Application of snowmelt runoff model (SRM) in mountainous watersheds: A review

Shalamu ABUDU*, Chun-liang CUI Muattar SAYDI James Phillip KING

1. Xinjiang Water Resources Research Institute, Urumqi 830049, P. R. China

2. Civil Engineering Department, New Mexico State University, Las Cruces, NM 88001, USA

Application of snowmelt runoff model (SRM) in mountainous watersheds: A review

Shalamu ABUDU*1,2, Chun-liang CUI1, Muattar SAYDI1, James Phillip KING2

1. Xinjiang Water Resources Research Institute, Urumqi 830049, P. R. China

2. Civil Engineering Department, New Mexico State University, Las Cruces, NM 88001, USA

The snowmelt runoff model (SRM) has been widely used in simulation and forecast of streamflow in snow-dominated mountainous basins around the world. This paper presents an overall review of worldwide applications of SRM in mountainous watersheds, particularly in data-sparse watersheds of northwestern China. Issues related to proper selection of input climate variables and parameters, and determination of the snow cover area (SCA) using remote sensing data in snowmelt runoff modeling are discussed through extensive review of literature. Preliminary applications of SRM in northwestern China have shown that the model accuracies are relatively acceptable although most of the watersheds lack measured hydro-meteorological data. Future research could explore the feasibility of modeling snowmelt runoff in data-sparse mountainous watersheds in northwestern China by utilizing snow and glacier cover remote sensing data, geographic information system (GIS) tools, field measurements, and innovative ways of model parameterization.

snowmelt runoff model; temperature; precipitation; snow cover area; remote sensing; northwestern China

1 Introduction

Snow accumulation and ablation processes dominate the surface water cycle over much of the global land area poleward of about 40° latitude, particularly in the continental interiors (Adam et al. 2009). Hence, understanding snow processes is significant to water resources management in these regions. It is known that the effect of global climate change on hydrologic systems, especially on mountain snow and glacier melt, can modify the timing and amount of runoff in mountainous watersheds. Therefore, accurate streamflow simulation and forecast is of great importance to water resources management and planning (Abudu et al. 2010), and can provide a firm basis for forecasts of water resources availability while minimizing the risk and loss from floods caused by rapid snow and glacier melt. Severalsnowmelt forecasting models have been developed to suit specific needs and hydrologic conditions. They are data-intensive and/or complicated to handle. Very few models can handle varied hydrologic conditions in general (Tekeli et al. 2005). The snowmelt runoff model (SRM), which uses snow cover information and meteorological data as input variables, has been most widely used both in simulation and forecast. SRM was developed by Martinec (1975) and used in small European basins. To date the model has been used by various agencies, institutes, and universities in more than 100 basins of various sizes ranging from 0.76 to 917 444 km2and at various elevations (Martinec et al. 2008).

SRM is designed to simulate and forecast daily streamflow in mountainous basins where snowmelt is a major runoff factor. Most recently, it has also been used to evaluate the effect of a changing climate on seasonal snow cover and runoff (Rango et al. 2008). More than 80% of applications of SRM have been performed by independent users. SRM also successfully underwent tests by the World Meteorological Organization (WMO) with regard to runoff simulation (WMO 1986). SRM calculates the daily discharge of a basin as follows (Martinec et al. 2008):

whereQis the average daily discharge (m3·s-1);cis the runoff coefficient expressing the losses as a ratio of runoff to precipitation, withcSreferring to snowmelt andcRto rain;ais the degree-day factor (cm·℃-1·d-1), which indicates the snowmelt depth resulting from one degree-day;Tis the number of degree-days (℃·d); ΔTis the adjustment by temperature lapse rate when extrapolating the temperature from the station to the average elevation of the basin or zone (℃·d);Sis the ratio of the snow cover area (SCA) to the total area;Pis the precipitation contributing to runoff (cm), which is determined by the preselected threshold temperatureTCRIT: if precipitation is determined to be new snow, it is kept in storage over the hitherto snow-free area until melting conditions occur;Ais the area of the basin or zone (km2);kis the recession coefficient indicating the decline of discharge in a period without snowmelt or rainfall:k=Qm+1Qm(mandm+1 are the sequence of days during a true recession flow period); andnis the sequence of days during the discharge computation period. Eq. (1) is written for a time lag between the daily temperature cycle and the corresponding discharge cycle of 18 hours. In this case, the number of degree-days measured on thenth day corresponds to the discharge on the (n+1)th day. Various lag times can be introduced by a subroutine. The conversion from cm·km2·d-1to m3·s-1(conversion from runoff depth to discharge) is.

Runoff computations by SRM appear to be relatively easily understood. There are three model input variables: temperature, precipitation, and SCA, and eight model parameters: temperature lapse rate, runoff coefficient (for rain and snow), degree-day factor, recession coefficient, critical temperature, rainfall-contributing area, and lag time. All the variables andparameters have to be specified for model input on a daily step.

SRM calculates the daily average streamflow on the (n+1)th day by the addition of snowmelt and precipitation contributing to runoff and discharge on the preceding day. Generally, a seasonal or yearly runoff volume can be estimated with a known discharge in a watershed, as long as the model input variables are provided and parameters are determined. In addition to these, the area-elevation curve of the basin is also required. Model parameters can be well estimated and optimized if the given basin or climate characteristics, such as soil conditions, vegetation cover, antecedent precipitation, and runoff data, are available. SRM can be used to simulate the daily streamflow of a snowmelt season, in a year, or in a sequence of years, to provide short-term and seasonal runoff forecasts, and to evaluate the potential effect of climate change on the seasonal snow cover and runoff (Martinec et al. 2008; Prasad and Roy 2005).

SRM generally has good performance and less data requirements, and is computationally simple (Tekeli et al. 2005). Moreover, it has undergone rigorous testing for various conditions, sizes of basins, and geographic locations (Engman et al. 1989). In SRM, the discharge is estimated by temperature conditions in the form of degree-days. The degree-days are used to determine the melt rate of the snowpack in the area of the basin covered by snow as observed from satellites. The precipitation input is used to add any rainfall runoff to the snowmelt component (Dey et al. 1989). The various applications of the snowmelt runoff model mainly differ in the selection of data sources, determination and calculation of parameters, and adaptability of the model structure for different study basins. Leavesley (1989) and Xu and Singh (1998) discussed the existing problems of SRM and stated that the more universal problems are generally associated with data constraints, whereas the more unique problems are associated with model formulation and the climatic and physiographic characteristics of a region. With the application of satellite data to monitoring snow accumulations, the model has been refined over a period of time for its applications to large basins, some in inaccessible terrain, for simulation periods of only a few days in the entire year (Dey et al. 1989). Because the quality of the input data is directly related to the model accuracy and efficiency, the calculation and determination of variables and parameters is most important to SRM. Hence, according to previous studies, the SRM research worldwide has been mainly concentrated on the acquisition of variables, optimization of parameters, and proper analyses of the hydrological and physical characteristics of a basin. This paper discusses some of the worldwide SRM research, mainly from the point of view of the validation of climatic factors, parameters, and mapping of SCA. Specific examples are introduced with emphases placed on the SRM research in northwestern China.

2 Selection of model inputs and parameters

SRM uses air temperature, precipitation, and SCA during the snowmelt period as essential input data. Usually, temperature and precipitation data can be obtained frommeteorological observations within and/or outside drainage basins, and SCA can be extracted from remote sensing satellite data. However, the methods for calibration of the SRM model variables vary largely, depending on the data availability in a specific study watershed. In addition, proper selection of model parameters such as the temperature lapse rate, runoff coefficient, degree-day factor, recession coefficient, and critical temperature also affects the model performance. The temperature lapse rate is of great significance. Temperature measured at one or more stations has to be extrapolated to the mean elevation of each elevation zone. Error in the lapse rate can result in incorrect computation of degree-days that distort the whole process of snow melting. The recession coefficient (k) is also a very important parameter as it determines the proportion of daily meltwater production that appears immediately in the runoff, particularly in a large basin with a large elevation range.

2.1 Temperature and precipitation

SRM is a degree-day model that calculates runoff by converting the degree-days above the critical temperature into melt depths by the degree-day factor. As a principal variable of SRM, temperature values affect the melting process of snow cover, and together with the critical temperature, determine a precipitation event such as rain or snow. Ideally, temperature and precipitation are measured within the basin and at each mean elevation for each zone. Seldom is the case in real basins that climatic data have to be extrapolated from one or more stations, some of which may not be in the basin (Engman et al. 1989). Good-quality climatic data observations ensure successful SRM performance while less favorable conditions for measuring temperature and precipitation in larger basins, in which only the data from the normal hydro-meteorological network are available, lead the SRM performance to deteriorate with the decreasing quality of basic data (Rango and Martinec 1981).

Previous studies have utilized different methods to obtain climatic factors. Dey et al. (1989) presented the application of SRM in the Western Himalayas. They used ambient air temperature values from one climate station outside of the basin and degree-day values determined by extrapolating the ambient air temperature to the mean elevation according to certain temperature lapse rates. No information about the source of precipitation was given. The simulation results were somewhat less accurate because of sparse climate observations in such a large basin and poor quality of input data set. Rango and Martinec (1981) stated that the most accurate simulation is provided by SRM when the climate station is located at the hypsometric mean elevation of the basin or at the elevation zone and inside the basin boundaries. These guidelines are hard to meet, however, because the ground observation is limited in most high and remote mountainous basins, in which few, if any, hydro-meteorological stations are available. Jesko et al. (1999) used temperature from three different stations at different elevation levels and obtained zonewide temperature by extrapolating the station altitude to mean zone altitude with a global temperature lapse rate. Acertain precipitation lapse rate was also used to avoid the problem of underestimating the precipitation in the study basin of Massa-Blatten, in the Swiss Alps. Monthly average temperature and precipitation disaggregated from monthly total values, though not recommended and inadequate for day-to-day runoff computation, were also used in a SRM test in large Himalayan basins with acceptable results (Seidel et al. 2000). Hydro-meteorological data of air temperature and precipitation for daily and ten-day average periods were used in SRM performance in the Beas Basin, India (Prasad and Roy 2005). Temperature and precipitation were also obtained by means of a certain interpolation method in cases where there were no climatic stations in the study basin but some around it, such as the SRM simulation of the Elaho River Basin, in British Columbia by Nitin (2004), who used an elevationally detrended spline interpolation function with the values measured at stations near the study basin. Daily temperature and precipitation data in the latest study on SRM in the Moroccan High Atlas Mountains were obtained throughout zones by certain lapse rates that had been calculated or observed according to the time series of recorded data within the respective basins (Boudhar et al. 2009).

It is a common way to obtain the data from meteorological observations within and/or outside drainage basins of accessing climatic factors’ value, but it must be ascertained that they are representative of the total study area. Richard and Gratton (2001) pointed out that air temperature is critical to SRM, especially for large basins, and extrapolation of air temperature using a certain lapse rate tested only at one or very limited stations might not be reasonable. More accurate and reliable temperature lapse rates can be specified if a number of weather stations at representative elevations in a given river basin are accessible or available (Li and Williams 2008). Richard and Gratton (2001) presented the synthetic regional weather station and individual weather station methods to estimate possible impacts of different air temperature acquisition on the snowmelt runoff modeling. Results showed that a weather station should be located within the most representative land cover of the study area to provide input data for accurate hydrological modeling. Otherwise, to obtain air temperature using the synthetic regional weather station method is more viable.

Temperature and precipitation can be acquired in various new ways. Tekeli et al. (2005) obtained climatic factors for SRM inputs in the Karasu Basin of the Euphrates River in eastern Turkey from numerical weather model forecasts of the mesoscale model version 5 (MM5). Li and Williams (2008) applied SRM to the Tarim Basin, one of the arid mountainous watersheds in northwestern China. In their research, one more new data source was used to obtain climatic factors and model parameters. They developed an enhanced temperature-index model that incorporates shortwave solar radiation and snow albedo from the degree-day factor. For the temperature-index model, temperature values came from two available long-term weather stations. Precipitation values were obtained in two different ways: First, the precipitation value from one station record was extrapolated for the mean altitudes of each elevation zones using a certain precipitation lapse rate recommended by Martinec et al. (2008). Second, the dailyprecipitation data, which was derived from multi-satellite observations with a spatial resolution of 1° by 1°, came from the Global Precipitation Climatology Project (GPCP),. However, both types of precipitation values proved to be less reliable in their study, according to the results.

2.2 Snow cover area

Being one of the three main input variables of SRM, SCA is of great significance for model accuracy. Errors in determining SCA are directly proportional to the resulting errors in the calculated snowmelt (Rango and Martinec 1981). SCA can be mapped by terrestrial observation (in a very small basin), by aircraft photography (especially during a flood emergency) and, most efficiently, by satellites (Martinec et al. 2008). The remote sensing data can provide detailed inputs to the snowmelt runoff modeling due to the highly spatial and temporal variability of snow cover information (Nagler et al. 2008). Moreover, SRM is optimized for input of remotely sensed snow cover data while many other models have not been designed for remote sensing data input. Hence, of the three kinds of SCA calculating methods, SCA obtained from satellite images is the most popular and preferred.

In remote sensing, numerous sensors have been used to map snow cover. Four of the most important and widely used sensors have included the Landsat 5 Thematic Mapper (TM), the National Oceanic and Atmospheric Administration-Advanced Very High Resolution Radiometer (NOAA-AVHRR), the European Remote Sensing-Synthetic Aperture Radar (ERS-SAR), and the Moderate Resolution Imaging Spectroradiometer (MODIS) (Poon 2004). Detection of SCA from satellites started in 1966 when NOAA began to use remote sensing to provide weekly estimates of snow cover in the Northern Hemisphere (Klein and Barnett 2003). Attention to SCA increased with the studies performed by Martinec (1975) and Rango and Martinec (1979). Rango (1983) applied remotely sensed snow cover data in snowmelt runoff modeling while Dozier (1984) demonstrated that band 5 of Landsat 5 was useful in discrimination between clouds and snow, as stated in Prasad and Roy (2005). Dey et al. (1989) obtained SCA from NOAA-4 satellite images and used SRM in the large Kabul River Basin in the Western Himalayas with acceptable results. Kumar et al. (1991) used Landsat Multispectral Scanner (MSS) data to obtain SCA, and, with other meteorological data, proved the applicability of SRM with adequate accuracy in the Bean Basin, in the Himalayas, India. Baumgartner and Apfl (1997) used NOAA-AVHRR data to analyze the snowmelt runoff process in the alpine basins in Switzerland and found that, for long-term monitoring programmers, only NOAA-AVHRR data offers an acceptable compromise concerning spatial and temporal resolution.

Cloud cover in satellite images is a major issue that contributes negatively to the extraction of daily snow maps; it remains the NOAA’s challenge in a situation of cloud cover. To overcome this potential data loss caused by cloud cover, the MODIS composite 8-day maximum snow-cover extent data are more preferable in current studies than the utilization oftraditional daily images when running SRM. Seidel et al. (2000) also used NOAA-AVHRR data in the runoff modeling for the Ganges-Brahmaputra River Basin in northern India. The best cloud-free 10-day composites were selected for each month so that the snow coverage was evaluated in terms of monthly values in the basin. Nitin (2004) selected 12 sets of MODIS 8-day snow cover images with the least amount of cloud cover to simulate SRM in the Elaho River Basin, British Columbia. The snow cover value of a particular month’s MODIS 8-day snow cover image was used for all the days of that month for SCA daily inputs. Results indicated that using the MODIS 8-day snow-cover extent data for a year-round simulation can roughly describe the general trend of seasonal snowmelt runoff, but cannot precisely interpret the sharp daily fluctuations (Nitin 2004). More intensive description of satellite images to map SCA was provided by Tekeli et al. (2005) in their study on SRM in the eastern part of Turkey. They discussed the advantages and disadvantages of different satellite images on the basis of research conducted by Klein and Barnett (2003) and pointed out that NOAA-AVHRR data needs to be processed to classify snow from the other features, whereas the snow cover maps of MODIS contain snow cover data ready to be used for SRM. They also confirmed that MODIS’s geolocation accuracy is much better than that of AVHRR. In their study a comparison between MODIS daily snow cover data (MOD10A1) andin situsnow depth measurements were made. In the comparison, in order to lessen the effect of cloud cover, the possible cloud-free MODIS daily snow cover data acquired few days before or after the specific day of ground data measurement were chosen. Good agreement between them indicated that MODIS snow mapping correctly captures the presence of snow on the ground surface. Finally, the MODIS 8-day snow cover products (MOD10A2) were used to map SCA for SRM by minimizing the cloud cover and maximizing the snow cover (Tekeli et al. 2005).

Another new method was implemented by Nagler et al. (2008) when forecasting the daily runoff in the drainage basin ?tztal in the Austrian Alps. In this study they mapped SCA using satellite images of MODIS and radar images of Envisat ASAR (Advanced Synthetic Aperture Radar). Snow maps from ASAR and MODIS revealed systematic differences, and the results suggest that ASAR underestimated snow-cover extent in comparison with optical imagery. Accordingly, SCA obtained from satellite images was used to forecast short-term runoff in this study. The latest research has found that snow-cover maps for SRM inputs can also be obtained from further simulation by a snow water equivalent (SWE) model with a certain empirical relationship between SCA and SWE (Boudhar et al. 2009). They examined SRM in the Moroccan High Atlas Mountains by deriving SCA via both satellite images of SPOT-VEGETATION sensors and SWE modulation with spatially interpolated meteorological data. Results show that both estimates were satisfactory in streamflow simulation at seasonal time scales. However, in the case of SCA tested by SWE based on bulk precipitation or unreliable temperature, it is prone to failure in describing the real SRM patterns of the watershed, which necessitates more accurate streamflow predictions that have to be obtained from the remotely sensed data.

2.3 Initial estimation of model parameters

SRM parameters including the temperature lapse rate, runoff coefficient, degree-day factor, recession coefficient, critical temperature, rainfall-contributing area, and time lag are also of vital importance to the model performance. When these parameters cannot be calibrated by historical data, they can be measured or estimated by hydrological judgment taking into account the basin characteristics, physical laws, theoretical relations, and empirical regression relations (Martinec et al. 2008). More generally, as argued by Rango and Martinec (1981), variables and parameters that were either measured or determined beforehand are preferred to optimization techniques aiming at the best possible agreement between the simulated and measured runoff values (Rango and Martinec 1981). This has been the case in many previous and subsequent applications of SRM (Dey et al. 1989; Kumar et al. 1991; Nitin 2004; Tekeli et al. 2005; Boudhar et al. 2009).

Usually, the parameters of the temperature lapse rate, degree-day factor, and recession coefficient can be evaluated more accurately by on-site measurements, weather records, and hydrological data analysis with certain empirical formulas. The estimation of the runoff coefficient, critical temperature, rainfall-contributing area, and lag time is more directly derived from basin characteristics such as geographical location, vegetation cover, topography, soil condition, evapotranspiration, physical laws, and complex interactions between them. Model parameters are unique for a given basin according to its geographic location and hydro-meteorological characteristics. In many studies, model parameters were chosen and modified based on the SRM manual description by the user to adjust the simulation results in view of the deterministic model according to the basin characteristics, and hydrological and physical conditions (Seidel et al. 1998; Richard and Gratton 2001; Jesko et al. 1999). However, it is suggested that SRM parameters be evaluated based on more hydrological judgments in a specific basin, rather than mechanical calibration or optimization (Martinec et al. 2008).

In conclusion, SRM has less data requirements in comparison with other hydrological models. However, accurate performance of SRM relies heavily on the input data quality. In most mountainous areas hydro-meteorological stations at higher elevation zones are sparse, and available observation records can hardly ensure the accuracy of the computation of variables and parameters. Through a review of previous research on the model inputs and parameterization, we have observed that: (1) Of the model variables, temperature is the major factor dominating the process of snowmelt. It is fundamental to obtain accurate and representative temperature values for modeling. Long time-series weather records at different hydro-meteorological stations, if any, with a same time scale are recommended to evaluate temperature values. (2) It is a general method to extrapolate the precipitation by a certain altitude gradient in a basin with a large elevation range. Close attention should be paid to validating a proper altitude gradient because the increase of precipitation amounts with altitude does not continue indefinitely. It is usually closely related to vertical air flow, terrainfeatures, and the possibility of water vapor transport in a basin. For instance, on the southern slopes of the Tianshan Mountains the largest amount of precipitation occurs at the elevation range between 2 500 m and 3 000 m, whereas on the northern slopes, it occurs at 2 000 m to 3 000 m, and above that altitude there is no significant increment in precipitation amount with altitude (Hu 2004). (3) Of a number of satellite images, the MODIS snow cover products are the most popular and desirable for reasons such as geolocation accuracy, minimized cloud cover, and free access. One more point is that satellite images for snow extraction are accessible during intermittent periods, and certain interpolation needs to be carried out to obtain successive daily input data regarding temperature, discharge data, and seasonal trends of basin climate. The interpolation method between two periods of satellite images is also a very interesting point for further study. (4) Determination of model parameters is related to the basin size to some extent. Usually, it seems more complex in a large basin covering a variety of climatic zones than in a small basin with relatively simple features. In a small basin, some of the parameters can be evaluated on a basin-wide basis, whereas in a large basin, the parameters have to be specified for different elevation zones respectively.

3 SRM applications in northwestern China

SRM research started comparatively late in China, and most of the research were applied to northwestern China, where most snow-dominated mountainous watersheds are located. Some pilot studies have been conducted since 2000. Most of these studies have sought to simulate the snowmelt runoff of mountainous watersheds by simply using SRM (Eq. (1)). An early attempt was made by Feng et al. (2000) to simulate the daily snowmelt runoff in the Manas River Basin, in the Tianshan Mountains in Xinjiang, China. In their study, they used NOAA-AVHRR satellite images and calculated the temperature lapse rate according to previous studies in the local area, which produced preferable modeling results accordingly. Liu et al. (2007) also tested SRM in the Tashikuergan River Basin simply by calculating the daily discharge of snowmelt water. The contribution of precipitation to the daily discharge was neglected based on the assumption that precipitation can hardly contribute to the average daily discharge in this basin mainly because of the extremely dry climate. Their study included a relatively superficial investigation, both in terms of the selection of data sources and calculation of parameters. Further research is needed in order to reveal the real snowmelt runoff characteristics of this region.

Besides these studies, more specific research was conducted by Ma and Cheng (2003) in the Gongnaisi River Basin in the western Tianshan Mountains; it performed quite well compared with the other application results from all around the world. Apart from SRM, a further test was made to study the response of snow cover and snowmelt runoff to future climatic change. A comparison was conducted between snowmelt runoff hydrographs under present and future climatic conditions. In their study, selected data sources were reliable,parameters were calculated and determined reasonably, and related descriptions were comparatively distinct. Another significant achievement of their study was that SRM was proven to be the ideal method in data-scarce regions, particularly in remote and inaccessible high-mountain watersheds in arid areas. One more example of successful simulation of SRM comes from the research of Liu et al. (2006) in the Dongkemadi River Basin at the headwaters of the Yangtze River on the Tibetan Plateau. Differing from previous research in northwestern China, they simulated the effects of different divisions of basin zone areas on the hydrological process and used temperatures at different stations as the driving factor to affect the hydrological simulation. Results proved the general agreement that SRM is very sensitive to temperature. Accordingly, an appropriate zone division and representative air temperature were selected as the final simulation scenario for the Dongkemadi River Basin’s hydrologica1 process.

An important issue when applying snowmelt runoff models is the physical accessibility and the lack of hydrometeorological measurements in high-altitude mountainous watersheds. Particularly in the northwestern region of China where most snow-dominated basins are located, a lack of data is a common problem. Hence, the application of SRM in data-scarce mountainous watersheds has attracted the attention of scholars in China. A typical example was the analysis of the snowmelt runoff process in the Kaidu River Basin in the Tianshan Mountains, China by Zhang et al. (2006). The basin is the largest in area compared with all other basins that have been studied in China to date. In their study they did some extensive work to determine the climatic input variables which had great significance to the model accuracy. Specifically, two groups of coefficients for temperature and precipitation were introduced in their study and the coefficients’ values were then adjusted empirically according to simulation results by iterative experiments. For such data-scarce remote mountainous basins where data extrapolated from hydro-meteorological stations may be unreliable, this method is recommended and is of real significance to improving the modeling and forecasting accuracy. Li and Wang (2008) studied the application of SRM in the Upper Heihe River Basin where water resources are supplied mostly by snowmelt runoff and limited hydro-meteorological data are available. In their study, some existing problems regarding the variables and parameters of SRM research in China were analyzed. Results show that SRM is an advisable method to examine the snowmelt runoff patterns in arid land watersheds where human observation is sparse. However, according to their results, further intensive observation on hydro-meteorological data would be necessary to the identification of the climatic factors and day-degree factors with higher accuracy so that the snowmelt runoff process could be simulated more accurately.

Li and Williams (2008) developed an enhanced temperature-index model that incorporated shortwave solar radiation and snow albedo to the degree-day factor in the basin of the Yarkant River, which is the largest tributary of the Tarim River, in Xinjiang, China. Their researchexplored the feasibility of modeling snowmelt runoff in a data-sparse mountainous watershed by modifying existing snowmelt models to develop an enhanced temperature-index model, which uses satellite-derived snow cover data and varied degree-day factors based on shortwave solar radiation and snow albedo. The latest study in northwestern China on SRM research was the application of SRM to spring flood forecasting in the Manasi River Basin (Zhang et al. 2009). In their study, the simulation of the snowmelt process was made using both the WinSRM 1.10 and the self-developed snow-melting runoff simulation and forecast software, SRSFS 1.0. Simulation results proved to be very close to one another according to the accuracy criteria of SRM. Moreover, this study forecasted the daily runoff of the study basin in the Spring of 2004 with SRSFS 1.0 based on the China Meteorological Administration (CMA) T213 meteorological data to obtain zonal temperature and precipitation. They used the CMA T213 meteorological data innovatively for snowmelt runoff forecasting and presented a new method of meteorological data mining, which needs further examination.

Table 1 shows some representative applications of SRM in mountainous watersheds in northwestern China. As can be seen, the model accuracies are relatively acceptable although most of the basins lack measured climatic data. Compared with other snowmelt models, SRM requires limited climatic data as inputs. This feature facilitates the successful application of the model in the snow-dominated moun tainous basins of northwestern China where measured hydro-meterological data are seriously limited and/or even not available at all. However, in previous applications of SRM in the region, the validation of SRM variables and parameters relied heavily on the previous definitions without necessary local investigation and adjustments that would contribute positively to the performance of SRM. More research should be encouraged to explore the feasibility of modeling snowmelt runoff in datasparse mountainous watersheds in northwestern China, utilizing snow and glacier cover remote sensing data, geographic information system (GIS) tools, field measurements, and computer simulation.

Table 1 Representative SRM applications in northwestern China

4 Conclusions

Due to global climate change and global warming, snowmelt runoff research is considered more essential than ever before to predicting water resources availability, programing water usage and management, and designing water allocation on a sustainable and long-term basis. Improving the accuracy of snowmelt runoff forecasts diminishes the loss from floods caused by rapid snow and glacier melt. In arid lands where mountain-fed rivers are the only available water resources for the needs of the public, agricultural irrigation, hydropower, and other uses, this situation seems more urgent and necessary. Of the continental river basins in northwestern China, mountain snowmelt water is regarded as the main water source of piedmont regions where human existence and production activities are mainly distributed in the oases scattered throughout the river basins. Therefore, snowmelt water is a vital resource to the safety of people living there. Hence, accurate simulation and forecast of snowmelt runoff has become a real necessity for this region.

This paper evaluates worldwide applications of SRM, particularly in snow-dominated mountainous watersheds in northwestern China. Selection of input variables and parameterization of SRM in the previous studies are described in detail. Previous studies suggest that the temperature and precipitation are the major factors dominating the process of snowmelt. When performing the extrapolation of temperature and precipitation, close attention should be paid to validating proper lapse rate and altitude gradient. The processing of satellite images is another major issue in the snowmelt runoff modeling. A certain interpolation needs to be carried out to obtain successive daily snow cover data between acquired satellite images based on the factors such as temperature, discharge data, and the seasonal trend of a basin climate. In addition, it is prudent to select different parameters based on the basin scale. For example, the parameters can be evaluated on a basin-wide basis in a small basin, whereas in a large basin, parameters have to be specified for different elevation zones since a large basin covers a variety of climatic zones.

SRM has been applied successfully to mountainous watersheds around the world for the last several decades. With improvement of input climatic variables and parameter selection methods, SRM has a potential in forecasting streamflow and evaluating the effects of climate change on runoff in the mountainous watersheds of northwestern China, especially data-scarce watersheds in high-elevation regions. The limited number of applications of SRM in northwestern China showed that the model accuracies are relatively acceptable although most of the basins lack measured climate data. The increasing availability of remote sensing data facilitates the successful application of the model in the snow-dominated mountainous basins of northwestern China, where measured hydro-meterological data are seriously limited and/or not available at all. However, in previous applications of SRM in the region, the validation of SRM variables and parameters relied heavily on former definitions without necessary local investigation and adjustments that would contribute positively to the performance of SRM.More research should be encouraged on the feasibility of modeling snowmelt runoff in data-sparse mountainous watersheds in northwestern China where only limited hydrometeorological measurements are available, by utilizing snow and glacier cover remote sensing data, GIS tools, field measurements, and innovative ways of model parameterization.

Abudu, S., Cui, C., King, J. P., and Abudukadeer, K. 2010. Comparison of performance of statistical models in forecasting monthly streamflow of Kizil River, China.Water Science and Engineering, 3(3), 269-281. [doi:10.3882/j.issn.1674-2370.2010.03.003]

Adam, J. C., Hamlet, A. F., and Lettenmaier, D. P. 2009. Implications of global climate change for snowmelt hydrology in the twenty-first century.Hydrological Processes, 23(7), 962-972. [doi:10.1002/hyp.7201]

Baumgartner, M. F., and Apfl, G. M. 1997. Remote sensing, geographic information systems and snowmelt runoff models: An integrated approach. Baumgartner, M. F., Schultz, G. A., and Johnson, A. I., eds.,Remote Sensing and Geographic Information Systems for Design and Operation of Water Resources Systems, 73-101. Wallingford: International Association of Hydrological Sciences.

Boudhar, A., Hanich, L., Boulet, G., Duchemin, B., Berjamy, B., and Chehbouni, A. 2009. Evaluation of the snowmelt runoff model in the Moroccan High Atlas Mountains using two snow-cover estimates.Hydrological Science Journal, 54(6), 1094-1113. [doi:10.1623/hysj.54.6.1094]

Dey, B., Sharma, V. K., and Rango, A. 1989. A test of snowmelt-runoff model for a major river basin in Western Himalayas.Nordic Hydrology, 20(3), 167-178. [doi:10.2166/nh.1989.013]

Dozier, J. 1984. Snow reflectance from Landsat-4 thematic mapper.IEEE Transactions on Geosciences and Remote Sensing, GE-22(3), 323-328. [doi:10.1109/TGRS.1984.350628]

Engman, E. T., Rango, A., and Martinec, J. 1989. EXSRM, an expert system for snowmelt runoff model (SRM).New Directions for Surface Water Model (Proceedings of the Baltimore Symposium). Washington, D.C.: International Association of Hydrological Sciences.

Feng, X. Z., Li, W. J., Shi, Z. T., and Wang, L. H. 2000. Satellite snow cover monitoring and snowmelt runoff simulation of Manas River in Tianshan Region.Remote Sensing Technology and Application, 15(1), 18-21. (in Chinese)

Hu, R. J. 2004.Physical Geography of the Tianshan Mountain in China. Beijing: China Environmental Science Press. (in Chinese)

Jesko, S., Martinec, J., and Seidel, K. 1999. Distributed mapping of snow and glaciers for improved runoff modeling.Hydrological Processes, 13(12-13), 2023-2031. [doi:10.1002/(SICI)1099-1085(199909)13:12/ 13＜ 2023::AID-HYP877＞3.0.CO;2-A]

Klein, A. G., and Barnett, A. C. 2003. Validation of daily MODIS snow cover maps of the Upper Rio Grande River Basin for the 2000-2001 snow year.Remote Sensing of Environment, 86(2), 162-176. [doi:10.1016/ S0034-4257(03)00097-X]

Kumar, V. S., Haefner, H., and Seidel, K. 1991. Satellite snow cover mapping and snowmelt runoff modeling in Beas Basin.Proceedings of the Vienna Symposium on Snow, Hydrology and Forests in High Alpine Areas, 101-109. International Association of Hydrological Sciences.

Leavesley, G. H. 1989. Problems of snowmelt runoff modeling for a variety of physiographic and climatic conditions.Hydrological ScienceJournal, 34(6), 617-634. [doi:10.1080/02626668909491371]

Li, H. Y., and Wang, J. 2008. The snowmelt runoff model applied in the Upper Heihe River Basin.Journal of Glaciology and Geocryology, 30(5), 769-775. (in Chinese)

Li, X. G., and Williams, M. W. 2008. Snowmelt runoff modeling in an arid mountain watershed, Tarim Basin, China.Hydrological Processes, 22(19), 3931-3940. [doi:10.1002/hyp.7098]

Liu, J. F., Yang, J. P., Chen, R. S., and Yang, Y. 2006. The simulation of snowmelt runoff model in the Dongkemadi River Basin, headwater of the Yangtze River.Acta Geographica Sinica, 61(11), 1149-1159. (in Chinese)

Liu, W., Li, Z. L., and Li, K. B. 2007. Snowmelt runoff modeling in Tashikuergan River Basin, Xinjiang, China.Technical Supervision in Water Resources, (3), 43-46. (in Chinese)

《教育大辭典》將親職教育定義為：“對父母實施的教育，其目的是改變或提升父母的教育觀念，使父母獲得撫養(yǎng)、教育子女的知識和技能”。現(xiàn)實中，我國的親職教育體系起步晚，發(fā)展還很不完善，主要依靠官方的家長學校、社會媒體的傳播和代際經(jīng)驗的傳遞，內(nèi)容零散。缺乏科學系統(tǒng)的規(guī)劃和專業(yè)人士的推進指導，主體單一，執(zhí)行力不足，且覆蓋面受限。難以為家長提供與時俱進的育兒理念、系統(tǒng)化的育兒技能，更不能滿足新形勢下為人父母者日益迫切的個性化的教育需求。

Ma, H., and Cheng, G. D. 2003. A test of snowmelt runoff model (SRM) for the Gongnaisi River Basin in the western Tianshan Mountains, China.Chinese Science Bulletin, 48(20), 2253-2259.

Martinec, J. 1975. Snowmelt runoff model for stream flow forecasts.Nordic Hydrology, 6(3), 145-154. [doi:10.2166/nh.1975.010]

Martinec, J., Rango, A., and Roberts, R. T. 2008.Snowmelt Runoff Model (SRM) User’s Manual, 19-39. New Mexico: New Mexico State University Press.

Nagler, T., Rott, H., Malcher, P., and Müller, F. 2008. Assimilation of meteorological and remote sensing data for snowmelt runoff forecasting.Remote Sensing of Environment, 112, 1408-1420. [doi:10.1016/j.rse. 2007.07.006]

Nitin, M. V. 2004. Snowmelt runoff modeling using MODIS in Elaho River Basin, British Columbia.Environmental Informatics Archives, 2, 526-530.

Poon, S. K. M. 2004.Hydrological Modeling Using MODIS Data for Snow Covered Area in the Northern Boreal Forest of Manitoba. M. E. Dissertation. Alberta: University of Calgary.

Prasad, V. H., and Roy, P. S. 2005. Estimation of snowmelt runoff in Beas Basin, India.Geocarto International, 20(2), 41-47. [doi:10.1080/10106040508542344]

Rango, A., and Martinec, J. 1981. Accuracy of snowmelt runoff simulation.Nordic Hydrology, 12(4-5), 265-274. [doi:10.2166/nh.1981.021]

Rango, A. 1983. Application of simple snowmelt runoff model to large river basin.Proceedings of the Fifty-first Annual Western Snow Conference, 89-99. Fort Collins: Colorado State University.

Rango, A., Martinec, J., and Roberts, R. T. 2008. Relative importance of glacier contributions to water supply in a changing climate.World Resources Review, 20(3), 487-503.

Richard, C., and Gratton, D. J. 2001. The importance of the air temperature variable for the snowmelt runoff modelling using the SRM.Hydrological Processes, 15(18), 3357-3370.

Seidel, K., Ehrler, C., and Martinec, J. 1998. Effects of climate change on water resources and runoff in an alpine basin.Hydrological Processes, 12(10-11), 1659-1669. [doi:10.1002/(SICI)1099-1085(199808/09) 12:10/11＜1659::AID-HYP687＞3.0.CO;2-4]

Seidel, K., Martinec, J., and Baumgartner, M. F. 2000. Modeling runoff and impact of climate change in large Himalayan basins.Proceedings of the International Conference on Integrated Water Resources Management for Sustainable Development(ICIWRM-2000). International Association of Hydrological Sciences.

Tekeli, A. E., Akyurek, Z., ?orman, A. A., ?ensoy, A., and Sorman, A. U. 2005. Using MODIS snow cover maps in modeling snowmelt runoff process in the eastern part of Turkey.Remote Sensing of Environment, 97(2), 216-230.

World Meteorological Organization (WMO). 1986.Intercomparison of Models of Snowmelt Runoff (Operational Hydrology Report). Geneva: Secretariat of the World Meteorological Organization.

Xu, C. Y., and Singh, V. P. 1998. A review on monthly water balance models for water resources investigations.Water Resources Management, 12(1), 20-50. [doi:10.1023/A:1007916816469]

Zhang, P., Wang, J., Liu, Y., and Li, Y. 2009. Application of SRM to flood forecast and forewarning of Manasi River Basin in Spring.Remote Sensing Technology and Application, 24(4), 456-461. (in Chinese)

Zhang, Y. C., Li, B. L., Bao, A. M., Zhou, C. H., Chen, X., and Zhang, X. R. 2006. Simulation of snowmelt runoff model in the Kaidu River Basin.Science in China Series D: Earth Sciences, 36(s2), 24-32. (in Chinese)

(Edited by Yun-li YU)

This work was supported by the National Natural Science Foundation of China (Grant No. 51069017), the Special Fund for Public Welfare Industry of Ministry of Water Resources of China (Grant No. 201001065), the Open-End Fund of Key Laboratory of Oasis Ecology, Xinjiang University (Grant No. XJDX0206-2010-03), and the Open-End Fund of the China Institute of Water Resources and Hydropower Research (Grant No. IWHR-SKL-201104).

*Corresponding author (e-mail:shalamu@yahoo.cn)

Received Feb. 21, 2011; accepted Dec. 26, 2011