Spatial Drought Assessment Using Remote Sensing and GIS techniques in Northwest region of Liaoning, China FUJUN SUN, MENG-LUNG LIN, CHENG-HWANG PERNG, QIUBING WANG, YI-CHIANG SHIU & CHIUNG-HSU LIU Department of Tourism Aletheia University No.32, Chen-Li St., Tamsui District, New Taipei City 25103, Taiwan mllin1976@mail.au.edu.tw http://mllin1976.epage.au.edu.tw Abstract: - In this study, we used the time series images of spectral reflectance from TERRA satellite to compute normalized difference vegetation index (NDVI) and normalized difference moisture index (NDMI) in the Chaoyang City, western Liaoning, north-eastern China. Time-series data of NDVI and NDMI derived from satellite images of the Moderate Resolution Imaging Spectroradiometer (MODIS) were used to show the growing curve of corn field and the impact of precipitation. The study assessed the spatial impact of drought on the time-series data of NDVI and NDMI in western Liaoning, north-eastern China. In addition, the standard precipitation index was used to detect metrological drought events. The results help us to compare the drought events detected from precipitation and satellite-derived images. Thus, quick spatial drought assessment techniques can be improved to effectively detect drought events spatially and temporally. Key-Words: - Drought assessment, Standard precipitation index, Remote sensing, MODIS 1 Introduction Drought monitoring assessment using satellite data has been an important issue in environmental monitoring and assessment [1-5]. Food security and safety have also been critical issues under the increasing population of the whole world. Therefore, developing quick drought monitoring and assessment using remotely sensed imagery to identify regional impact of drought for cropping systems is important. Satellite imagery was comprehensively applied to monitor and assess vegetation dynamics, drought condition, land surface moisture, cropping system, desertification [6-8]. Drought indices derived from satellite imagery have been widely used to identify spatial extents of drought [9-13]. The indices are useful for detection and monitoring large area vegetation stress resulted from drought or soil oversaturation following flooding and excessive rains. Another study was designed to use NOAA-AVHRR imagery to compute normalized difference vegetation index (NDVI) and vegetation condition index (VCI) indices to correlate precipitation data in the northwest of Iran [1]. They concluded that NOAA-AVHRR derived NDVI well reflects precipitation fluctuations in the study area and is useful for drought risk management. The NOAA-AVHRR data were applied to compute satellite derived drought indices, including the NDVI, Anomaly of Normalized Difference Vegetation Index (NDVIA), Standardized Vegetation Index (SVI), Vegetation Condition Index (VCI), Land Surface Temperature (LST) and NDVI (LST/NDVI), the Vegetation Health Index (VH), and the Drought Severity Index (DSI) [14]. Therefore, drought indices derived from satellite imagery are helpful for detecting, monitoring, and assessing drought and its spatial information at a regional scale. Remote sensing techniques have been applied over a number of regions worldwide to monitor vegetation and drought conditions using satellite derived indices, such as normalized difference vegetation index (NDVI) and normalized difference moisture index (NDMI) [1-4]. Therefore, we used time-series data of NDVI and NDMI to assess vegetation condition during the period between May and September in the year 2009. These drought related indices have been used to assess drought condition. The use of the local G statistic to define clusters may help to filter out spatial dependence or provide evidence for the occurrence of outliers [6-7]. The local G statistic is computed for SVI and SMI to identify drought affected area as a spatial filter. In this study, we used satellite derived drought indices to assess spatial drought-affected areas. The ISBN: 978-1-61804-175-3 355
drought indices adopted in this study are NDVI, NDMI, SVI and SMI. Further, the local G statistic method was applied to find out the drought-affected areas. Standard precipitation index (SPI) was also applied to observe long term trend of drought in the study area. The study was carried out in the western Liaoning, north-eastern China. The main land cover types are corn field and forest, which cover more than 73% of the study area s 19,718 km2 (Fig. 1 & Fig. 2). Corn field is the dominant cropping system in the study area. The growing season of corn field is from May to September. 2 Study Area Fig. 1 The Map of Chaoyang City and weather stations in the western Liaoning, north-eastern China. Fig. 2 Study area is restricted to the corn field and forest zones of Chaoyang, western Liaoning, north-eastern China. ISBN: 978-1-61804-175-3 356
3 Materials and Methods Standard precipitation index (SPI) was proposed by McKee et al. and widely applied to detect drought events. The drought category of SPI value is listed as below (Table 1). Table 1. Drought category of SPI value (McKee et al., 1993) SPI value Drought Category Time in Category 0 to -0.99 Mild drought 34.1% -1.00 to -1.49 Moderate drought 9.2% -1.50 to -1.99 Severe drought 4.4% -2.00 Extreme drought 2.3% This study tried to map drought affected areas using MODIS data on the TERRA satellite. The NDVI and NDMI data (2001-2008 used as the base period) analyzed in this study were 8-day composites for the study area located at western Liaoning, excluding cloud pixels. These data were used to calculate SVI to assess vegetation conditions and SMI to assess soil moisture conditions. NDVI and NDMI are utilized: NDVI NIR RED = (1) NIR + RED where RED and NIR are the surface reflectance in the visible (620-670 nm) and Near-Infrared (NIR) (841-876 nm) regions of the electromagnetic spectrum of MODIS on the TERRA satellite, respectively, for pixel I during season j for year k. NDMI NIR MIR = (2) NIR + MIR where NIR and SWIR is the surface reflectance in the MIR (2105 2155 nm) region of the electromagnetic spectrum of MODIS on the TERRA satellite, respectively, for pixel i during season j for year k. SVI ( NDVI NDVI ) ij = (3) σndviij NDVIij NDVI σndviij where where,, and are the multiyear average NDVI for pixel i in 8-day j, 8-day NDVI for pixel i in week j for year k, standard deviation of NDVI for pixel i in 8-day j respectively. SMI where where ( NDMI NDMI ) = (4) σndmi NDMIij NDMI, ij ij σndmiij, and are the multiyear average NDMI for pixel i in 8-day j, 8- day NDMI for pixel i in week j for year k, standard deviation of NDMI for pixel i in 8-day j respectively. A key advantage of using local G statistic (G i * ) method is that G i * is capable of indicating the spatial clusters of interested geographic phenomena [15]. The method was computed for SVI and SMI is to examine the drought conditions of the western Liaoning, northeastern China. The equation for Gi* was calculated for each pixel using the following equation: * j Wij ( d) x j Wi x Gi = (5) * * s[ W ( n W ) /( n 1) ] 1/ 2 i i where W * i is the count of the pixels within a distance, d, of the central pixel i, d defines the size of the kernel by the number of pixels from the central pixel, i, x is the mean, s is the standard deviation, and n is the total number of pixels in the image. The resultant ranges of probability of values for the statistic are mapped to illustrate the spatial distribution at P < 0.05 for negative values of G * i. 4 Results and Discussions This study is successful in delineating detailed spatial information concerning the drought affected areas of the 2009 drought event in the western Liaoning, China. The spatial distributions of NDVI and NDMI indicate that the vegetation and moisture conditions of the drought event (Julian Day 233~240) in the year 2009. The drought indices, SVI and SMI, derived from MODIS images enhance drought identification mapping spatially. The results show that the indices significantly helped quick drought monitoring and assessment. The use of the local Getis statistic (G i * ) provides insights on the spatial relationships of drought affected area. Specially, the location of significant * G i values identified areas where the drought affected areas occur and are spatially clustered (Figure 3). These clusters can be viewed as spatial filtering of drought affected area. Therefore, this spatial information may then be used to delineate a drought map and help governments to identify high priority areas of local water resource allocation during the drought period. The time series of SPI for Chaoyang and Yebaishou, western Liaoning, are shown in Fig.4 and 5 for I = 3, 6 and 9 months for the period 1959-2010. When the time periods are small (3 and 6 months), the SPI moves frequently above and below zero (Fig. 4 & 5). ISBN: 978-1-61804-175-3 357
Fig. 3. Delineation of spatial filtering for drought event showing drought affected areas where A indicates the area identified by no drought indices (SVI, and SMI); B indicates the area identified by one of the drought indices; C indicates the area identified by all drought indices. Fig. 4. SPI time series calculated for Chaoyang, western Liaoning, 1959-2010 using time scales of 3, 6 and 9 months. ISBN: 978-1-61804-175-3 358
Fig. 5. SPI time series calculated for Yebaishou, western Liaoning, 1959-2010 using time scales of 3, 6 and 9 months. 5 Conclusion A GIS-based geocomputing technique of spatial autocorrelation is used for detecting the spatial extent of drought affected area. There is good spatial agreement revealed between the different satellite-derived drought indices (SVI, and SMI). The observational results demonstrate that the drought indices derived from MODIS images are significant help in drought assessment spatially. It can be seen that drought affected areas identified by the GIS-based geocomputing techniques provide useful spatial information as spatial filtering for drought assessment. Furthermore, the SPI has adopted to assess the metrological drought events in the study area. The results of metrological drought events can be used to compare with the satellitederived drought indices to improve the spatial drought assessment techniques. The results demonstrate the importance of the spatial filtering approach (using the local G-statistic) for drought assessment to provide useful spatial information. However, use of the local G statistic for spatial filtering is only the first step for identifying drought affected area. The actual drought affected areas need to be identified and confirmed by field investigations. In conclusion, this novel approach is useful for quick large scale geographical monitoring and assessment of drought affected area. References: [1] P. R. Bajgiran, A. A. Darvishsefat, A. Khalili, & M. F. Makhdoum, Using Avhrr-Based Vegetation Indices for Drought Monitoring in the Northwest of Iran, Journal of Arid Environments, Vol.72, 2008, pp. 1086-1096. [2] C. Bhuiyan, R. P. Singh, & F. N. Kogan, Monitoring Drought Dynamics in the Aravalli Region (India) Using Different Indices Based on Ground and Remote Sensing Data, International Journal of Applied Earth Observation and Geoinformation, Vol.8, No.4, 2006, pp. 289-302. [3] H. C. Claudio, Y. Cheng, D. A. Fuentes, J. A. Gamon, H. Luo, W. Oechel, H. L. Qiu, A. F. Rahman, & D. A. Sims, Monitoring Drought Effects on Vegetation Water Content and Fluxes in Chaparral with the 970 Nm Water Band Index, Remote Sensing of Environment, Vol.103, No.3, 2006, pp. 304-311. [4] S. M. Herrmann, A. Anyamba, & C. J. Tucker, Recent Trends in Vegetation Dynamics in the African Sahel and Their Relationship to Climate, ISBN: 978-1-61804-175-3 359
Global Environmental Change Part A, Vol.15, No.4, 2005, pp. 394-404. [5] L. S. Unganai & F. N. Kogan, Drought Monitoring and Corn Yield Estimation in Southern Africa from Avhrr Data, Remote Sensing of Environment, Vol.63, No.3, 1998, pp. 219-232. [6] M. L. Lin & C. W. Chen, Application of Fuzzy Models for the Monitoring of Ecologically Sensitive Ecosystems in a Dynamic Semi-Arid Landscape from Satellite Imagery, Engineering Computations: International Journal for Computer-Aided Engineering and Software, Vol.27, No.1, 2010, pp. 5-19. [7] M. L. Lin, C. W. Chen, Q. B. Wang, Y. Cao, J. Y. Shih, Y. T. Lee, C. Y. Chen, & S. Wang, Fuzzy Model-Based Assessment and Monitoring of Desertification Using Modis Satellite Imagery, Engineering Computations: International Journal for Computer-Aided Engineering and Software, Vol.26, No.7, 2009, pp. 745-760. [8] S. M. Quiring & S. Ganesh, Evaluating the Utility of the Vegetation Condition Index (Vci) for Monitoring Meteorological Drought in Texas, Agricultural and Forest Meteorology, Vol.150, No.3, 2010, pp. 330-339. [9] L. Vergni & F. Todisco, Spatio-Temporal Variability of Precipitation, Temperature and Agricultural Drought Indices in Central Italy, Agricultural and Forest Meteorology, Vol.151, No.3, 2011, pp. 301-313. [10] M. L. Lin, Q. Wang, F. Sun, T. H. Chu, & Y. S. Shiu, Quick Spatial Assessment of Drought Information Derived from Modis Imagery Using Amplitude Analysis, World Academy of Science, Engineering and Technology, Vol.67, 2010, pp. 410-414. [11] W. T. Liu & R. I. N. Juárez, Enso Drought Onset Prediction in Northeast Brazil Using Ndvi International Journal of Remote Sensing, Vol.22, No.17, 2001, pp. 3483-3501. [12] A. Anyamba, C. J. Tucker, & J. R. Eastman, Ndvi Anomaly Patterns over Africa During the 1997/98 Enso Warm Event, International Journal of Remote Sensing, Vol.22, No.10, 2001, pp. 1847-1859. [13] F. Kogan & J. Sullivan, Development of Global Drought-Watch System Using Noaa/Avhrr Data, Advances in Space Research, Vol.13, No.5, 1993, pp. 219-222. [14] Y. Bayarjargal, A. Karnieli, M. Bayasgalan, S. Khudulmur, C. Gandush, & C. J. Tucker, A Comparative Study of Noaa-Avhrr Derived Drought Indices Using Change Vector Analysis, Remote Sensing of Environment, Vol.105, No.1, 2006, pp. 9-22. [15] D. Z. Sui & P. J. Hugill, A Gis-Based Spatial Analysis on Neighborhood Effects and Voter Turn-Out:: A Case Study in College Station, Texas, Political Geography, Vol.21, No.2, 2002, pp. 159-173. ISBN: 978-1-61804-175-3 360