China-Pakistan Economic Corridor Zone II Versions EN2 Vol 4 (3) 2019
A dataset of monthly temperature vegetation dryness index along the China–Pakistan Economic Corridor from 2000–2017
: 2018 - 08 - 01
: 2018 - 09 - 25
: 2019 - 07 - 22
773 5 0
Abstract & Keywords
Abstract: Drought disasters happen frequently along the China–Pakistan Economic Corridor (CPEC), which harm the development and safety of the countries along it. It is therefore important to conduct drought monitoring along CPEC. While soil water content (SWC) is a fundamental indicator of drought prediction, temperature vegetation dryness index (TVDI) can predict drought by inverting surface soil water content. Based on a combination of MODIS normalized differential vegetation index (NDVI) and land surface temperature (LST) products and SRTM DEM data, the study used NDVI-LST spatial characteristics to extract the spatial edge of dryness and wetness along CPEC from 2000 – 2017, through which to obtain the monthly TVDI. The study area is from 41°25'24.49"N to 23°45'24.49"N and from 60°53'57.97"E to 79°52'27.97"E. The data is in GeoTiff format, with a spatial resolution of 1 km. We then compared the data with the precipitation and soil water content observed by meteorological stations. Results show that TVDI has a significantly negative correlation with Standardized Precipitation Index (SPI) and SWC . The dataset can be used to support drought monitoring and prediction along CPEC.
Keywords: China–Pakistan Economic Corridor; temperature vegetation dryness index; land surface temperature; normalized differential vegetation index
Dataset Profile
TitleA dataset of monthly temperature vegetation dryness index along the China–Pakistan Economic Corridor from 2000 – 2017
Data corresponding authorZhang Yaonan (
Data authorsFeng Keting, Zhang Yaonan, Tian Deyu, Kang Jianfang
Time range2000 – 2017
Geographical scope41°25′24.49″N - 23°45′24.49″N, 60°53′57.97″E - 79°52′27.97″E
Spatial resolution1 km
Data volume634 MB
Data format*.tif
Data service system
Sources of fundingNational R&D Infrastructure and Facility Development Program of China “National Special Environment and Function of Observation and Research Station Shared Service Platform” (Y719H71006); Information Program of the Chinese Academy of Sciences “Construction and Application of ‘Technology Cloud’ in Cold and Arid Regions Environment Evolution” (XXH13506).
Dataset compositionThe dataset consists of 216 files of monthly temperature vegetation dryness index data in total. The files are named in the format of TVDI.AYYYYDDD.1_km_month.tif.
1.   Preface
China-Pakistan Economic Corridor, covering the whole country of Pakistan and the surrounding areas of Kashgar, Xinjiang (Figure 1), starts from Kashgar, Xinjiang, China and ends at Gwadar Port, Pakistan, with a total length of 3,000 km. Linking “Silk Road Economic Belt” in the north and connecting“21st Century Maritime Silk Road” in the south, it is a key hub crossing the North-South Silk Road, a trade corridor including roads, railways, oil and gas, and cable passages, as well as a key component[1][2] of the“Belt and Road” Initiative. The Corridor will have a major impact on the economic and social development of China and Pakistan, and will play an active role in demonstrating and promoting the construction of the “Belt and Road”[3] . Frequent droughts along the China-Pakistan Economic Corridor have severely affected the security and social development of countries along the route, restricting the implementation of the “Belt and Road” Initiative[4]. Therefore, it is necessary to use various types of data to conduct monitoring research on drought disasters along the China-Pakistan Economic Corridor, which will lays down a strong theoretical basis for combating drought and disasters and risk assessment, provides scientific and technological support for further mastering effective comprehensive drought indicators, and can develop into decision-making reference for drought-resistant production practices. Eventually, by using this research method, we can promote the scientific and technological cooperation in disaster monitoring, early warning, disaster relief and disaster reduction between China and countries along the “Belt and Road”.

Figure 1   Demonstration map of the study area
The monitoring and analysis of drought has long been a highly concerned issue among the government and academia[5]in a Networked Society. The traditional drought monitoring method, based on observations or experimental observations of ground stations, uses meteorological and hydrological data such as precipitation, temperature, evaporation, and runoff obtained by meteorological and hydrological observatories, as well as data on agrometeorological observations and various drought indicators, to conduct statistical analysis of the observed data and quantify the drought. Due to the limited spatial density of the observation sites, it is difficult to monitor the drought in an extensive, rapid and continuous way simply basing on the data from the ground observation sites. With the development and application of remote sensing technology, remote sensing drought monitoring has become an indispensable means of global drought and disaster reduction. It combines with traditional disciplines and complements each other's advantages, which can provide regional, continental and even global drought information[6]and acts as a macroscopic, rapid, objective and economical effective means[7] . In remote sensing drought monitoring, the combination of vegetation index and surface temperature for drought monitoring is widely used[8][9][10][11][12][13] . Among them, the Temperature Vegetation Dryness Index (TVDI) method is the most widely used. Sandholt[14]proposed a TVDI index to estimate soil surface water conditions basing on the relationship between vegetation index and surface temperature. Domestic scholars use the TVDI index to conduct drought monitoring at different spatial scales such as national levels, regional levels, and provincial levels. Qi Shuhua[15]conducted national drought monitoring with the TVDI method. The results show that the method is reasonable for large-scale evaluation of drought. Ran Qiong[16]performed elevation correction on the surface temperature, so that TVDI obtained by elevation correction can better reflect soil moisture. Yang Xiuhai[16]found that TVDI can basically reflect the surface soil moisture conditions, and it is feasible to use TVDI to monitor the summer drought in a dynamical way in the northwest. Yao Chunsheng[17] used the TVDI method to invert the soil moisture in Xinjiang. Zhang Shunqia[18]monitored and evaluated the drought in Sichuan by means of TVDI in 2006, whose results are basically consistent with the results of climate monitoring. Du Lingtong[19]used TVDI to monitor drought in Ningxia and analyzed the characteristics of drought changes over the past decade. Shasha[20]constructed three kinds of characteristic spaces of vegetation index and surface temperature using historical remote sensing data, and discussed the suitability of TVDI method in Longdong area of Gansu Province. However, when TVDI is used for drought monitoring, the terrain fluctuations in the study area and the difference between the north and south latitude spans could affect the surface temperature data and thus lead to the calculation error of TVDI, the accuracy of TVDI inversion would be accordingly reduced. Ran Qiong[21] considered the influence of elevation changes on the surface temperature and corrected the elevation of the surface temperature so that the TVDI obtained by the elevation correction can better reflect the soil moisture. Zhao Jiepeng[22]used geography latitude and ground elevation to correct the surface temperature so as to achieve difference-correction of solar radiation and atmospheric background in large areas, as well significantly improve the accuracy of TVDI’s monitoring soil moisture.
This dataset uses the TVDI method[14], the elevation and latitude to correct the surface temperature[21][22] and calculates the temperature vegetation drought index. Based on the MODIS vegetation index and surface temperature data and the Digital Elevation Model) , DEM) data, this dataset obtains the "China-Pakistan Economic Corridor" region monthly temperature vegetation drought index data set from 2000 to 2017, providing basic data for regional disaster research and decision-making.
2.   Data acquisition and processing Method
The data used in this dataset are MODIS vegetation index product MOD13A3, surface temperature product MOD11A2, SRTM DEM products, and weather station observation precipitation data and soil moisture data. The MODIS data comes from NASA Land Processes Distribution Active Archive Center (LPDAAC,; DEM data from SRTM dataset provided by the Geography Spatial Data Cloud (; precipitation and soil moisture data from the National Meteorological Data Sharing Service Platform ( The data production process includes data preprocessing, data reconstruction, calculation and evaluation, as shown in Figure 2.

Figure 2   Data Production Flow
Among them, the data preprocessing is data splicing, projection conversion, band extraction, resampling, etc. The extracted bands include Normalized difference vegetation index (NDVI), Land Surface Temperature (LST) and the corresponding Quality Assurance (QA); data reconstruction includes spatial interpolation, time series filtering, terrain correction, time series completion, time series reconstruction; calculation and evaluation of TVDI calculations, and the use of Standardized Precipitation Index (SPI) and soil moisture for the index evaluation.
2.1   Data preprocessing
The MODIS vegetation index product MOD13A3 has a time resolution of month, a spatial resolution of 1 km, a surface temperature product MOD11A2 with a time resolution of 8 days, a spatial resolution of 1 km, a data format of hdf, and a projection mode of sinusoidal map projection. The data is spliced, projected, band extracted, and resampled with the MODIS Reprojection Tool (MRT). MRT parameter setting: the output format is Geotiff, the output projection adopts geographic projection, the level surface WGS84, and the sampling method nearest neighbor sampling, and the pixel size 0.0083333333 degrees (1 km), to obtain data of NDVI, surface temperature and related quality control.
The DEM data format is Geotiff with a resolution of 90 m. The ArcGIS tool is applied to splice, resample, and crop the DEM data to generate DEM data of the China-Pakistan Economic Corridor with a resolution of 1 km.
Python language programming is applied, combined with the study area vector boundary and MODIS quality control data, realizing the data batch cutting and masking, eliminating the quality unreliable pixels, to generate the quality NDVI and surface temperature data sets of the China-Pakistan Economic Corridor.
2.2   Data Reconstruction Processing
Remote sensing data reconstruction aims to use a variety of statistical and numerical analysis methods to simulate missing data or improve the accuracy of inversion models, so as to interpolate missing observations, optimize time series data, and provide more complete basic data for related research. Remote sensing data reconstruction methods are divided into two types: spatial reconstruction and temporal reconstruction. The data is spatially and temporally reconstructed during the production process.
2.2.1   Processing of Spatial Interpolation
If there are mass unreliable pixels missing in the data pre-extracted NDVI and temperature data, it will be necessary to perform spatial interpolation, and apply the inverse distance weighting (IDW) method to achieve spatial interpolation of data. Python will be applied first to call the RasterToPoint module of the ArcGIS arcpy package to convert the raster dataset into point features, and then call the IDW module of the ArcGIS arcpy package to interpolate the points into a raster surface to implement data bulk space interpolation. The IDW method parameters are set as follows:
(1) Distance index: applied to control the saliency of the points around the interpolation value. The higher the value is, the smaller the data point image is with a longer distance. Generally, the value range is between 0.5 and 3. The most reasonable result can be obtained. This time a value of 2 has been chosen;
(2) Search radius: defines the input point for interpolating the missing pixel value, including variable search radius and fixed distance to specify the input sample point, select the variable search radius mode, and the number of nearest neighbor input sample points is 12;
(3) Cell size: Set to be the same as the input image data.
2.2.2   Processing of Time Series Filtering
On the basis of spatial reconstruction, S-G filtering is used to fit and reconstruct multi-phase images in time series. The SG filter fitting method is a convolution algorithm based on smoothed time series data and least squares principle proposed by Savitzky [23] etc. in 1964. It is a weighted average algorithm for moving windows, but its weighting coefficient is not A simple constant window, but by a least squares fit of a given higher-order polynomial over a sliding window, the expression is:
In equation (1), is the fitted value, is the original value of the pixel, is the coefficient when filtering the i th value, m is the width of the half filter window, and N is the length of the filter, equal to the width of the sliding array. 2m+1.
In the S-G filtering, the iterative result of the fitting effect index taking the minimum value is the optimal filtering effect, and the calculation formula is:
\({F}_{k}={\sum }_{i=0}^{N}\left|{Y}_{i}^{k}-{Y}_{i}^{0}\right|×{W}_{i}\) (2)
In equation (2), \({F}_{k}\) is the sequence fitting result index after the k th iteration, \({Y}_{i}^{0}\) and \({Y}_{i}^{k}\) are respective the i th value in the sequence after no iteration and the k th iteration, \({W}_{i}\) is the sequence number the weight of i th values, N is the filter length.
The processing flow is as follows:
(1) Determine the half width m of the filtering window and the order d of the polynomial fitting. Usually, the m value is 4-7, the d 2~4, and the data set processing selects m=4, d=2;
(2) Performing S-G filtering processing on the time-series data after spatial interpolation to generate time series data;
(3) Calculating the weight of each point in the sequence;
(4) Generate new time series data according to the weighted, spatially interpolated time series data and the S-G filtered sequence, and fit the new time series data by S-G filtering;
(5) Calculate the fitting effect index. If the fitting effect index reaches the minimum, exit the iteration; otherwise, return to step (4) to continue the iterative processing.
2.2.3   Processing of Corrected Temperature Data by Terrain
Studies have shown that [14], the accuracy of TVDI model inversion is affected by factors such as surface temperature, vegetation coverage, surface parameters, atmospheric conditions and solar radiation, while geographic latitude and ground elevation are two important factors affecting atmospheric background differences and solar radiation factor. The influence of the terrain fluctuations in the study area and the north-south latitude span on the MODIS surface temperature data will bring the calculation error of TVDI. Therefore, the surface temperature needs to be corrected [20-22]. The formula for the surface temperature correction is given by equation (3).
\({T}_{c}={T}_{s}+a×H+b×L+c\) (3)
In formula (3), \({T}_{c}\) is the corrected surface temperature, \({T}_{s}\) is the MODIS surface temperature before correction, H is elevation, L is latitude, a is elevation correction coefficient, b and c are latitude correction coefficients, respectively, where a is usually 0.006 (0.6) °C/100 m) [21-22], some studies have shown that the optimal value of a in Xinjiang is 0.003~0.005 (0.3~0.5°C/100 m), and the latitude correction coefficients b and c are respective 0.3~0.5, −20 ~−12 [22], in the production of this dataset, a takes a value of 0.003, b 0.4, and c −16.
2.2.4   Time series data processing
Due to the missing of partial image periods in the MODIS data sequence, it is necessary to complete the missing period image. During the treatment, the missing period images were complemented by the method of long-time average annual value of the same time period data. Since the time resolution of NDVI and surface temperature data required in the calculation is month, and the MODIS 1 km surface temperature has no monthly resolution data, the MOD11A2 data is hence converted into a monthly scale with the method of averaging the data within 8 days of a month.
2.3   Temperature vegetation index calculation
TVDI [14] is a soil moisture monitoring method based on NDVI-LST feature space, which has certain physical significance. It can provide more accurate and abundant drought information than NDVI or LST alone. The negative correlation between the slope of LST and NDVI and soil moisture is an important statistical feature in the feature space. As the surface vegetation coverage increases, the surface temperature begins to decline. When the local surface is dry and water-deficient, the surface temperature will rise rapidly; on the contrary, when the soil is moist, the surface temperature will increase slightly. According to the principle of feature space, the key to calculating TVDI is the fitting of wet and dry edges. The existing research results show that there is a significant negative correlation between NDVI and LST on the dry edge in the feature space, which indicates that when the vegetation is subjected to water supply, the surface temperature decreases as vegetation coverage increases. In most studies, NDVI and LST on the wet side are positively correlated or unrelated, but mostly positively correlated [24]. The calculation equation for TVDI is:
\(TVDI=\frac{{T}_{c}-{{T}_{c}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}{{{T}_{c}}_{\mathrm{m}\mathrm{a}\mathrm{x}}-{{T}_{c}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}\) (4)
\({{T}_{c}}_{\mathrm{m}\mathrm{a}\mathrm{x}}={a}_{1}+{b}_{1}×NDVI\) (5)
\({{T}_{c}}_{\mathrm{m}\mathrm{i}\mathrm{n}}={a}_{2}+{b}_{2}×NDVI\) (6)
In equations (4), (5), and (6), \({T}_{c}\) is the surface temperature of any pixel, and \({{T}_{c}}_{\mathrm{m}\mathrm{a}\mathrm{x}}\) and \({{T}_{c}}_{\mathrm{m}\mathrm{i}\mathrm{n}}\)are respectively the lowest and the highest surface temperatures corresponding to a certain NDVI, a1b1a2b2 are respectively the undetermined coefficients and the value range of TVDI is [0, 1]. For a set of NDVI and LST remote sensing images, the NDVI takes a step size of 0.01, and obtains the highest and lowest LST values for each NDVI. The least squares method is used to fit the dry and wet edge equations, and insert the equations (5), (6) into (4) to obtain TVDI. Figure 3 shows the fitting effect of the monthly dry and wet edges in the study area in 2009, and Table 1 shows the fitting results of the wet and dry edges.













Figure 3   Monthly NDVI-LST feature space and dry and wet edges fitting curve in 2009
It can be seen from Fig. 3 that as the vegetation index increases, the maximum surface temperature decreases, while the minimum surface temperature increases. From the fitting effect (Table 1), the dry edge slope is less than 0, the wet edge slope is greater than 0, the dry edge fitting correlation coefficient R2 is 0.8889, and the other 11 months are greater than 0.90. NDVI and LST are significantly negatively correlated (p<0.001); the wet-edge fitting correlation coefficient R2 has a minimum value of 0.5113 and a maximum value of 0.9283. NDVI has a significantly positive correlation with LST (p<0.001), and the overall effect of dry-wet edge fitting is sound. In addition, the constant term of the fitting equation of the wet and dry edge, the intercept of the dry and wet edge on the vertical axis (surface temperature) in the feature space, represents the surface temperature of the bare soil pixel when the water is sufficient or deficient. It can be seen from Fig.1 that the intercept of the wet and dry edge changes correspondingly with the change of temperature during the year, that is, the intercept is small in winter and is large in summer.
Table 1   Monthly dry and wet edges fitting equation and related coefficient in 2009
MonthDry edge equationR2Wet edge equationR2
3月\({{T}_{c}}_{\mathrm{m}\mathrm{a}\mathrm{x}}=-25.116×NDVI+45.402\)0.9655\({{T}_{c}}_{\mathrm{m}\mathrm{i}\mathrm{n}}=20.781×NDVI-\mathrm{ }3.754\)0.776
4月\({{T}_{c}}_{\mathrm{m}\mathrm{a}\mathrm{x}}=-22.560×NDVI+51.435\)0.9675\({{T}_{c}}_{\mathrm{m}\mathrm{i}\mathrm{n}}=19.035×NDVI+ 3.326\)0.7582
5月\({{T}_{c}}_{\mathrm{m}\mathrm{a}\mathrm{x}}=-14.977×NDVI+54.194\)0.8889\({{T}_{c}}_{\mathrm{m}\mathrm{i}\mathrm{n}}=20.063×NDVI+ 5.752\)0.8995
6月\({{T}_{c}}_{\mathrm{m}\mathrm{a}\mathrm{x}}=-16.780×NDVI+57.159\)0.9027\({{T}_{c}}_{\mathrm{m}\mathrm{i}\mathrm{n}}=19.858×NDVI+ 8.527\)0.9283
7月\({{T}_{c}}_{\mathrm{m}\mathrm{a}\mathrm{x}}=-30.291×NDVI+62.142\)0.9375\({{T}_{c}}_{\mathrm{m}\mathrm{i}\mathrm{n}}= 7.259×NDVI+15.157\)0.7943
8月\({{T}_{c}}_{\mathrm{m}\mathrm{a}\mathrm{x}}=-31.644×NDVI+60.217\)0.9452\({{T}_{c}}_{\mathrm{m}\mathrm{i}\mathrm{n}}= 8.726×NDVI+13.899\)0.5113
9月\({{T}_{c}}_{\mathrm{m}\mathrm{a}\mathrm{x}}=-31.321×NDVI+54.065\)0.9634\({{T}_{c}}_{\mathrm{m}\mathrm{i}\mathrm{n}}=28.402×NDVI+ 2.008\)0.7655
3.   Explanation of Data Sample
3.1   Naming Rules
The naming rules for the monthly temperature vegetation drought index data set for the China-Pakistan Economic Corridor from 2000 to 2017 are as follows: TVDI.AYYYYDDD.1_km_month.tif, the specific respective meanings are:
(1) TVDI: represents the temperature vegetation drought index product;
(2) AYYYYDDD: represents that the product time is DDD day of YYYY year (January 1st of each year is set as the first day);
(3) 1_km: represents that the product spatial resolution is 1 km;
(4) month: represents that the product is monthly data.
For example, TVDI.A2017032.1_km_month.tif represents the temperature vegetation drought index products that the spatial resolution is 1 km in February 2017.
3.2   Explanation of Data
It is shown in Table 2 that the product information of the monthly temperature vegetation drought index for the China-Pakistan Economic Corridor from 2000 to 2017. Among them, TVDI ranges from 0 to 1, and is magnified 10, 000 times during storage. The pixel value ranges from 0 to 10000, the padding value is −3000, and the data type is signed int16. The multiplication factor is multiplied by a scale factor of 0.0001.
Table 2   Information of Temperature Vegetation Drought Index Product
1Number of Wave Band1
2Pixel Value0~10000
3Data Typessigned int 16
4Scale Factor0.0001
5Padding Value−3000
6Line Number2120
7Column Number2277
8Pixel Size0.0083333333, 0.0083333333
9Coordinate SystemWGS84
Taking TVDI as the drought grading index, the drought grade is divided into 5 levels [15]: wet (0 ≤ TVDI ≤ 0.2), normal (0.2 < TVDI ≤ 0.4), light drought (0.4 < TVDI ≤ 0.6), moderate drought (0.6 <TVDI ≤ 0.8) and heavy drought (0.8 < TVDI ≤ 1). Figure 4 shows the drought level map divided by TVDI grading indicators.

(a)April, 2017

(b)September, 2017

Figure 4   Map of Drought Level in the China-Pakistan Economic Corridor
4.   Control and Evaluation for Data Quality
4.1   Data Quality Control
The quality of TVDI data products is related to that of NDVI and LST products. Therefore, the quality of TVDI products is guaranteed by quality control of NDVI and LST data products. According to the TVDI data production process (Fig. 2), quality control includes processing of quality trusted data processing and time-space reconstruction of extracted quality trusted data products. The untrusted pixels in the product are generally cloud coverage or areas with strong atmospheric effects. The quality of the trusted data is processed according to the MODIS product quality control document, and the untrusted pixels among them are eliminated. The untrusted pixels were spatially interpolated with its surrounding trusted pixel values to complement the missing pixels, then the time-series filtering reconstruction of the interpolated products, and improving data quality from both time and space.
Among them, the extracted quality trusted data processing, in accordance with the NDVI product data description and the QA value, divides the NDVI pixel into two types: trusted and untrusted, and the trusted ones are extracted. The processing flow is as follows:
(1) The pixel with the pixel value of 0 in the pixel reliability band (1 km monthly pixel reliability) is marked as a trusted pixel;
(2) There are a large number of glacier in the study area, so the pixels with the pixel value of 2 in 1 km monthly pixel reliability are marked as trusted pixels;
(3) 1 km monthly pixel reliability pixels with the pixel value of 3 are marked as untrusted pixels;
(4) 1 km monthly pixel reliability pixels with the pixel value of 1 are calibrated as pending pixels;
(5) For the pending pixels, the corresponding 1km monthly VI Quality pixel value 0–5 digits are compared to determine the credibility of the cell: when the 0–1 bit value is 0, it is determined to be a trusted cell; When the 0–1 bit value is 1, the credibility is determined according to the 2–5 bit value;
(6) Mask process, eliminate untrusted pixels, and extract trusted pixels, thus generating new NDVI data.
For the LST data, according to the LST Product Quality Control (QC) document description, the quality trusted pixels are extracted. The processing flow is as follows:
(1) The pixels in the quality control band QC_Day with a 0–1 bit value of 0 are calibrated as trusted cells;
(2) Cells with 0–1 bit values​of 2 and 3 in QC_Day are marked as untrusted cells;
(3) The pixels with the 0–1 bit value of 1 in QC_Day are marked as the pending pixels;
(4) For the pending pixels, the value of the corresponding pixel 2–7 bits in QC_Day are compared to determine the confidence of the pixels: when the value of 2–3 bits is 0, the calibration is a trusted cell; when 2– When the 3-bit value is 1, the corresponding 4–5 bit value is 0, and the 6–7 bit value 0 is labeled as a trusted cell; other pixels are labeled as untrusted cells;
(5) Mask process, eliminate untrusted pixels, and extract trusted pixels, thus generating new data of surface temperature.
4.2   Correlation Analysis between TVDI and SPI of Stations & Soil Moisture
SPI has been proven to be a good indicator of meteorological drought. TVDI is also a method of soil moisture monitoring. Therefore, TVDI is evaluated by SPI and measured soil moisture. Since the observation data of meteorological stations in Pakistan have not been obtained, only the monthly scale SPI calculated by the precipitation data of 7 meteorological stations in China and the soil moisture data of 3 meteorological stations are available to verify the validity of TVDI. The SPI of 7 meteorological stations in 1960–2016 was calculated, and the TVDI values of the corresponding pixels of the seven meteorological stations in 2000–2016 were extracted, and the correlation between the TVDI of each weather station and the monthly SPI and soil moisture was analyzed. The results (Fig.5) show that TVDI has a negative correlation with SPI and soil moisture. The correlation coefficient between TVDI and monthly scale SPI is −0.389, the minimum is −0.158, except for the correlation between TVDI and SPI in Ulugqat station (p=0.024) which passed the significance test of p<0.05, the correlation between TVDI and SPI at other stations passed the significance test of p<0.01; the correlation coefficient between TVDI and measured soil moisture ranged from −0.656 (max) to −0.217 (min). The correlation of humidity passed the significance test of p < 0.01.



Figure 5   Correlation Diagram between TVDI and SPI, Soil Moisture in Meteorological Stations



Figure 5   Correlation Diagram between TVDI and SPI, Soil Moisture in Meteorological Stations



Figure 5   Correlation Diagram between TVDI and SPI, Soil Moisture in Meteorological Stations



Figure 5   Correlation Diagram between TVDI and SPI, Soil Moisture in Meteorological Stations



Figure 5   Correlation Diagram between TVDI and SPI, Soil Moisture in Meteorological Stations
5.   Usage and Recommendations of Data
The China-Pakistan Economic Corridor monthly temperature vegetation drought index dataset 2000-2017 reflects the characteristics of regional drought on the monthly scale. The data is stored on a monthly basis in the format of GeoTiff. Common GIS and remote sensing software such as ArcGIS and ENVI are applied in data reading and operation. TVDI is represented by the pixel value of the image. Users can classify the data according to their own grading standard to generate a drought level map when use.
Thank the USGS for providing MODIS data, thank the Geospatial Cloud for providing DEM data, and thank the National Meteorological Data Sharing Service Platform for providing data of precipitation and soil moisture.
[1] Ding Siyang, Zhu Wenquan, Jiang Yuan, etc. Evaluation of Ecological Status of China-Pakistan Economic Corridor Based on RS and GIS [J]. Journal of Beijing Normal University (Natural Science Edition), 2017, 53(03): 358-365.
[2] Wang Huizhi, Hui Zhifan, Xu He. Ecological Environment Risk Analysis and Coping Strategies in the Construction of “China-Pakistan Economic Corridor” [J]. Journal of Hohai University (Philosophy and Social Sciences Edition), 2017, 19(01): 65-68, 90-91.
[3] Hu Angang, Ma Yingjun, Gao Yuning. “Belt and Road Initiative”: Creating an Updated Version of the Opening up and a Global Open Dividend [J]. Journal of Hohai University (Philosophy and Social Sciences Edition), 2016, 18(04): 1-5, 89.
[4] Cui Peng, Su Fenghuan. Application of Domestic High-resolution Satellites in the “Belt and Road Initiative” Natural Disaster Risk Management [J]. Satellite Application, 2016(10): 8-11.
[5] Guo Ni, Wang Xiaoping. Advances in Technology of Remote Sensing Drought Application & Technical Problems and Development Oppurtunity[J]. Journal of Arid Meteorology, 2015, 33(01): 1-18.
[6] Liu Qinhuo, Xin Jingfeng, Xin Xiaozhou, etc. Agricultural Drought Remote Sensing Monitoring Method Based on Surface Temperature and Vegetation Index [J]. Science & Technology Review, 2007 (06): 12-18.
[7] Liu Huan, Liu Ronggao, Liu Shiyang. Remote Sensing Monitoring Method for Drought and its Application Development [J]. Geo-Information Science, 2012, 14(02): 232-239.
[8] KARNIELI A, AGAM N, PINKER R T, et al. Use of NDVI and land surface temperature for drought assessment: Merits and limitations [J]. Journal of Climate, 2010, 23: 618–633.
[9] KOGAN F. Application of vegetation index and brightness temperature for drought detection [J]. Advanced in Space Research, 1995, 15(11): 91-100.
[10] KOGAN F. World droughts in the new millennium from AVHRR-based vegetation health indices [J]. Eos, Transactions American Geophysical Union, 2002, 83(48): 557.
[11] KOGAN F, STARK R, GITELSON A, et al. Derivation of pasture biomass in Mongolia from AVHRR-based vegetation health indices [J]. International Journal of Remote Sensing, 2004, 25: 2889-2896.
[12] LAMBIN E F, EHRLICH D. The surface temperature-vegetation index space for land cover and land-cover change analysis [J]. International Journal of Remote Sensing, 1996, 17(3): 463-487.
[13] Wang Pengxin, WAN Z M, Gong Jianya. Drought monitoring model based on vegetation index and land surface temperature [J]. Advance in Earth Sciences, 2003, 18(4): 527-533.
[14] SANDHOLT I, RASMUSSEN K, ANDERSEN J. A simple interpretation of the surface temperature/vegetation index space for assessment of surface moisture status [J]. Remote Sensing of Environment, 2002, 79(2): 213-224.
[15] Qi Shuhua, Wang Changyao, Niu Zheng. National drought monitoring research with temperature vegetation drought index (TVDI) [J]. Journal of Remote Sensing,2003(05):420-427, 436.
[16] Yang Xiuhai, Zhuo Ga, Luo Bu. Drought monitoring in Northwest China based on MODIS data [J]. Pratacultural Sciences, 2011, 28(08): 1420-1426.
[17] Yao Chunsheng, Zhang Zengxiang, Wang Xiao. Inversion of soil moisture in Xinjiang with temperature vegetation drought index method (TVDI) [J]. Remote Sensing Technology and Application, 2004(06): 473-478.
[18] Zhang Shunqian, Qing Qingtao, Hou Meiting, etc. Remote Sensing Monitoring and Impact Assessment of Drought in Sichuan Based on Temperature Vegetation Dryness Index [J]. Transactions of the CSAE, 2007(09): 141-146, 294.
[19] Du Lingtong, Hou Jing, Hu Yue, etc. Characteristics of drought change in Ningxia from 2000 to 2010 based on remote sensing temperature vegetation drought index [J]. Transactions of the CSAE, 2015, 31(14): 209-216.
[20] Sha Sha, Guo Ni, Li Yaohui, etc. Applicability of Temperature Vegetation Dryness Index (TVDI) in Soil Moisture Monitoring in Eastern Gansu Province [J]. Journal of Desert Research, 2017, 37(01): 132-139.
[21] Ran Qiong, Zhang Zengxiang, Zhang Guoping, etc. DEM correction of national soil moisture inversion of temperature vegetation drought index [J]. Science of Soil and Water Conservation, 2005 (02): 32-36, 50.
[22] Zhao Jiepeng, Zhang Xianfeng, Liao Chunhua, etc. Remote Sensing Inversion Model of Soil Moisture in Large Arid Area Based on TVDI [J]. Remote Sensing Technology and Application, 2011, 26(06): 742-750.
[23] SAVITZKY A, GOLAY M J E. Smoothing and Differentiation of Data by Simplified Least Squares Procedures[J]. Analytical Chemistry. 1964, 36 (8): 1627-1639.
[24] Sha Sha, Guo Ni, Li Yaohui, etc. Application Status and Problems of Temperature and Vegetation Drought Index TVDI in China [J]. Journal of Arid Meteorology, 2014, 32(01): 128-134.
Data citation
Feng KT, Zhang YN, Tian DY, Kang JF. A dataset of monthly temperature vegetation dryness index along the China-Pakistan Economic Corridor from 2000 – 2017. Science Data Bank, 2018. (August 1, 2018). DOI: 10.11922/sciencedb.640.
Article and author information
How to cite this article
Feng KT, Zhang YN, Tian DY, Kang JF. A dataset of monthly temperature vegetation dryness index along the China-Pakistan Economic Corridor from 2000 – 2017. China Scientific Data 4 (2019). (July 18, 2019). DOI: 10.11922/csdata.2018.0051.zh.
Feng Keting
Mainly responsible for the work: program design, data processing and analysis.
Male, a native of Haiyuan County, Ningxia Hui Autonomous Region, Ph.D., engineer, research orientation for geoscience big data applications.
Zhang Yaonan
Mainly responsible for the work: overall program guidance.
Male, a native of Tianshui City, Gansu Province, Ph.D., researcher, research orientation for geoscience big data.
Tian Deyu
Mainly responsible for the work: data preprocessing.
Male, a native of Sizi Wangqi, Inner Mongolia Autonomous Region, master's degree, research orientation for remote sensing in cold and dry areas.
Kang Jianfang
Mainly responsible for the work: MODIS data collection.
Female, a native of Qinan, Gansu Province, master's degree, engineer, research orientation for the big data application in the cold and dry areas.
National R&D Infrastructure and Facility Development Program of China “National Special Environment and Function of Observation and Research Station Shared Service Platform” (Y719H71006); Information Program of the Chinese Academy of Sciences “Construction and Application of ‘Technology Cloud’ in Cold and Arid Regions Environment Evolution” (XXH13506).
Publication records
Published: July 22, 2019 ( VersionsEN1
Updated: July 22, 2019 ( VersionsEN2
Released: Sept. 25, 2018 ( VersionsZH2
Published: July 22, 2019 ( VersionsZH3
Updated: July 22, 2019 ( VersionsZH4