Mapping of wind energy potential over the Gobi Desert in Northwest China based on multiple sources of data

Abstract

In recent years, wind energy has been a fast-growing alternative source of electrical power due to its sustainability. In this paper, the wind energy potential over the Gobi Desert in Northwest China is assessed at the patch scale using geographic information systems (GIS). Data on land cover, topography, and administrative boundaries and 11 years (2000–2010) of wind speed measurements were collected and used to map and estimate the region’s wind energy potential. Based on the results, it was found that continuous regions of geographical potential (GeoP) are located in the middle of the research area (RA), with scattered areas of similar GeoP found in other regions. The results also show that the technical potential (TecP) levels are about 1.72–2.67 times (2.20 times on average) higher than the actual levels. It was found that the GeoP patches can be divided into four classes: unsuitable regions, suitable regions, more suitable regions, and the most suitable regions. The GeoP estimation shows that 0.41 billion kW of wind energy are potentially available in the RA. The suitable regions account for 25.49%, the more suitable regions 24.45%, and the most suitable regions for more than half of the RA. It is also shown that Xinjiang and Gansu are more suitable for wind power development than Ningxia.

Keywords

wind energy / alternative energy / sustainability / Northwest China / Gobi Desert

Introduction

China’s economy is in a transitional phase and consumes a huge amount of energy. It is, therefore, of great social significance to quicken the pace of the development of renewable energy. Wind energy has become a fast-growing alternative source of electrical energy in recent years. It allows for a reduced dependence on fossil fuels, thus contributing to the sustainable development of energy sources and mitigation of climate change risks. Total global wind generation capacity reached 392.93 million kW at the end of 2015, 21.68 million kW of which was installed in the first half of the year (WWEA, 2015). China’s wind capacity showed astonishing growth of more than 10 million kW within six months during 2015. The total capacity of the wind turbines installed in China by mid-2015 was equal to 2% of the world’s electricity demand (WWEA, 2015).

A wind resource assessment can provide an overall estimate of the mean energy content over a large area (Dabbaghiyan et al., 2016). It is very important that a reliable assessment of wind resources be made before they are utilized. Over the last decade, many studies on the assessment and prospects for renewable energy sources with different methodologies, especially wind energy, have been carried out across the world in countries, such as Iran (^{Keyhani et al., 2010}), Sweden (^{Siyal et al., 2015}), India (Lolla et al., 2015), Brazil (Lima and Filho, 2010), Northern Cyprus (Solyali et al., 2016), Algeria (Himri et al., 2008), the Ukraine (Sobchenko and Khomenko, 2015), in addition to numerous others. Meteorological distribution and Weibull distribution were used to conduct assessments of wind energy potential by Keyhani et al. (2010) and Dabbaghiyan et al. (2016) in Iran and by De Araujo Lima and Bezerra Filho (2010) in Brazil. Based on the statistical data of 11 years of wind speed measurements, Keyhani et al. (2010) found that the wind energy potential of the region may be adequate for non-grid connected electrical and mechanical applications, such as wind generators for local consumption, battery charging, and water pumping. Dabbaghiyan et al. (2016) tried to determine whether four locations in Bushehr (a province of Iran) were suitable for the generation of wind energy. ^{Siyal et al. (2015)} assessed the wind energy potential in Sweden by using a GIS-based approach. Results suggest that the wind energy potential and the land area available for wind energy installations are sufficient to meet Sweden’s future renewable energy targets. ^{Lolla et al. (2015)} conducted an assessment of the grid-connected wind energy potential in India based on the wind resource estimation from the Modern-Era Retrospective Analysis for Research and Applications (MERRA) Reanalysis. The results show that the southern grid has the most potential for renewable energy followed by the western, eastern, and northeastern grids. Sobchenko and Khomenko (2015) studied the wind energy content and spatiotemporal distribution characteristics in the Ukraine by applying probabilistic analysis techniques. They provided a preliminary assessment of wind resources for different regions of the Ukraine and determined the site-related temporal variability. In Northern Cyprus, Algeria, and other countries, researchers have also conducted wind energy assessments using different methods and data. In addition, research has been conducted at the global scale, including studies of the factors that have an impact on evaluations of wind energy potential. ^{Mostafaeipour (2010)} analyzed offshore wind speeds at the global scale and also studied the feasibility of installing wind turbines in the Persian Gulf, Caspian Sea, Urmia Lake, and Gulf of Oman based on satellite data. Nguyen et al. (2016) studied the factors that influence studies of wind energy potential and found that the estimated capacities depended greatly on the design of the simulation model used. This study also highlighted the importance of capturing wind and load data points that correspond to periods of extremely high demand.

Researchers in China have also conducted studies on the status of wind energy and its prospects (Xu et al., 2010), an energy analysis of a Chinese wind farm (Yang et al., 2013), and a comparative study on the differences in wind energy investment patterns between China and the European Union (Ydersbond and Korsnes, 2016). Assessments of wind energy potential have been reported for various regions in China, e.g., Henan (Zhang et al., 2011), Shandong (^{Wang, 2007}), Jiangsu (^{Xu, 2011}; He et al., 2016), Xinjiang (Bao et al., 2006), Shenyang (Jiang et al., 2009), Fujian (Sun et al., 2012) and the sea (Chang et al., 2015). The wind energy potential in these studies was calculated for an administrative unit, in addition to providing the qualitative distribution of wind speed or wind energy, or quantifying the wind power density. Additional research has also been conducted by Li et al. (2007) to summarize wind energy assessment methods, Zhu and Xue (1981,1983) and Xue et al. (2001) to summarize wind energy content and spatiotemporal distribution characteristics, and Fan and Wang (2016) to examine the factors that impact the evaluation of wind energy. However, significant differences have been observed between these study results and few validations have been conducted. Some of the research focused on large ocean or land areas and did not take the land cover types into consideration. The aim of this paper is to simulate and map the distribution of the wind energy potential area in the Gobi Desert in Northwest China based on multiple sources of data. The results discussed in this paper improve upon those of previous studies by showing a reduction in the topography’s influence on the wind speed interpolation results, while preliminary validation was also carried out. The biggest difference from the previous research is that this study calculated different types of wind energy potential and mapped it at the patch (a relatively homogeneous area that differs from its surroundings in Landscape Ecology) scale, which breaks the restriction of administrative units.

This paper is divided into five parts: 1) introduces an assessment of the wind energy potential in China and abroad; 2) gives a general overview of the study area, the data, and the estimation, calculation, and validation methods used in this work; 3) presents the results based on those methods; 4) provides a wind energy potential calculation; and 5) concludes the paper in a summarization of research results.

Materials and methods

Study area and data

Study area

China is rich in wind resources, with the highest concentration of available onshore wind energy observed in the northwest region. This paper focuses on wind energy development on unused land (mainly in the Gobi Desert) in Northwest China, which comprises the Xinjiang Uygur Autonomous Region, the Ningxia Hui Autonomous Region, and Gansu Province (Fig. 1). The research area (RA) is a vast territory with a sparse population and of great significance due to potential development of the unused land in this region. The RA accounts for 25% of the country as a whole, with an area of 2,177,700 km², situated between 32°11′N and 48°10′N latitude and 73°40′E and 108°46′E longitude. Xinjiang, the largest region of the RA, covers an area of 1,660,000 km², while Gansu covers 453,700 km², accounting for 20.83%, and the smallest area of Ningxia covers 64,000 km².

Data

In this study, meteorological, digital elevation model (DEM), land cover, and Google Earth data were all used. Details concerning these data are given below.

1) Meteorological data

The wind speed measurements of meteorological stations used in this work were collected at 10 m height for an 11-year period (2001-2010) and were provided by the China meteorological data network (http://data.cma.gov.cn/), which includes 143 sites in the RA. The data comprise daily averages of wind speed, temperature, air pressure, relative humidity, wind duration, and wind direction.

2) DEM, slope, and aspect

The SRTM DEM data used in this study is a joint product of NASA (National Aeronautics and Space Administration) and NIMA (National Imagery and Mapping Agency) for the year 2000. The data have a spatial resolution of 90 m and was downloaded from the geospatial data cloud of the Chinese Academy of Sciences (http://www.gscloud.cn/). The slope and aspect data used in this study were all derived from the SRTMDEM data - reference Burrough and McDonell (1998) for the algorithm used. The slope and aspect data, therefore, had the same resolution as the DEM. The DEM, slope, and aspect data all had to be resized to a resolution of 1 km before use.

3) Land-use and land cover change (LUCC) data

The land-use and land cover change data were acquired from the Resources and Environment Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn/). These data had a spatial resolution of 1 km and were dated from 2010. The LUCC was classified using Landsat TM/ETM imagery and was generated by artificial visual interpretation. There were two levels of land-use type. Level one consisted of six types: cropland, forest, grassland, water, unused land, and residential areas. Level two consisted of 25 types, giving a more detailed classification than level one. One of the purposes of this study was to select areas of geographical potential (GeoP) in the Gobi Desert that are more suitable for building wind power plants than other landscapes. The detailed level-two classification of unused land proved useful for this selection. The level-two unused land classes were desert, semi-desert, saline-alkali soil, marsh, bare soil, and bare rock. The Gobi Desert consists of both semi-desert and desert.

4) Google Earth data

Google Earth, a ‘virtual globe’ and geographical information program released in 2005, is now widely used. The images provided by Google Earth were used to validate the wind energy potential data.

Wind energy potential estimation and mapping methods

In this study, geographical potential (GeoP) was mapped based on the theoretical potential (TheoP), and the technical potential (TecP) was estimated. The TheoP is defined as kinetic energy available per wind mass unit as a function of the wind speed and the air density. The GeoP is defined as the potential related to the suitable areas for wind energy installation. Both the desert and semi-desert regions of the Gobi Desert were used in this study. The TecP is defined as the feasible potential that can be installed. A flow chart describing the data processing (Fig. 2) is shown below.

The TheoP calculation

The TheoP is defined as kinetic energy available per wind mass unit as a function of the wind speed and the air density and is generally considered to be a better indicator of wind resources than wind speed (Al-Nassar et al., 2005). It is expressed in Watts per square meter (W/m²) and takes into account the frequency distribution of the wind speed and the dependence of wind power on the air density and the cube of the wind speed. In terms of the wind speed, the average TheoP can be expressed as Eq. (1):

(1)\mathrm{T}\mathrm{h}\mathrm{e}\mathrm{o}\mathrm{P}={\displaystyle \frac{\mathrm{1}}{\mathrm{2}}}\rho V^{\mathrm{3}}.

This equation for the TheoP is referred to in the Wind Resource Assessment Handbook, edited by the Renewable Resources Lab in the USA (Bailey et al., 1997) and The State Standard of the People’s Republic of China: Wind energy resources assessment method (2002). Many other calculation methods are discussed in Manwell et al. (2002), Gani et al. (2016), and Grassi et al. (2015). As shown in Eq. (1), TheoP is the theoretical potential, V is the working wind speed (m/s) (which ranges from 3 to 20 m/s) and r is the daily air density (kg/m³).

The daily air density can be calculated using Eq. (2):

(2)\rho ={\displaystyle \frac{P}{R\times T}},

where P is the daily average air pressure (Pa or N/m²), T is the daily average air temperature (Kelvin), and R is the gas constant for dry air (287 J/(kg·K)).

The wind speed is processed before calculating the TheoP based on the constraints of wind energy installations.

Wind speed estimations in unknown locations

There are 143 measuring stations available in Northwest China. This scattered distribution of stations may generate uncertainties when estimating the wind speed in unknown locations far from the observations. Previous work demonstrated the efficiency of interpolation methods of climate data by using Kriging (Luo et al., 2008). Six methods of spatial interpolation (IDW, Kriging, Natural Neighbor, Spline, Topo to Raster, Trend) were also compared in this study to determine their suitability for estimating wind speed surfaces. The results indicated that Kriging is most likely to produce the best estimate of a continuous surface for wind speed. Kriging is an interpolation method to predict the variable of interest in unassembled locations from data observed at known locations. Kriging uses variograms to express the spatial variation. It minimizes the error of predicted values that are estimated by spatial distribution of the predicted values. One advantage of Kriging is that it can use the covariance between two or more variables, thus the Kriging approach was used in this work. A simple introduction to Kriging interpolation is shown below.

Kriging weights the surrounding measured values to derive a prediction for an unmeasured location. The general formula for Kriging interpolators is formed as a weighted sum of the data:

(3)\widehat{Z}(s_{\mathrm{0}})={\displaystyle \sum _{i=\mathrm{1}}^N{\lambda }_{i}Z(s_i)},

where Z(s_i)= the measured value at the i^th location, {\lambda }_i= an unknown weight for the measured value at the i^th location, s_{\mathrm{0}}= the prediction location, and N= the number of measured values.

In the Kriging method, the weights are based not only on the distance between the measured points and the prediction location, but also on the overall spatial arrangement of the measured points. To use the spatial arrangement in the weights, the spatial autocorrelation must be quantified. Thus, in ordinary Kriging, the weight, {\lambda }_i, depends on a fitted model to the measured points, the distance to the prediction location, and the spatial relationships among the measured values around the prediction location.

We proposed a model to select the wind speed interpolation results. Algorithms were then used to estimate the wind energy potential.

The working wind speed selection

Wind speed in unknown locations is not easily estimated due to the strong influence of local topographical features (Şahin and Aksakal, 1998). Due to aerodynamics, wind speed may vary significantly from one region to another, especially in complex terrain (Dabbaghiyan et al., 2016). Thus, to remove the influence of variations in topography, we formed a wind speed conversion table (Table 1) relating the wind speed for flat surfaces to that for other inhomogeneous surfaces at a height of 10 m, based on Chock and Cochran (2005) and ^{Zardi (2015)}.

In the table, 6 m/s is a demarcation point for the wind speed over a flat surface. We obtained new wind speed values that considered the effects of complex terrain by using a wind speed partial adjustment model based on Table 1. Large height differences in the RA (elevation variance from –160 m to 7311 m) could lead to unacceptable errors when wind speed processing is based on the conversion table. The best solution was to divide the RA into different regions (Fig. 3(a)) with similar altitudes (Table 2) before processing. The conversion table and the regional division of the RA are shown in Table 1 and Table 2, respectively. The significance of the different hill position names and Dh can be obtained from Fig. 3(a), a map of the RA divisions, and Fig. 3(b), which illustrates the different hill position names.

Wind direction and its relationship to the slope, the gradient of the slope, and the wind speed itself were taken into consideration in the processing model. The perennial prevailing winds in the RA typically blow from the northwest. Thus, the terrain in the RA was divided into eight parts (Fig. 4) and the ratio of wind speed for the different categories of aspect to the wind speed for flat land was calculated (Table 1). The ground layer, at a height of 10 m, was stable (Xu et al., 2012); therefore, the stable layer parameters shown in Table 1 were chosen when interpolation data were selected. This aspect plays a fundamental role as the wind resource assessment is a critical task when assessing the whole region’s potential.

The GeoP estimation and mapping

After the TheoP calculation, the GeoP was selected based on the LUCC data. The GeoP is the result of the TheoP selected by the desert, Gobi, saline-alkali soil, marsh, bare soil, and bare rock land cover types. According to the technical criteria in GB/T18710-2002⁾, we divided the GeoP calculation results into four classes: unsuitable regions (GeoP<150 W/m²), suitable regions (150 W/m²≤GeoP<250 W/m²), more suitable regions (250 W/m²≤GeoP<350 W/m²), and most suitable regions (GeoP≥350 W/m²). Each class has been estimated and mapped as the GeoP using geographic information systems (GIS). Data on land cover and administrative boundaries in GIS format were used.

The TecP calculation

Once the GeoP had been estimated and mapped, the TecP could be calculated using Eq. (4):

(4)\mathrm{TecP}=e*f*\mathrm{GeoP}*t,

where TecP is the technical potential, GeoP is the geographical potential, t (hours) is the duration of the working wind speed in a year, e is the efficiency of the wind turbine (percentage of energy that a wind turbine can extract from the available wind power), and f is the ratio of the distance between the wind power devices and their diameters.

As an example, after the wind blows over a 1-m diameter wind power conversion device, a distance 10 times that of the diameter, on all sides, is required before the wind speed returns to its original value, thus, effectively covering an area of 100 m². In this case, 10,000 conversion devices are needed to completely cover an area of 1 km² ({\displaystyle \frac{{\mathrm{10}}^{\mathrm{6}}}{\mathrm{10}0}}=\mathrm{10},\mathrm{000}). According to the Betz limit, commonly used during past decades (De Broe et al., 1999; Keyhani et al., 2010), a wind turbine cannot extract more than 59.3% of the available wind power. However, at present, the full efficiency of a wind turbine averages about 30%. Consequently, the amount of TecP_example can be calculated for an area of 1 km² using Eq. (5):

(5){\mathrm{TecP}}_{\mathrm{example}}=0\mathrm{.3}*\mathrm{10}*\mathrm{GeoP}*t,

where TecP_example is the technical potential of the 1-m diameter wind turbine (in kWh), 0.3 is the current full efficiency of a wind turbine, 10 is the ratio of the distance between the wind power devices and the diameters of the wind power devices, GeoP is the geographical potential, and t represents the duration of the working wind speed in a year.

The wind energy potential validation

In addition to selecting suitable areas for wind power development, a preliminary validation of the wind energy potential was carried out as part of this study, beginning with the distribution of the GeoP area. Wind power plants are often built in high wind energy potential areas. It is reasonable to validate the GeoP distribution area by seeking wind power plants in wind energy potential areas. This paper provides an indirect remote-sensing based method to validate the wind energy GeoP by seeking the wind power plants. High-resolution imagery provided by Google Earth was used. The validation was based on the use of remote sensing imagery to find the real distribution of wind turbine plants. If there are wind plants in the GeoP area, the result is reasonable. The value of TecP was then validated. First, the Google Earth images covering the areas of the GeoP were searched for wind turbines. Then, from several typical wind turbine locations found in different regions, three were selected at random and their sizes noted. The next step was to calculate the annual TecP for these sites based on the methodology described earlier in this paper. According to the hub height of the wind plants, we converted the annual theoretical values to 10 m in height based on Bailey et al. (1997). Finally, the annual TecP for these sites was compared to the annual converted theoretical values of the real power plants.

Results and discussion

In this study, the wind speed interpolation result was improved by reducing the topographic influence. The yearly average wind energy potential for the Gobi Desert in Northwest China was simulated over an 11-year period from 2000 to 2010. Based on these data, the GeoP values were calculated and the spatial distribution was analyzed. The TecP for different areas within the RA was also calculated. The main results obtained from the present study are summarized below.

Estimating the wind speed in unknown locations

The Kriging approach was used in this study to estimate the wind speed in unknown locations, and an accuracy assessment was performed. Of the 143 measurements used in the Kriging interpolation, 98 were for interpolation and the remainder for validation. The results are shown in Fig. 5. The R (relation) of the measured and interpolated data is 0.42, the RE (relative error) is 0.30, and the RMSE (root mean square error) is 0.78. It is true, however, that the interpolation is rough and cannot be used in the study. Therefore, this type of data must be adjusted near to the measured data after the interpolation. This adjustment is based on DEM and wind direction according to the model in Section 2.2.3 of this paper. After the adjustment, the R of the measured and simulated data is 0.89, the RE is 0.11, and the RMSE is 0.33, allowing for the simulation results to be used in this study.

The geographical potential

The GeoP map (Fig. 6) shows the distribution and classification of the suitable land in the Gobi Desert. The calculated GeoP results are distributed as regular patches. Some of them are continuous, while others are scattered. The continuous patches are found east of Xinjiang (mainly in Piqan, Ruoqiang, Hami, and Yiwu Counties, Barköl Kazakh Autonomous County, Mori Kazakh Autonomous County, and Urumqi) and north of Gansu (mainly in Subei Mongolian Autonomous County, as well as in Guazhou, Dunhuang, Yemen, and Jinta Counties). The scattered patches are spread over the southwest of Xinjiang (mainly in Hoboksar Mongolian Autonomous County, and Habahe, Buerjin, Jimunai, Tuoli, Bole, and Wuqia Counties), the center of Gansu (mainly in Minqin County), and in the Ningxia Hui Autonomous Region (mainly in the Lingwu, Yanchi, and Shapotou regions).

Secondly, we know that the GeoP areas were divided into four classes. The highest concentration of the ‘most suitable regions’ was located east of Xinjiang (mainly in Hami, Piqan, Yiwu, Barköl Kazakh Autonomous County, Mori Kazakh Autonomous County, and Urumqi) and north of Gansu (mainly in Subei Mongolian Autonomous County and Guazhou County). Scattered areas were also found in Bole, Tuoli, and Wuqia Counties of the Xinjiang Uygur Autonomous region. The regions classified as ‘more suitable’ were located east of Xinjiang and north of Gansu, and surrounded by the most suitable regions. The “suitable regions” were primarily scattered along the western edge of Xinjiang, the middle of Gansu, and in the Ningxia region.

The total GeoP, at 10 m above ground level, was estimated to be 0.41 billion kW, with the GeoP of the suitable region accounting for 0.104 billion kW or 25.49% of the total. The more suitable region accounted for 24.45% of the total or about 0.099 billion kW. The most suitable region accounted for more than half of the GeoP at about 0.204 billion kW. The actual exploitable amount of wind power can be calculated by taking the actual exploitation of the available wind power at 10% of the potential available and also taking the active area of the wind power generation devices into account. If the diameter of a wind power generation device is 1 m, its effective area will be 0.5²×p= 0.785 m², giving an area factor of 0.785. For the RA, the total exploitable GeoP was calculated to be 31.92 million kW, with the amounts for the suitable and more suitable regions both at 8.13 million kW, and the total for the most suitable region being 15.98 million kW.

Wind energy potential validation

The wind energy potential is difficult to validate because it is invisible and information about the location is minimal. In addition, both the layout of the wind power plants and field surveys are required to determine the amount of power that they can actually generate.

In this study, we validate the GeoP distribution results (Fig. 6) by verifying the existence of wind power plants in the GeoP areas based on the method discussed in Section 2.4. Figure 6 verifies the distribution of numerous wind power plants in accordance with the GeoP results. In Xinjiang, the wind power plants are primarily located in Tacheng, Burqin, Urumqi, Hami, and Barköl Kazakh Autonomous County as well as in Guazhou, Yumen, Minqin, and Dingxi in Gansu, and Lingwu and Yanchi in Ningxia. One or more plants have been found in each region. Wind plants have also been found outside of the GeoP such as Zhangye, Tongwei, and Jingtai, because these areas are not in the Gobi Desert land cover.

Three of these plants were chosen at random and the location and layout of each site was found by interpretation of high-resolution Google Earth imagery (Fig. 7). The three sites, all in Xinjiang, were the Ala Shan Kou (Fig. 8), Xiaocaohu (Fig. 9) and Shisanjianfang (Fig. 10).

The annual TecP (2001-2010) for the three wind power plants was calculated based on the wind energy potential estimation results using a wind turbine conversion efficiency of 30%. We also investigated the annual theoretical wind power generation in reality of the three validation sites and converted the value to a 10 m height. The TecP of regions A, B, and C is 0.19 billion kWh/yr, 0.33 billion kWh/yr, and 0.40 billion kWh/yr, respectively, while the amounts of wind power actually generated are 0.11 billion kWh/yr, 0.15 billion kWh/yr, and 0.15 billion kWh/yr, respectively. The values obtained from the TecP are thus about 1.72-2.67 times (2.20 times on average) higher than the real values. On this basis, the TecP results were adjusted and the amount of wind energy in different regions of the RA was calculated.

Discussion

The spatial distribution of wind speed was interpolated based on meteorological site data and the GIS spatial analysis method. By comparing the results of six interpolation methods, Kriging performed best in terms of accuracy and thus was adopted in this study. However, the influence of the complex terrain was neglected in the Kriging interpolation, especially the narrow effects of Hexi Corridor. Thus, this study proposed an advanced wind speed model based on multiple factors, which improved the accuracy of the interpolation results by reducing the topographical influences. The wind energy GeoP is the selection results of TheoP from DEM and LUCC data and was mapped at the patch scale, which makes significant improvement to the provincial and regional scales. The indirect remote-sensing method, easier and more convenient than field investigations, was used to validate the wind energy GeoP by verifying the existence of wind power plants on remote sensing images.

Selection of a suitable prevailing wind direction and acquisition of high-quality DEM and LUCC data are crucial to effectively simulate wind energy potential. In this study, the estimate of the wind energy potential might be different from the actual wind power generated by the plants, as our simulation and estimate were carried out under ideal conditions. We used the perennial prevailing winds direction to simplify the simulation processes, but in reality, the situations of wind direction are very complex and changeable. Based on these findings, this work will be improved in future studies. High-quality DEM and LUCC data, including both spatial resolution and high classification accuracy, have a positive influence on the accuracy of simulation and estimate.

Estimation of TecP

Global level

Based on the wind energy potential conversion factor of 30% used above, a series of results for TecP was obtained. For the years 2000 to 2010, the gross TecP of the RA was about 243.92 billion kWh/yr. If all of this wind energy had actually been used for power generation, it would have saved 97.57 million tons of standard coal (1 kWh= 0.4 kg standard coal), which is equivalent to a savings of 48.78 billion CNY (1 kg standard coal= 500 CNY). Further investigation showed that the total wind power generation in the year 2014 was 31.25 billion kWh/yr, equivalent to 12.81% of the total potentially available. Thus, more than 80% of the total available wind energy in the RA needs to be developed and utilized. The TecP of the most suitable region is about 150.44 billion kWh/yr, which accounts for 61.68% of the total; the more suitable region accounts for 19.92% and the suitable region the remainder.

If the maximum wind power conversion factor (59.3%) is used, the total TecP in the RA would be about 482.16 billion kWh/yr. Total electricity consumption in China was 5420.34 billion kWh in 2013. This amount of wind energy could provide 8.9% of China’s total demand. It can be seen, then, that wind power, a green energy resource, has a bright future in China.

Provincial level

The TecP at other administrative levels was also calculated - again based on a conversion factor of 30%. For the years 2000 to 2010, a comparison showed that Xinjiang had the largest TecP in the RA, with Gansu in second place and Ningxia last. This is not only due to the area of each province, but also to the complex terrain. The regions of Xinjiang and Gansu consist of many valleys and mountains that are natural corridors for wind and are, therefore, suitable for the development of wind energy. In contrast, the Helan Mountains, in the northwest area of Ningxia, form a natural wind barrier for this autonomous region. In addition, Ningxia occupies a small area of the Gobi desert. For these reasons, Ningxia has less available wind energy than Xinjiang and Gansu.

The TecP of Xinjiang is about 153.37 billion kWh/yr, accounting for 62.88% of the total (Table 3) in the RA. This could potentially lead to a savings of 61.35 million tons of standard coal (equivalent to 30.67 billion CNY) and a reduction in CO₂ emissions of 152.94 million tons. The TecP of the regions classified as suitable is 27.15 billion kWh/yr, equivalent to 17.70% of the total potential for Xinjiang with a savings of 5.43 billion CNY, whereas the more suitable regions account for 20.88% of the total potential (about 32.03 billion kWh/yr). This amount of energy is equivalent to a reduction of about 31.94 million tons in CO₂ emissions. The TecP of the most suitable region is about 94.19 billion kWh/yr, the exploitation of which could save 36.68 million tons of standard coal.

Gansu’s TecP is approximately 89.51 billion kWh/yr, accounting for 36.69% of the total in the RA. The use of this amount of wind energy would lead to a savings of 35.8 million tons of standard coal, equivalent to 17.90 billion CNY, and a reduction of 89.25 million tons in CO₂ emissions. The TecP amounts available in the suitable, more suitable, and most suitable regions are 16.70 billion kWh/yr, 16.56 billion kWh/yr, and 56.25 billion kWh/yr, respectively. The TecP of the most suitable region accounts for 62.84% of the total for Gansu, equivalent to a savings of 11.25 billion CNY, whereas the other two regions account for 37.16%. The exploitation of wind energy in these regions could lead to a savings of 13.30 million tons of standard coal or a reduction of 33.16 million tons in CO₂ emissions. It would also make a significant contribution to Gansu’s GDP.

Ningxia is not as rich in wind energy resources as Xinjiang and Gansu. It has a TecP resource of about 1.05 billion kWh/yr, which accounts for only 0.43% of the total technical wind energy potential in the RA. If all of the TecP in the Ningxia Gobi Desert was exploited, it would lead to a reduction of 1.05 million tons in CO₂ emissions, equivalent to a savings of 0.42 million tons of standard coal, and thus, of significant benefit to the environment. Due to the wind obstruction by the Helan Mountains, Ningxia does not have any regions classified as ‘more suitable’ in terms of wind energy potential. In addition, the most suitable region is very small. In contrast to Xinjiang and Gansu, the TecP of the suitable region in Ningxia accounts for the largest portion available, about 1.04 billion kWh/yr or 99.05% of the total. The use of this resource could lead to a reduction in CO₂ emissions of 1.04 million tons and to a savings of 0.21 billion CNY. It can be seen from the above analysis that Xinjiang and Gansu account for most of the TecP electricity generation and are more suitable for wind power development.

County level

The amounts of TecP in each county are shown in Fig. 11. It can be seen that Gansu Province, northern Subei Mongolian Autonomous County, Guazhou, and Yumen are the largest producers of electricity, followed by Dunhuang, Minqin, and Jinta. Other regions have only small capacities. The TecP of northern Subei Mongolian Autonomous County is 22.04 billion kWh/yr. The exploitation of this potential would lead to a savings of 8.82 million tons of standard coal. The suitable region accounts for 0.20%, the more suitable region 10.76%, and the most suitable region 89.05%. Guazhou’s potential is 7.53 billion kWh/yr, with the suitable region accounting for 3.42%, the more suitable region 45.23%, and the most suitable region the remainder. If all the available wind energy in Guazhou was used to generate electricity, 1.51 billion CNY would be saved and 7.51 million tons of CO₂ emissions avoided. Yumen is also a big wind energy producer, located outside the most suitable region. Its potential is approximately 2.69 billion kWh/yr, with the suitable region accounting for 81.29%, and the more suitable region 18.71%. The amounts of TecP in Dunhuang, Minqin, and Jinta are 2.23, 2.38, and 1.67 billion kWh/yr, respectively, equivalent to the consumption of 2.51 million tons of standard coal in total.

With the exception of Huinong County, Ningxia contains no areas classified as ‘most suitable’ in terms of TecP; however, the potential of this area is only about 0.01 billion kWh/yr. Lingwu, Yanchi, and Zhongwei have the largest potential in Ningxia at 0.12, 0.18, and 0.07 billion kWh/yr, respectively, equivalent to burning 0.15 million tons of standard coal. The areas with the next largest TecP are Qingtongxia, Wuzhong, and Tongxin at 0.03, 0.002, and 0.03 billion kWh/yr, respectively. If this TecP was exploited, it would lead to a reduction of 0.06 million tons in CO₂ emissions. The potential in other counties is so small that they are not mentioned here.

Hami, Barköl Kazakh Autonomous County, and Piqan County have the largest TecP followed by Yiwu, Mori Kazakh Autonomous County, and Ruoqiang County. The figures for other areas are minimal. The total TecP for Hami is 31.73 billion kWh/yr, of which the suitable, more suitable, and most suitable regions account for 10.06%, 15.99%, and 73.95%, respectively. The amount of TecP available in Piqan County is 9.10 billion kWh/yr, of which the suitable, more suitable, and most suitable regions account for 21.33%, 22.39%, and 56.27%, respectively. The total TecP for Barköl Kazakh Autonomous County is 6.40 billion kWh/yr, of which the suitable, more suitable, and most suitable regions account for 12.02%, 25.88%, and 62.10%, respectively. The figures for Yiwu, Mori Kazakh Autonomous County, and Ruoqiang County are 6.07, 2.15, and 2.40 billion kWh/yr, respectively.

Conclusions

In this study, the yearly average available onshore wind energy potential in Northwest China was mapped and estimated for the 11-year period from 2000 to 2010 by applying scenarios related to the topography and land use constraints (DEM and land cover) and technical criteria related to wind energy installations (refer to (Bao et al., 2006)). Additionally, we presented a preliminary validation of the estimated wind power potential using a remote sensing interpretation method. The results can be used as a guide for large-scale wind energy development in the Gobi Desert. The most important outcomes of the study can be summarized as follows.

1) The geographical potential (GeoP) of the RA is distributed regularly. The areas of similar GeoP are continuous in some areas, and scattered in others. The continuous regions are located in the middle of the RA and comprise the more suitable and most suitable regions, while other regions (mostly the suitable) are more scattered.

2) After adjusting the simulation results, the total geographical potential of the RA was 0.41 billion kW, of which the suitable region accounts for 25.49%, the more suitable region 24.45% and the most suitable region more than half at about 0.204 billion kW.

3) For a wind power conversion factor of 30%, it is concluded that Xinjiang has the largest TecP in the RA at about 153.37 billion kWh/yr, or 62.88% of the total, followed by Gansu, with 36.69%, and Ningxia with 0.43%. Xinjiang and Gansu comprise most of the TecP in Northwest China. The terrain characteristic of these areas is more suitable for wind energy development.

4) Xinjiang has the largest TecP in Northwest China, and Hami, Barköl Kazakh Autonomous County, and Piqan County have the largest TecP in Xinjiang. Gansu also has a large TecP, with northern Subei Mongolian Autonomous County, Guazhou, and Yumen being the largest producers of wind power. The areas of GeoP in Ningxia are classified mainly as ‘suitable’, and with the exception of Huinong County, none are classified as ‘most suitable’. Lingwu, Yanchi, and Zhongwei are the largest TecP areas in Ningxia. If wind energy becomes more developed in the counties mentioned above, there is the potential for them to produce a large amount of wind power in the near future.

Finally, it is worth mentioning that the work described in this paper is only a preliminary study carried out to estimate the wind energy potential of the Gobi Desert areas in Northwest China with the aim of providing decision support for wind energy development in that region. In estimating and mapping the wind energy potential, the regional divisions, and also the parameters, should be finely adjusted and more influencing factors considered so improved estimates can be made.

References

[1] Al-Nassar W, Alhajraf S, Al-Enizi A, Al-Awadhi L (2005). Potential wind power generation in the State of Kuwait. Renew Energy, 30(14): 2149–2161

[2] Bailey B H, McDonald S L, Markus M J, Elsholz K V (1997). Wind Resource Assessment Handbook: Fundamentals for Conducting a Successful Monitoring Program. New York: AWS Scientific, Inc.

[3] Bao N, Liu J, Ni W, Ye Z (2006). Wind energy resource analysis at Debauching wind farm in Xinjiang region. Acta Energize Solaris Sinica, 27: 1073–1077 (in Chinese)

[4] Burrough P A, McDonell R A (1998). Principles of Geographical Information Systems. Oxford University Press

[5] Chang R, Zhu R, Merete B, Charlotte B H, Xing X, Jiang Y (2015). Offshore Wind Resources Assessment from Multiple Satellite Data and WRF Modeling over South China Sea. Remote Sens, 7(1): 467–487

[6] Chock G Y K, Cochran L (2005). Modeling of topographic wind speed effects in Hawaii. J Wind Eng Ind Aerodyn, 93(8): 623–638

[7] Dabbaghiyan A, Fazelpour F, Abnavi M D, Rosen M A (2016). Evaluation of wind energy potential in province of Bushehr, Iran. Renew Sustain Energy Rev, 55: 455–466

[8] De Araujo Lima L, Bezerra Filho C R (2010). Wind energy assessment and wind farm simulation in Triunfo-Pernambuco, Brazil. Renew Energy, 35(12): 2705–2713

[9] De Broe A M, Drouilhet S, Gevorgian V (1999). A peak power tracker for small wind turbines in battery charging applications. IEEE Trans Energ Convers, 14(4): 1630–1635

[10] Fan X, Wang W (2016). Spatial patterns and influencing factors of China’s wind turbine manufacturing industry: a review. Renew Sustain Energy Rev, 54: 482–496

[11] Gani A, Mohammadi K, Shamshirband S, Altameem T, Petković D, Ch S (2016). A combined method to estimate wind speed distribution based on integrating the support vector machine with firefly algorithm. Environ Prog Sustain Energy, 35(3): 867–875

[12] Grassi S, Veronesi F, Schenkel R, Peier C, Neukom J, Volkwein S, Raubal M, Hurni L (2015). Mapping of the global wind energy potential using open source GIS data. In: 2nd International Conference on Energy and Environment: Bringing Together Engineering and Economics

[13] He Z, Xu S, Shen W, Long R, Yang H (2016). Overview of the development of the Chinese Jiangsu coastal wind-power industry cluster. Renew Sustain Energy Rev, 57: 59–71

[14] Himri Y, Rehman S, Draoui B, Himri S (2008). Wind power potential assessment for three locations in Algeria. Renew Sustain Energy Rev, 12(9): 2495–2504

[15] Jiang X, Ming X, Zhang J, Zhang S, Zhang S, Song J (2009). Distribution characteristics and exploitation and utilization of wind energy resources of Shenyang City, China. Resources Science, 31: 1764–1771 (in Chinese)

[16] Keyhani A, Ghasemi-Varnamkhasti M, Khanali M, Abbaszadeh R (2010). An assessment of wind energy potential as a power generation source in the capital of Iran. Tehran. Energy, 35(1): 188–201

[17] Li Z, Zhu R, He X, Zhang D (2007). Study on the assessment technology of wind energy resource. Acta Meteorol Sin, 65: 708–717

[18] Lolla S, Roy S B, Chowdhury S (2015). Wind and solar energy resources in India. Energy Procedia, 76: 187–192

[19] Luo W, Taylor M C, Parker S R (2008). A comparison of spatial interpolation methods to estimate continuous wind speed surfaces using irregularly distributed data from England and Wales. Int J Climatol, 28(7): 947–959

[20] Manwell J F, McGowan J G, Rogers A L (2002). Wind Energy Explained: Theory, Design and Application. Amherst: Wiley

[21] Mostafaeipour A (2010). Feasibility study of offshore wind turbine installation in Iran compared with the world. Renew Sustain Energy Rev, 14(7): 1722–1743

[22] Nguyen C, Ma C, Hailu A, Chalak M (2016). Factors influencing calculation of capacity value of wind power: a case study of the Australian National Electricity Market (NEM). Renew Energy, 90: 319–328

[23] Şahin A Z, A Aksakal (1998). Wind power energy potential at the northeastern region of Saudi Arabia. Renew Energy, 14(1–4): 435–440

[24] Siyal S H, Mörtberg U, Mentis D, Welsch M, Babelon I, Howells M (2015). Wind energy assessment considering geographic and environmental restrictions in Sweden: a GIS-based approach. Energy, 2015, 83: 447–461

[25] Sobchenko A, Khomenko I (2015). Assessment of regional wind energy resources over the Ukraine. Energy Procedia, 76: 156–163

[26] Solyali D, Altunc M, Tolun S, Aslan Z (2016). Wind resource assessment of Northern Cyprus. Renew Sustain Energy Rev, 55: 180–187

[27] Sun Y, Wang R, Liu J, Xiao L, Yang D (2012). Assessment of onshore wind energy potential in Fujian Province based on GIS. Resources Science, 34: 1167–1174 (in Chinese)

[28] Wang J (2007). The analysis and calculation of the wind energy resources in Shandong Province. Dissertation for Master degree. Lanzhou: Lanzhou University (in Chinese)

[29] WWEA (2015). Bulletin special issue 2015. World Wind Energy Association

[30] Xu J, He D, Zhao X (2010). Status and prospects of Chinese wind energy. Energy, 35(11): 4439–4444

[31] Xu X (2011). The study on some issues in evaluation, exploitation and utilization of wind resources in Jiangsu Province. Dissertation for PhD degree. Nanjing: Nanjing University of Information Science and Technology, China (in Chinese)

[32] Xu Y, Wan X, Fu C, Liu C (2012). Study of wind speed interpolation in complex terrain—A case of Jilin Province. Yunnan Geographic Environment Research, 24: 78–81 (in Chinese)

[33] Xue H, Zhu R, Yang Z, Yuan C (2001). Assessment of wind energy reserves in China. Acta Energiae Sloaris Sinica, 22: 167–170 (in Chinese)

[34] Yang Q, Chen G, Liao S, Zhao Y H, Peng H W, Chen H P (2013). Environmental sustainability of wind power: an emergy analysis of a Chinese wind farm. Renew Sustain Energy Rev, 25: 229–239

[35] Ydersbond I M, Korsnes M S (2016). What drives investment in wind energy? A comparative study of China and the European Union. Energy Research & Social Science, 12: 50–61

[36] Zardi D (2015). Mountain Meteorology| Valley Winds. Reference Module in Earth Systems and Enviromental Sciences. Encyclopedia of Atmospheric Sciences, 114–134

[37] Zhang H W, Chen H L, Zhang H (2011). Utilization and spatial-temporal distribution of wind energy resources in Henan Province based on GIS. Journal of Natural Resources, 26: 1021–1029 (in Chinese)

[38] Zhu R, Xue H (1981). The calculation and distribution of the wind energy in China. Meteorology, 8: 26–28 (in Chinese)

[39] Zhu R, Xue H (1983). Division of wind energy in China. Acta Energize Solaris Sinica, 4: 123–132 (in Chinese)

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag GmbH Germany