blöschl G. , 2002: Scaling of soil moisture: A hydrologic perspective. Annu. Rev. Earth Planet. Sci., 30 , 149–180.
Wu, W., Geller M. A. , and Dickinson R. E. , 2001: The response of soil moisture to long-term variability of precipitation. J. Hydrometeor., 3 , 604–613.
Zhao, D. M., Su B. K. , and Zhao M. , 2006: Soil moisture retrieved from satellite images and its application to heavy rainfall simulation in eastern China. Adv. Atmos. Sci., 26 , 299–316.
Zhu, Q. G., Lin J. R. , Shou Z. W. , and Tang D. S. , 2000: The Principle and Method of Synoptic Meteorology. (in Chinese). China Meteorological Press, 649 pp.
Time series of comparisons for SWI (100 cm; %), in situ soil moisture at different layers (10, 20, and 50 cm; %), and precipitation (column; mm day−1) covering 32.9°–34.88°N, 112.5°–116°E: (a) area averaged and (b) the anomaly for the years 1992–2000.
Citation: Journal of Hydrometeorology 9, 3; 10.1175/2007JHM965.1
Comparisons of (a) annual and (b)–(d) interannual [(b) average, (c) anomaly, (d) JJA anomaly] variations between soil water (SWI and in situ soil moisture; %) and precipitation (column; mm day−1) in Fig. 1.
Citation: Journal of Hydrometeorology 9, 3; 10.1175/2007JHM965.1
Time series of comparisons between area-averaged SWI (solid; %) and precipitation (column; mm day−1) in (a) EC, (b) NC, and (c) SWC for the years 1992–2000.
Citation: Journal of Hydrometeorology 9, 3; 10.1175/2007JHM965.1
Same as in Fig. 3, but for the anomalies.
Citation: Journal of Hydrometeorology 9, 3; 10.1175/2007JHM965.1
Comparisons of interannual variations between SWI (solid) and precipitation (column) anomaly in Fig. 4: (a)–(c) whole year, (d)–(f) JJA; (a), (d) EC, (b), (e) NC, (c), (f) SWC.
Citation: Journal of Hydrometeorology 9, 3; 10.1175/2007JHM965.1
Comparisons of annual variations between area-averaged SWI (solid) and precipitation (column) in Fig. 3: (a) EC, (b) NC, and (c) SWC.
Citation: Journal of Hydrometeorology 9, 3; 10.1175/2007JHM965.1
(a)–(l) Annual variations of monthly averaged precipitation for the years 1992–2000, January–December (mm day−1).
Citation: Journal of Hydrometeorology 9, 3; 10.1175/2007JHM965.1
Same as in Fig. 7, but for SWI (%).
Citation: Journal of Hydrometeorology 9, 3; 10.1175/2007JHM965.1
Correlation coefficients (R) between area-averaged SWI, precipitation and their anomalies, averaged precipitation (mm day−1), standard deviation (SD) for the precipitation (SDP; mm day−1), and the ratio of SDSWI and SDP in EC, SWC, and NC.
