1. Introduction
Hydrological variables such as soil moisture, streamflow, and evapotranspiration are important variables for water resources management. Hydrological and land surface models can simulate the spatial distribution and temporal variation of these hydrological variables and predict extreme events (floods and drought). These models require meteorological data and static land surface data as input and use these data to calculate the kinetic and thermodynamic processes. Therefore, there are significant uncertainties in hydrological models, including meteorological inputs, model framework, vegetation conditions (Zhang et al. 2001), and soil types (Xia et al. 2015). Land cover, soil, and topography are three primary watershed properties that contribute to hydrological variability through rainfall–runoff response and erosion processing (Baker and Miller 2013; Yu et al. 2013). Regional hydrological models are sensitive to land-use and soil maps with varying accuracy, especially in simulations of runoff.
Land use can impact regional and local climate, hydrological processes, and water resources significantly (Sertel et al. 2010). The definition of land use determines the albedo, root depth, and roughness length, which impact evapotranspiration, momentum fluxes, and heat fluxes (Wei and Zhang 2010; Sy et al. 2017). Soil moisture, infiltration rate, and evapotranspiration could affect the hydrology of a region (Lee and Berbery 2012; Yachongtou et al. 2019). Ghaffari et al. (2010) showed that when land use changed from grassland to bare land, the runoff of the Iranian Zanjanrood watershed increased. Xu et al. (2016) found that increasing the forest in the Yinghe basin could increase the watershed evapotranspiration and reduce runoff, while increasing the cultivated land could reduce evapotranspiration and increase runoff. It was found that the runoff would reduce after switching grassland to woodland in the Qilian Mountains (Tian et al. 2016). Zhang et al. (2017) indicated that the surface runoff and streamflow in the Loess Plateau region have increased through urban expansion, farmland decrease, and water use change for irrigation. The land-use data presented by various land-use maps are different, and these differences in land-use types can influence the performance of the model simulation. Most hydrological models use the Moderate-Resolution Imaging Spectroradiometer (MODIS) land-use map, but its accuracy is not very good in China (Gong 2009). CNLUCC (China Land Use Land Cover Remote Sensing Monitoring Dataset) is currently a high-precision land-use map covering the whole of China, which has played an important role in national land resource surveys, hydrology, and ecological research (Xu et al. 2018). Therefore, CNLUCC is used in this research to test the impacts of land-use maps on hydrological processes.
The soil-type map is an essential input for hydrological models. The models use these soil types and their physical properties such as soil hydraulic and thermal property parameters to simulate the soil hydrothermal processes. Soil-type maps impact hydrological simulations (An et al. 2018; Lehmann et al. 2018). Osborne et al. (2004) found that soil types can influence the partitioning between surface and subsurface runoff. De Lannoy et al. (2014) compiled a new soil-type map using the HWSD version 1.21 (Harmonized World Soil Database) and the STATSGO2 (the State Soil Geographic) data for use in large-scale land surface models. And they found that the new soil-type map could reduce the bias of simulated soil moisture versus in situ and satellite data. Zheng and Yang (2016) evaluated the effects of FAO (Food and Agriculture Organization) and GSDE (Global Soil Dataset for Earth System Science) soil-type maps on regional land water circulation simulations with the Noah-MP land surface model and showed that flood intensity was greatly affected by the soil-type map. They found that sandier soil types result in lower soil moisture, lower evapotranspiration, and higher subsurface runoff. Livneh et al. (2015) compared FAO and STASGO2 soil-type maps and found that the performance of the more spatially detailed map was better in reproducing hydrologic variability and extreme events. And STATSGO2 led to the greater separation between fast and slow runoff responses. Dy and Fung (2016) used the GSDE map to update the WRF default soil-type map and found that the performance of predicting near-surface temperature and relative humidity was improved. They found that the soil map exerts a strong effect on the volumetric soil moisture content over a long period of time. Kearney and Maino (2018) added the new generation of a soil-type map (HWSD1.21 and STATSGO2) to the climate model GEOS-5 and found that the ability to simulate soil moisture at fine spatial and temporal resolution at the continental scale has improved. Lu et al. (2019) applied a new soil-type map and revised hydrologic soil parameters in the Noah land surface model, and they found that these changes reduced heat flux and enhanced latent heat flux from surface to the atmosphere and improved the model forecast skill in temperature and humidity over North China. The default soil-type map in WRF-Hydro is FAO, but the resolution of this map is relatively coarse, therefore a higher-resolution soil-type map GSDE is used in this research to understand the model sensitive to the soil-type map.
The analysis of the impact of land-use maps on streamflow targets the annual streamflow (Guo et al. 2008; Zhang et al. 2017), but does not include its impact on the monthly and daily streamflow. The above studies focused on individual sensitivity of hydrological models on either land-use or soil-type map. And the combined influence of soil type and land use is less investigated. This paper aims to evaluate the impacts of land-use and soil-type maps on hydrological processes at the watershed scale by applying a distributed hydrological model WRF-Hydro for the Xijiang River basin. Two soil-type maps and two land-use maps are used in this research. Soil moisture, soil temperature, evapotranspiration, surface runoff, underground runoff, heat flux, and absorbed shortwave radiation are used to quantify the effects of land-use and soil-type maps. And their sensitivities to land-use/soil-type maps are analyzed. Besides, the performance of these underlying surface maps is evaluated by comparing the observed and simulated streamflow and soil moisture. It has been found that the spatial distribution of soil moisture in different reanalysis and satellite products varies greatly (Zeng et al. 2016), and for reanalysis products, the underlying parameters (such as soil-type maps) may influence the accuracy of the products. Therefore, this study aims to inform the selection of underlying maps for reanalysis products and models.
2. Data and method
a. Study area
The study region is the Xijiang River basin, the largest watershed of the Pearl River basin. It contributes to the economic development of South China and meets most of the water supply and demand of the Pearl River Delta. The basin has a tropical and subtropical climate, hot and humid, with annual mean temperature ranging from 19° to 22°C and annual mean precipitation ranging from 1200 to 2200 mm. The precipitation and streamflow of this basin exhibit seasonal variability due to the monsoon rainfall. From April to September is the flood season in the Xijiang basin with the annual rainfall and streamflow accounting for 65% and 75%, respectively. The river originates in the Maxiong Mountain, travels through Yunnan, Guizhou, Guangxi, and Guangdong Provinces, and finally flows into the South China Sea. It is 2214 km long with an area of 34 600 km2. The topography of this basin is shown in Fig. 1.
Location and topography (m) of the Xijiang River basin above the Gaoyao station, where the triangles represent the hydrological stations, the gray points represent the automatic soil moisture station, the pink stars are the agrometeorological stations, and the blue line represents the river.
Citation: Journal of Hydrometeorology 24, 7; 10.1175/JHM-D-22-0095.1
b. Datasets
WRF-Hydro default soil-type map (abb. FAO) is produced by the United States Department of Agriculture (USDA) in 1991, dividing the soil type into 12 basic types based on the content of each composition. The U.S. region has a spatial resolution of 30″, but the rest is 5′. This map uses the 1:5 million soil distribution map in China, and the soil profile data are derived from 60 profiles in the World Emission List Project. Consequently, the correctness and accuracy of the data in China are questioned. Dai et al. (2013) established a soil composition map at 30″ resolution based on the original observation data in China, and Wei et al. (2013) fused soil texture datasets from different regions and countries to establish a global soil texture map (GSDE). GSDE soil-type map was established based on a 1:1 million Chinese soil distribution map and an original database of nearly 9000 soil profiles, and implemented strict quality control in the integration and inspection of profile data. It converted soil particle size data from Chinese and international systems to American systems and used the connection method and distance connection method to form the new soil properties map (Wei et al. 2013). This soil-type map contains 8 layers of different soil depths. The top soil types required to drive WRF are produced from the first 5 layers, and the bottom soil types required for WRF are made from layers 6 to 8. The percentages of sand, silt, and clay are calculated by weighted averaging, with soil layer thickness used as the weight, and the soil is then classified into soil categories in accordance with the USDA 16-class soil classification system (Dy and Fung 2016).
The default land-use map in WRF-Hydro is MODIS produced by Boston University (Friedl et al. 2002, 2010). Land use is divided into 20 categories according to IGBP (International Geosphere-Biosphere Program). The accuracy of this land-use map in China is lower than the global average accuracy (Gong 2009) and is not up to date, which is one of the most important factors restricting the accuracy of numerical model simulation. CNLUCC is currently China’s most accurate land-use remote sensing monitoring map (Xu et al. 2018), and it has been used in the national land resources survey, hydrology, and ecological research. The spatial distributions of land-use maps are shown in Fig. 2. From the figure, it can be seen that the differences between these two maps are large.
The spatial distribution of soil-type (FAO and GSDE) and land-use (MODIS and CNLUCC) maps, and the differences between these maps. SaCL = sandy clay loam, and SiCL = silty clay loam.
Citation: Journal of Hydrometeorology 24, 7; 10.1175/JHM-D-22-0095.1
The digital elevation model (DEM) and the China Meteorological Forcing Dataset (CMFD) are used to drive the offline WRF-Hydro model whose initial condition is provided by the ERA5. DEM data are from the Hydrological Data and Maps Based on Shuttle Elevation Derivatives at Multiple Scales (HydroSHEDS; Lehner et al. 2008), which were developed on the high-resolution elevation data of the Shuttle Radar Topography Mission (SRTM). These data have been widely used in hydrological and landslide studies (Wang et al. 2016). In this research, the resolution of 15″ (about 500 m) is used. The CMFD developed by the Institute of Tibetan Plateau Research has a reasonable consistency with the observation data (Chen et al. 2011; He et al. 2020). It has been applied in hydrological and land models, as well as in assessing the impacts of climate change in China (Tan et al. 2021). The 3-hourly CMFD data starts from 1979 with a spatial resolution of 0.1° × 0.1°. The ERA5 reanalysis data developed by the European Centre for Medium-Range Weather Forecasts (ECMWF) represent the meteorological variability and land variability reasonably (Hersbach et al. 2020), and it has been widely used in hydrology, climate, and synoptic studies (Tarek et al. 2020; Probst and Mauser 2022). The 6-hourly ERA5 data with 30-km spatial resolution are used.
The observed monthly soil moisture in 1992–2013 (Wang and Shi 2019), observed daily soil moisture in 2017/18 (A. Wang 2022, personal communication), and observed daily streamflow in 2005–13 (Pearl River Water Resources Commission of the Ministry of Water Resources) are used to evaluate the model performance. These three observed datasets are respectively sourced from agrometeorological stations (http://data.cma.cn/), automatic soil moisture stations, and the Water Yearbook of China. The spatial distribution of these stations is shown in Fig. 1.
c. Model
WRF-Hydro is a distributed hydrological model,, which infuses routing modules (base flow, saturated subsurface flow, overland flow, and channel) with the existing Noah-MP land surface model (Gochis et al. 2018). The resolution between the Noah-LSM and routing modules is flexible, which allows for higher resolution in the routing processes than in the land surface model. The model has been used for a wide range of projects, including flood prediction, water resources forecasting, and land–atmosphere coupling studies (Kerandi et al. 2018). WRF-Hydro is both a hydrological model (so-called offline model) and a coupled framework of hydrological and atmospheric models. The offline version 5 is used in the research. The spatial resolution of the Noah-LSM is 5000 m, while it is 500 m in routing processes. All routing modules are activated, and the parameterization scheme options used for NoahMP are shown in Table 1.
Parameterization scheme options used in this study.
The meteorological forcing data are from CMFD. The CMFD does not have U and V components, and the wind speed values are assigned to U and all zero values are assigned to V (Gochis et al. 2018). The initial condition file is generated by WRF, and the high-resolution condition file is generated by ArcGIS (Sampson and Gochis 2018). Its generation details can be found in Gochis et al. (2018). The model is run over a period of time from January 1979 to December 2013, where 1979–91 is used as model spinup. The vegetation fraction data are derived from the 5-yr AVHRR NDVI climatology (1985–1990) at a resolution of 0.144° (Gutman and Ignatov 1998).
To further assess the impact of land-use and soil-type maps on the simulation performance of WRF-Hydro, four cases are performed. The cases’ names and differences are as follows: 1) FAO-MODIS, using the FAO soil-type map and MODIS land-use map; 2) FAO-CNLUCC, using the FAO soil-type map and CNLUCC land-use map; 3) GSDE-MODIS, using the GSDE soil-type map and MODIS land-use map; 4) GSDE-CNLUCC, using the GSDE soil-type map and CNLUCC land-use map.
d. Statistical metrics
3. Results
a. Comparison of land-use and soil-type maps
The top and bottom soil type spatial distribution of FAO and GSDE soil-type maps in the Xijiang River basin, and the differences between these maps are shown in Fig. 2. From the figure, it can be seen that compared with FAO the resolution of GSDE is significantly increased, and the differences between these two maps are large. For the top layer, clay is replaced by clay loam in the west, clay loam replaces silty clay in the central, and sandy clay loam is replaced by clay in the west and southwest. The differences between these two maps in the bottom layer exist in the southwest, central, and north of the basin, and clay replaces clay loam in the central. The soil type is sandy loam in FAO in the southwest and north, while it is loam, silty clay loam, clay loam, and clay in GSDE.
As is shown in Fig. 2, the differences in land-use maps are considerably large for the whole basin. Compared with the MODIS, the proportions of evergreen needleleaf forest, evergreen broadleaf forest, closed shrublands and grasslands increase, and the proportions of mixed forests and woody savannas decrease in CNLUCC. Overall, the forest type increased by 2.5%, and the urban agriculture land-use type decreased by 15% in CNLUCC.
b. Influence on hydrological variables
In this paper, soil-type and land-use influence on hydrological variables is tested through the study of hydraulic, thermal, and radiation variables. The spatial distributions of annual mean soil moisture of FAO-MODIS and spatial distributions of the relative change of other cases are shown in Fig. 3. The relative change means the difference between FAO-MODIS and another case divided by FAO-MODIS multiplied by 100. From Figs. 3e–h, it can be seen that the influence of the land-use map on top soil moisture is larger than on the deep soil moisture. When forests change to other types or other types change to forests, the soil moisture change is larger. When grasslands are replaced by croplands or when one forest type is replaced by another forest type, soil moisture changes little. Land use affects soil moisture at different levels, and this may relate to the root depth of vegetation. The depth of forests is four layers, while it is three layers for savannas, grassland, and cropland. The vegetation absorbs the water from the soil due to the transpiration of vegetation. Therefore, soil moisture increases when forests change to other types for the fourth layer.
(a)–(d) Annual mean soil moisture (m3 m−3) at different depths in FAO-MODIS, and (e)–(p) annual mean soil moisture at different depths in other cases relative change to the annual mean soil moisture of FAO-MODIS. The dots represent statistically significant differences in monthly values at the 95% confidence interval.
Citation: Journal of Hydrometeorology 24, 7; 10.1175/JHM-D-22-0095.1
Similar to previous studies (Xia et al. 2015), the modeled soil moisture is strongly dependent on soil types. As shown in Figs. 3i–l, when the soil-type map changes from FAO to GSDE, soil moisture changes greatly. The soil moisture in the southwest and north of the basin increases greatly, and it decreases in the west of the basin. In FAO, the soil type is sandy clay loam in the southwest and north of the basin, but it is clay loam, silty clay, and clay in GSDE. The differences between these soil properties are large (Figs. S1–S9 in the online supplemental material). Sandy clay loam has more sand than other soil types. When the content of sand increases, the soil has smaller pore spaces, the water storage of soil is poorer (Chen and Dudhia 2001), and this leads to lower soil moisture. Therefore, soil moisture in the southwest and north of the basin increases when the GSDE map is used. In the west part, when FAO changes to GSDE, clay changes to clay loam. And the soil moisture decreases when the soil-type map changes to GSDE. Clay has larger porosity, field capacity, and wilting point than clay loam (Fig. S1), and the soil moisture is higher. The sand of clay loam is higher than that of silty clay. When clay loam changes to silty clay (in the central part), the soil moisture is higher. Porosity affects the water storage capacity of the soil, so the distribution of soil moisture is consistent with the change of soil texture, and is greatly affected by soil type.
The impact of land-use maps on soil moisture is much smaller than the impact of soil-type maps on soil moisture. The influence of soil-type maps on seasonal soil moisture simulation is tested (Figs. S10–S13). And the spatial distribution of seasonal (spring, summer, autumn, winter) soil moisture simulation is similar to the annual mean soil moisture, that is the soil moisture in the southwest and north of the basin increases largely, and it decreases in the west of the basin when the soil-type map changes from FAO to GSDE.
The spatial distributions of annual surface runoff, underground runoff, soil temperature, and evapotranspiration are shown in Fig. 4. From this figure, it can be seen that when the land-use map changes from MODIS to CNLUCC, surface runoff increases in the west part, while it decreases in the corner of the northwest and southeast. It relates to the change in forests. If the land use changes to the forest, the surface runoff decreases. And if land use changes from forest to other types, the surface runoff increases. Similar conditions exist in some other regions (Xu et al. 2016; Tian et al. 2016). The influence of different types of forests is small. Land-use maps also influence the soil temperature, when cropland changes to the forest, the soil temperature decreases. Evapotranspiration is sensitive to land-use maps. When MODIS changes to CNLUCC, savannas change to forests in the central part, and evapotranspiration increases, which relates to plant transpiration. Forests, which have deeper roots than other vegetation, absorb soil water and transport more water to the atmosphere through transpiration, reducing surface runoff and increasing evapotranspiration. Besides, the green vegetation fraction (GVF) also influences evapotranspiration. The GVF is 0.7 for the evergreen needleleaf forest, while it is 0.95 for the evergreen broadleaf forest. And the evapotranspiration of the evergreen broadleaf forest is larger than that of the evergreen needleleaf forest (south corner of the basin).
As in Fig. 3, but for annual mean surface runoff (mm month−1), underground runoff (mm month−1), soil temperature (K), and evapotranspiration (mm month−1).
Citation: Journal of Hydrometeorology 24, 7; 10.1175/JHM-D-22-0095.1
From Fig. 4, it can be seen that the surface runoff is larger for clay loam, silty clay, and clay than for sandy clay loam. And it is larger for silty clay and clay than for clay loam. When sand clay loam changes to silty clay or clay loam changes to silty clay, the underground runoff and soil temperature decrease. When clay changes to clay loam, the underground runoff, and soil temperature increase. Similar to the findings of Osborne et al. (2004), the soil type can influence the partitioning between surface and subsurface runoff. For evapotranspiration, when sand clay loam changes to silty clay, the evapotranspiration increases (southwest and east part of the basin).
The impact of land-use maps on surface runoff is much smaller than the impact of soil-type maps, but it has a greater impact on evapotranspiration. Soil-type maps influence the simulation of soil moisture, and the change in soil moisture influences the surface runoff. And the attributes of soil type also influence the simulation of soil moisture. Therefore, the soil-type maps have a greater impact on the simulation of surface runoff. Evapotranspiration, including evaporation and transpiration, and land use largely influence the transpiration. Therefore, the land-use map has a greater impact on evapotranspiration. The influence of land-use and soil-type maps on soil temperature is small. The changes in soil temperature between these cases are between −1 and 1 K, and the relative changes are low.
Figure 5 shows the distribution of sensible heat flux in FAO-MODIS and the relative changes in the other three cases. Sensible heat flux in this context refers to the transfer of thermal energy from the land surface to the lowest atmospheric layer, driven by a gradient in temperature between the two. When the land-use maps change from MODIS to CNLUCC, the sensible heat increases for the east part and decreases for the central part of the basin. When the soil-type map changes, the sensible heat increases in the upper part of the basin, but decreases in the southwest and east of the basin. It is mainly due to the difference in sand and gravel content which affects the change in sensible heat by causing the difference in soil moisture. The sensible heat is more sensitive to the land-use map than the soil-type map. Evapotranspiration can influence the atmosphere temperature: when evapotranspiration increases the atmosphere temperature decreases. And the changes in temperature influence the sensible heat. Evapotranspiration is sensitive to land-use maps, therefore, sensible heat is sensitive to land-use maps. Latent heat is the heat absorbed or emitted by water during phase transformation. When the land-use map changes, the spatial distribution of relative change of latent heat flux is similar to that of evapotranspiration. If evapotranspiration increases, more water is transformed into the atmosphere, and leads to higher latent heat flux. The difference in soil properties causes changes in soil water content, which directly causes latent heat difference. From the figure, it can be seen that the land-use maps’ influence on surface heat fluxes (sensible heat flux and latent heat flux) is larger than the influence of soil-type maps. The change in surface heat fluxes is related to the change of evapotranspiration, and the land-use map has a great impact on evapotranspiration. Therefore, the surface heat fluxes are sensitive to the land-use map. The effects on the radiation variables are analyzed by surface albedo and absorbed shortwave (total absorbed shortwave radiation). When the land-use map changes to CNLUCC, surface albedo reduces and absorbed shortwave increases for most areas of the basin, and soil-type maps have less impact on surface albedo and absorbed shortwave.
As in Fig. 3, but for the annual mean sensible heat flux (W m−2), latent heat flux (W m−2), albedo, and absorbed shortwave (W m−2).
Citation: Journal of Hydrometeorology 24, 7; 10.1175/JHM-D-22-0095.1
Changes in the annual average of soil moisture, soil temperature, surface runoff, underground runoff, evapotranspiration, sensible heat flux, latent heat flux, albedo, and absorbed shortwave of the basin are presented in Fig. 6. From this figure, it can be seen that the soil-type and land-use maps have little influence on the temporal change of hydrological variables. When the land-use map changes to CNLUCC, the annual mean surface runoff, underground runoff, and albedo get lower, while soil moisture, sensible heat, and absorbed shortwave increase. When the soil-type map changes to GSDE, the soil moisture increases largely which is related to the soil properties. At the same time, the surface runoff increases, and underground runoff decreases. The evapotranspiration and latent heat flux change little in the whole basin, as the increased parts and decreased parts of this basin are close.
Annual average (a) soil moisture (m3 m−3), (b) soil temperature (K), (c) surface runoff (mm month−1), (d) underground runoff (mm month−1), (e) evapotranspiration (mm month−1), (f) sensible heat flux (W m−2), (g) latent heat flux (W m−2), (h) albedo, and (i) absorbed shortwave (W m−2) time series of four cases in the average of Xijiang River basin.
Citation: Journal of Hydrometeorology 24, 7; 10.1175/JHM-D-22-0095.1
c. Simulation verification
The performance of the soil-type and land-use maps is verified at soil moisture observation stations and hydrological stations. The time series of four soil moisture stations are shown in Fig. 7. The depth of the soil moisture is 0–0.1 m, and the depth of the simulated soil moisture is the first layer (0–0.1 m). From the figure, it can be seen that the soil-type and land-use maps can influence the simulation of soil moisture. Figure 7a shows the station located at 26.22°N, 104.08°E, and the top soil type and land use are different on different maps, but the bottom soil types are similar. From the results of this station, it can be seen that when using the CNLUCC land-use map the model performance is improved, and the simulation effect of using the GSDE soil type is also greatly improved. When the FAO map changes to the GSDE map, clay changes to clay loam at this station; the porosity and field capacity decrease so that the soil moisture decreases. The results of GSDE-CNLUCC are the best compared with other cases. The location of Fig. 7b station is 23.93°N, 108.10°E, where the top soil type is the same in different soil-type maps, while the bottom soil type and land use are different. From the results, it can be seen that the differences among these four cases are very small. When MODIS changes to CNLUCC, cropland changes to shrubland. The transpiration is larger in shrubland, and this type gets lower soil moisture. At the Cangwu station, the soil type is similar in GSDE and FAO soil maps, while the land use is different in the land-use maps (Fig. 7d). It can be seen that the differences among these four cases are very small. Figure 7c shows the performance of simulated soil moisture at the station with constant land use but soil type changing with map changes. When the FAO map changes to the GSDE map, sandy clay loam changes to clay loam. And the higher porosity and field capacity in clay loam type leads to more soil moisture. It can be seen that when modifying the soil-type map from FAO to GSDE, the performance improves after 2012.
The annual mean soil moisture (m3 m−3) time series of the four cases and observation at (a) Xuanwei station, (b) Duan station, (c) Tiandeng station, and (d) Cangwu station (stations circled by the box in Fig. 1) in 1992–2013, and in some periods the observed soil moisture is missing.
Citation: Journal of Hydrometeorology 24, 7; 10.1175/JHM-D-22-0095.1
Figure 8 shows boxplots depicting the differences between each case and 15 agrometeorological stations. From the figure, it can be seen that the differences in CC in these four cases are small. The RMSE has changed largely among the four cases, however, the interquartile range of the RMSE of GSDE-MODIS and GSDE-CNLUCC are smaller than FAO-MODIS. The median RMSE of GSDE-MODIS and GSDE-CNLUCC is lower than in other cases. These facts indicate that both land-use and soil-type maps have an influence on soil moisture simulation. After using the GSDE soil-type map and CNLUCC land-use maps, the model performance in simulating soil moisture is slightly increased.
The boxplot of monthly CC, and RMSE (m3 m−3) of 15 agrometeorological soil moisture stations for four cases, where red lines are the median values.
Citation: Journal of Hydrometeorology 24, 7; 10.1175/JHM-D-22-0095.1
The spatial distribution of daily CC and RMSE of automatic soil moisture stations in FAO-MODIS in 2017–18 and the spatial distribution of the differences of CC and RMSE between other cases and FAO-MODIS are shown in Fig. 9. In Fig. 9, redder colors indicate higher performance. From this figure, it can be seen that when MODIS changes to CNLUCC, the CCs increase in most areas of the basin, and the change RMSEs is slight. When FAO changes to GSDE, the CCs increase in the central parts of the basin, and RMSEs increase in the upper and central parts of the basin. When both land-use and soil-type maps change, the performance increase is larger than with only one map change at some stations, but there are some stations that have poor performance compared with only one map change. In all, when the CNLUCC land-use map and GSDE soil-type map are used, the simulation performance of soil moisture in some areas of the basin is improved, and the performance decreases in some areas. The performance of simulated soil moisture decreased in certain areas of the basin when using the GSDE soil-type map and CNLUCC land-use map. This may be related to the attributes of the soil types. The attributes (Figs. S1–S9) of the soil types used in this research are the default attributes (or parameters), and the attributes could influence the results. Lu et al. (2019) revised the default attributes in the Noah land surface model, and they found that these changes improved the model performance. To illustrate, two stations are selected and the time series of the simulated and observed soil moisture are shown (Figs. S15 and S16). When the FAO soil-type map changes to GSDE, clay changes to clay loam (Fig. S16a) in the upper part of the basin and sandy clay loam changes to clay loam (Fig. S16b) in the southwest of the basin. However, the default attributes of clay loam may not be accurate enough, leading to biases in simulated soil moisture. Figure S16b indicates that when using the FAO soil-type map, the simulated soil moisture is lower than the observed soil moisture, while using the GSDE soil-type map shows the opposite. However, due to the attributes of clay loam, the simulated soil moisture is too high. Thus, the influence of the attributes requires further research. The spatial distribution of observed and simulated soil moisture is shown in Fig. S14. From the figure it can be seen that when the FAO soil-type map is used the soil moisture overestimated/overestimated in the upper/lower part of the basin, and the soil moisture is lower/higher when GSDE soil-type map used. The change of soil moisture (in the central part) is small among these four cases. The differences in soil moisture in these cases are related to the attributes of the soil types. The differences in attributes (Figs. S1–S9) of some soil types are small, and the differences are larger for some soil types. And these influence the soil moisture simulation.
The spatial distribution of (a) CC and (e) RMSE (m3 m−3) of soil moisture stations in FAO-MODIS, and the spatial distribution of the differences of (b)–(d) CC and (f)–(h) RMSE (m3 m−3) between those three cases and FAO-MODIS.
Citation: Journal of Hydrometeorology 24, 7; 10.1175/JHM-D-22-0095.1
Table 2 gives the results of three statistical metrics at nine stations of four cases of simulated daily streamflow. When different soil-type and land-use maps are applied, the soil types and land use in the watershed of the nine stations are changed with the change of the maps. From the CC and NSE, it can be seen that when GSDE soil-type and CNLUCC land-use maps are used, the model performance of simulating streamflow is improved at most stations. When the soil-type map changes from FAO to GSDE and the land-use map changes from MODIS to CNLUCC, soil types and land uses change accordingly. The attributes of soil types and land uses can influence the model performance of simulating streamflow. Soil type and land use have a large impact on the BIAS of simulated streamflow. For example, in the basin where the Gaoche station is located, the soil type has changed from clay loam to clay, and the land use has changed from cropland to grassland. When the soil type changes from clay to silty clay, the soil moisture, and evapotranspiration decrease, which may lead to more streamflow. For the Shihuichang station, the performance of simulating streamflow gets worse in terms of NSE when land use changes, but the performance improves when the GSDE soil type is used. The Gaoyao station is located at the outlet of the basin, and from the results it can be seen that the performance of simulating streamflow is improved when using GSDE soil-type and CNLUCC land-use maps. It can be seen from the results of each station that land-use and soil-type maps have an influence on streamflow simulation, and the influence is larger in reducing BIAS and smaller in improving CC. The difference among these cases in simulating monthly streamflow variation is small, and both cases can simulate the monthly change of the streamflow. For the Wuzhou and Gaoyao stations, the model overestimates the streamflow in the flood periods. When FAO soil type and MODIS land-use maps are used, the BIASs are 55.5% and 57.1% for these two stations. After using the GSDE soil-type map and CNLUCC land-use map, the BIASs reduce to 52.9% and 55.2%. The performance of simulating flood periods improves at the Wuzhou and Gaoyao stations after using the GSDE soil-type map and CNLUCC land-use map. The simulated streamflow is influenced by many variables (such as soil moisture and evapotranspiration). The performance of simulating these variables is different among the four cases. When these variables are simulated better, the performance of simulating the streamflow is better.
Quantitative analysis results of model performance on daily streamflow simulation of four cases.
4. Conclusions and discussion
Land-use and soil-type maps are two underlying conditions that influence the uncertainties of hydrological models, and this study focuses on the impacts and uncertainties coming from land-use and soil-type maps. Using FAO and GSDE soil-type maps and MODIS and CNLUCC land-use maps, we compared these different sources of underlying maps and assessed their influence on hydrological simulation. Compared with FAO soil-type map, the top soil types of GSDE are different for almost all the basins. When FAO changes to GSDE, the clay mainly changes to clay loam, the sandy clay loam mainly changes to silty clay and clay loam, and the clay loam mainly changes to silty clay. The soil properties, such as porosity, field capacity, saturation soil conductivity, and wilting point soil moisture, have large differences among these soil types. These properties have a large influence on model simulation. The differences between the two land-use maps are large. When MODIS changes to CNLUCC, the proportions of evergreen needleleaf forest, evergreen broadleaf forest, closed shrublands, and grasslands increase, and the proportions of mixed forests and woody savannas decrease.
The simulated soil moisture and surface runoff are more sensitive to the soil-type map compared with the land-use type. The soil-type map of the Xijiang River basin also has an effect on heat flux simulation, but the sensitivity of radiation variables to the soil-type map is lower compared with the land-use type. The simulated evapotranspiration and heat flux are more sensitive to land-use maps compared with the soil-type change in the Xijiang River basin, whereas simulated soil moisture does not present high sensitivity. When GSDE soil-type and CNLUCC maps are used, the performance of simulated soil moisture and streamflow is better in most areas.
The accuracy of land-use and soil-type maps should be considered when applying the hydrological and land surface models in practice. Hydrological simulations are sensitive to land-use and soil-type maps. Soil-type and land-use maps can influence land surface properties such as soil parameters, albedo, and roughness length, and these variations can influence the climate system by altering the water and energy fluxes between land and atmosphere. Little change in land surface characteristics could lead to significant climatic responses (Zhao and Wu 2017). From the contrast of the impact of different land-use and soil-type maps, it can be seen that land-use and soil-type maps can impact the partition of energy flux, soil moisture redistribution, and runoff generation. These maps influence different models and different parameterizations of physical processes may differ. This research is based on the offline model, and the influence of soil-type and land-use maps on the coupled model needs further research. Liu and Tian (2010) showed that China had experienced substantial land-use change due to human activities and environmental changes over the past 300 years. Therefore, changes in soil-type and land-use maps should be considered when predicting future climate change. Consequently, before each project, researchers need to select accurate soil-type and land-use maps.
Acknowledgments.
This work was supported by the National Nature Science Foundation of China (Grants 42088101 and 42175170) and the National Major Science and Technology Infrastructure Project “Earth System Numerical Simulation Device (EarthLab) Project.” We appreciate Prof. Aihui Wang for providing the observed soil moisture data and the team of Prof. Kun Yang for providing the CMFD dataset.
Data availability statement.
The dataset on which this paper is based is too large to be retained or publicly archived with available resources. Documentation and methods used to support this study are available from http://data.tpdc.ac.cn/en/data/8028b944-daaa-4511-8769-965612652c49/.
REFERENCES
An, N., C. S. Tang, S. K. Xu, X. P. Gong, B. Shi, and H. I. Inyang, 2018: Effects of soil characteristics on moisture evaporation. Eng. Geol., 239, 126–135, https://doi.org/10.1016/j.enggeo.2018.03.028.
Baker, T. J., and S. E. Miller, 2013: Using the Soil and Water Assessment Tool (SWAT) to assess land use impact on water resources in an east African watershed. J. Hydrol., 486, 100–111, https://doi.org/10.1016/j.jhydrol.2013.01.041.
Chen, F., and J. Dudhia, 2001: Coupling an advanced land surface hydrology model with the Penn State NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129, 569–585, https://doi.org/10.1175/1520-0493(2001)129<0569:CAALSH>2.0.CO;2.
Chen, Y. Y., K. Yang, J. He, J. Qin, J. C. Shi, J. Y. Du, and Q. He, 2011: Improving land surface temperature modeling for dry land of China. J. Geophys. Res., 116, D20104, https://doi.org/10.1029/2011JD015921.
Dai, Y. J., S. G. Wei, Q. Y. Duan, B. Y. Liu, S. H. Fu, and G. Y. Niu, 2013: Development of a China dataset of soil hydraulic parameters using pedotransfer functions for land surface modeling. J. Hydrometeor., 14, 869–887, https://doi.org/10.1175/JHM-D-12-0149.1.
De Lannoy, G. J. M., R. D. Koster, R. H. Reichle, S. P. P. Mahanama, and Q. Liu, 2014: An updated treatment of soil texture and associate hydraulic properties in a global land modeling system. J. Adv. Model. Earth Syst., 6, 957–979, https://doi.org/10.1002/2014MS000330.
Dy, C. Y., and J. C. H. Fung, 2016: Updated global soil map for the Weather Research and Forecasting model and soil moisture initialization for the Noah land surface model. J. Geophys. Res. Atmos., 121, 8777–8800, https://doi.org/10.1002/2015JD024558.
Friedl, M. A., and Coauthors, 2002: Global land cover mapping from MODIS: Algorithms and early results. Remote Sens. Environ., 83, 287–302, https://doi.org/10.1016/S0034-4257(02)00078-0.
Friedl, M. A., D. Sulla-Menashe, B. Tan, A. Schneider, N. Ramankutty, A. Sibley, and X. M. Huang, 2010: MODIS collection 5 global land cover: Algorithm refinements and characterization of new datasets. Remote Sens. Environ., 114, 168–182, https://doi.org/10.1016/j.rse.2009.08.016.
Ghaffari, G., S. Keesstra, J. Ghodousi, and H. Ahmadi, 2010: SWAT-simulated hydrological impact of land-use change in the Zanjanrood basin, Northwest Iran. Hydrol. Processes, 24, 892–903, https://doi.org/10.1002/hyp.7530.
Gochis, D. J., and Coauthors, 2018: WRF-Hydro V5 technical description. NCAR Tech. Note, 107 pp., https://ral.ucar.edu/sites/default/files/public/WRF-HydroV5TechnicalDescription_0.pdf.
Gong, P., 2009: Accuracy assessment of global land cover datasets based on global field stations (in Chinese). Prog. Nat. Sci., 19, 754–759.
Guo, H., Q. Hu, and T. Jiang, 2008: Annual and seasonal streamflow responses to climate and land-cover changes in the Poyang Lake basin, China. J. Hydrol., 355, 106–122, https://doi.org/10.1016/j.jhydrol.2008.03.020.
Gutman, G., and A. Ignatov, 1998: The derivation of the green vegetation fraction from NOAA/AVHRR data for use in numerical weather prediction models. Int. J. Remote Sens., 19, 1533–1543, https://doi.org/10.1080/014311698215333.
He, J., K. Yang, W. Tang, H. Lu, J. Qin, Y. Chen, and X. Li, 2020: The first high-resolution meteorological forcing dataset for land process studies over China. Sci. Data, 7, 25, https://doi.org/10.1038/s41597-020-0369-y.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Jordan, R., 1991: A one-dimensional temperature model for a snow cover: Technical documentation for SNTERERM.89. Special Rep. 91-16, Cold Region Research and Engineers Laboratory, U.S. Army Corps of Engineers, 61 pp.
Kearney, M. R., and J. L. Maino, 2018: Can next-generation soil data products improve soil moisture modelling at the continental scale? An assessment using a new microclimate package for the R programming environment. J. Hydrol., 561, 662–673, https://doi.org/10.1016/j.jhydrol.2018.04.040.
Kerandi, N., J. Arnault, P. Laux, S. Wagner, J. Kitheka, and H. Kunstmann, 2018: Joint atmospheric-terrestrial water balances for East Africa: A WRF-Hydro case study for the upper Tana River basin. Theor. Appl. Climatol., 131, 1337–1355, https://doi.org/10.1007/s00704-017-2050-8.
Lee, S. J., and E. H. Berbery, 2012: Land cover change effects on the climate of the La Plata Basin. J. Hydrometeor., 13, 84–102, https://doi.org/10.1175/JHM-D-11-021.1.
Lehmann, P., O. Merlin, and P. Gentine, 2018: Soil texture effects on surface resistance to bare-soil evaporation. Geophys. Res. Lett., 45, 10 398–10 405, https://doi.org/10.1029/2018GL078803.
Lehner, B., K. Verdin, and A. Jarvis, 2008: New global hydrography derived from spaceborne elevation data. Eos, 89, 93–94, https://doi.org/10.1029/2008EO100001.
Liu, M. L., and H. Q. Tian, 2010: China’s land cover and land use change from 1700 to 2005: Estimations from high-resolution satellite data and historical archives. Global Biogeochem. Cycles, 24, GB3003, https://doi.org/10.1029/2009GB003687.
Livneh, B., R. Kumar, and L. Samaniego, 2015: Influence of soil textural properties on hydrologic fluxes in the Mississippi River basin. Hydrol. Processes, 29, 4638–4655, https://doi.org/10.1002/hyp.10601.
Lu, B., W. Wang, Y. Yang, J. Q. Zhong, and M. Chen, 2019: Updated soil map and soil hydrologic parameters for WRF and their influences over North China during the warm season (in Chinese). Acta Meteor. Sin., 77, 1028–1040, https://doi.org/10.11676/qxxb2019.066.
Niu, G., and Z. Yang, 2006: Effects of frozen soil on snowmelt runoff and soil water storage at a continental scale. J. Hydrometeor., 7, 937–952, https://doi.org/10.1175/JHM538.1.
Osborne, T. M., D. M. Lawrence, J. M. Slingo, A. J. Challinor, and T. R. Wheeler, 2004: Influence of vegetation on the local climate and hydrology in the tropics: Sensitivity to soil parameters. Climate Dyn., 23, 45–61, https://doi.org/10.1007/s00382-004-0421-1.
Probst, E., and W. Mauser, 2022: Evaluation of ERA5 and WFDE5 forcing data for hydrological modelling and the impact of bias correction with regional climatologies: A case study in the Danube River Basin. J. Hydrol. Reg. Stud., 40, 101023, https://doi.org/10.1016/j.ejrh.2022.101023.
Sakaguchi, K., and X. Zeng, 2009: Effects of soil wetness, plant litter, and under-canopy atmospheric stability on ground evaporation in the Community Land Model (CLM3.5). J. Geophys. Res., 114, D01107, https://doi.org/10.1029/2008JD010834.
Sampson, K., and D. J. Gochis, 2018: WRF Hydro GIS pre-processing tools, version 2.2 documentation. NCAR, 39 pp.
Sertel, E., A. Robock, and C. Ormeci, 2010: Impacts of land cover data quality on regional climate simulations. Int. J. Climatol., 30, 1942–1953, https://doi.org/10.1002/joc.2036.
Sy, S., N. Noblet-Ducoudre, B. Quesada, I. Sy, A. M. Dieye, A. T. Gaye, and B. Sultan, 2017: Land-surface characteristics and climate in West Africa: Models’ biases and impacts of historical anthropogenically-induced deforestation. Sustainability, 9, 1917, https://doi.org/10.3390/su9101917.
Tan, Q., Z. Xu, Y. Zhao, J. Liu, C. Ban, X. Liu, and J. Wang, 2021: Applicability of China meteorological forcing dataset to the Nianchu River basin. J. Beijing Norm. Univ. Nat. Sci., 57, 372–379, https://doi.org/10.12202/j.0476-0301.2020261.
Tarek, M., F. P. Brissette, and R. Arsenault, 2020: Evaluation of the ERA5 reanalysis as a potential reference dataset for hydrological modelling over North America. Hydrol. Earth Syst. Sci., 24, 2527–2544, https://doi.org/10.5194/hess-24-2527-2020.
Tian, F., Y. H. Lu, B. J. Fu, T. H. Yang, G. Y. Qiu, C. F. Zang, and L. Zhang, 2016: Effects of ecological engineering on water balance under two different vegetation scenarios in the Qilian Mountain, northwestern China. J. Hydrol. Reg. Stud., 5, 324–335, https://doi.org/10.1016/j.ejrh.2015.11.015.
Wang, A., and X. Shi, 2019: A multilayer soil moisture dataset based on the gravimetric method in China and its characteristics. J. Hydrometeor., 20, 1721–1736, https://doi.org/10.1175/JHM-D-19-0035.1.
Wang, J., H. J. Wang, and Y. Hong, 2016: A high-resolution flood forecasting and monitoring system for China using satellite remote sensing data (in Chinese). Chin. Sci. Bull., 61, 518–528, https://doi.org/10.1360/N972015-00489.
Wei, S. G., and Coauthors, 2013: A China data set of soil properties for land surface modeling (with supplementary materials). J. Adv. Model. Earth Syst., 5, 212–224, https://doi.org/10.1002/jame.20026.
Wei, X., and M. Zhang, 2010: Quantifying streamflow change caused by forest disturbance at a large spatial scale: A single watershed study. Water Resour. Res., 46, W12525, https://doi.org/10.1029/2010WR009250.
Xia, Y., M. B. Ek, Y. Wu, T. Ford, and S. M. Quiring, 2015: Comparison of NLDAS-2 simulated and NASMD observed daily soil moisture. Part II: Impact of soil texture classification and vegetation type mismatches. J. Hydrometeor., 16, 1981–2000, https://doi.org/10.1175/JHM-D-14-0097.1.
Xu, J., J. Hua, and Z. Y. Yan, 2016: Study on effect of land cover change on runoff simulation result in a headstream watershed (in Chinese). J. Nanjing Univ. Nat. Sci., 52, 142–148.
Xu, X. L., J. Y. Liu, S. W. Zhang, R. D. Li, C. Z. Yan, and S. X. Wu, 2018: China’s multi-period land use land cover remote sensing monitoring dataset (CNLUCC). Data Registration and Publishing System of the Resource and Environmental Science Data Center of the Chinese Academy of Sciences, accessed 1 June 2022, https://www.resdc.cn/DOI/doi.aspx?DOIid=54.
Yachongtou, B., J. Liang, G. G. Zeng, X. X. Li, M. Q. Lei, and Q. Liu, 2019: Synergistic effect of land use and climate change on Liuyang River Basin runoff assessed by SWAT model (in Chinese). J. Water Res. Water Eng., 30, 88–94.
Yu, E. T., J. Q. Sun, and W. L. Xiang, 2013: High-resolution hindcast of record-breaking rainfall in Beijing and impact of topography. Atmos. Ocean. Sci. Lett., 6, 253–258, https://doi.org/10.1080/16742834.2013.11447090.
Zeng, Y., Z. Su, R. van der Velde, L. Wang, K. Xu, X. Wang, and J. Wen, 2016: Blending satellite observed, model simulated, and in situ measured soil moisture over Tibetan Plateau. Remote Sens., 8, 268, https://doi.org/10.3390/rs8030268.
Zhang, L., W. R. Dawes, and G. R. Walker, 2001: Response of mean annual evapotranspiration to vegetation changes at catchment scale. Water Resour. Res., 37, 701–708, https://doi.org/10.1029/2000WR900325.
Zhang, L., R. Karthikeyan, Z. Bai, and R. Srinivasan, 2017: Analysis of streamflow responses to climate variability and land use change in the Loess Plateau region of China. Catena, 154, 1–11, https://doi.org/10.1016/j.catena.2017.02.012.
Zhao, D., and J. Wu, 2017: The impact of land use and land cover changes on East Asian summer monsoon precipitation using the WRF-mosaic approach. Atmos. Sci. Lett., 18, 450–457, https://doi.org/10.1002/asl.788.
Zheng, H., and Z. L. Yang, 2016: Effects of soil-type datasets on regional terrestrial water cycle simulations under different climatic regimes. J. Geophys. Res. Atmos., 121, 14 387–14 402, https://doi.org/10.1002/2016JD025187.
