CC BY
17
Arifjanov Aybek Mukhamedjanovich, doctorof technical sciences, professor, Head of the Department of Hydraulics and Hydroinformation, Tashkent Institute of Irrigation and Agricultural Mechanization Engineers
E-mail: obi-life@mail.ru
Akmalov Shamshodbek Bakhtiyarovich, doctoral, researcher of Tashkent Institute of Irrigation and Agricultural Mechanization Engineers, Department of Hydrology and Hydrogeology E-mail: shamshodbekjon@mail.ru
Samiev Luqmon Nayimovich, Senior researcher, Department of Hydraulics and Hydroinformation,
Tashkent Institute of Irrigation and Agricultural Mechanization Engineers.
E-mail: luqmonsamiev@mail.ru
Tursunoy Ubaydullaevna Apakxo'jaeva, assistant, Department of Hydraulics and Hydroinformation,
Tashkent Institute of Irrigation and Agricultural Mechanization Engineers
CHOOSING AN OPTIMAL METHOD OF WATER EXTRACTION
FOR ARID REGIONS IN THE CASE OF BESHBULAK AND YANGIOBOD VILLAGES (SYRDARYA PROVICE, UZBEKISTAN)
Abstract: This article is about choosing an optimal method of water extraction in arid areas. Water is a very important asset in arid areas. In order to control water allocation, spetialists need fast and accurate information about water obj ects and irrigation sets. Remote sensing data are a useful tool for this. Many researchers have used these data with satisfying results. With remote sensing data a lot of methods and ways to extract water objects have been created. Choosing a more effective method has been an important goal for researches. In our paper a lot of water extraction methods are compared in the Beshbulak and the Yangiobod region. According to the results, the most accurate method of water extraction for arid zones is chosen. This method is water objects extracting with the NIR2 (near infrared) band of VHR (very high resolution) data.
Keywords: NDWI (normalized difference ofwater index), RS (remote sensing), Water management, object based image analysis (OBIA), eCognition, Arid regions.
1. Introduction water management has increased sharply. RS systems
Water extraction is very important for water ac- have been developing and launching different satellites
counting system. 10 years ago the VHR remote sensor with various possibilities. Water objects and irrigation
was created. It has a connection with IKONOS and sets of the country in small areas have been enlarged
QuickBird satellites' launches. Afterwards, more devel- by the spectral resolution of satellite sensors, enabling
oped satellites were launched [6]. The development in analysis of irrigation sets on land. Main objective of
RS was the reason to use this information more widely this research to choose optimal index of water extrac-
in water management. Since 2000 the usage of RS in tion for arid zone.
2. Method and materials
Case of study. The research areas are Beshbulak and Yangiobod villages which are situated in Syr-Darya Province, Uzbekistan.
Field data collection. It is obligatory to point out the appropriate method among them with the best results to classify the water objects in the province at present and in the future. It guarantees the creation of a useful method for specialists to observe the water objects. Thus, it is necessary to compare the water objects in WV2 trapezium in field experiments for validation. To realize this comparison, one needs to collect data about features of water objects. Taking this need into account, we have taken the information about water objects' location; their size and width in the WV2 trapezium.
Satellite data collection. From RS data for water extraction analysis used WorldView-2 sattelite data. For irrigation set extraction analysis we have chosen two zones. The first zone is well suplied with water WV2-2011 (Beshbulak vilage) and the second one is desert zone which has some problems with water supply WV2-2012 (Yangiobod village) (Tab. 1):
Table 1. - List of collected WordView-2 data and their technical characteristics (Source: digitalglobe.com 2014)
Collected data (WorldView-2 (Q3-09))
ID Year Month Date Latitude Longitude Cloud cover
WV2-2011 (Beshbulak village) 2011 07 21 40.501 69.05 0
WV2-2012 (Yangiobod village) 2012 09 12 40.481 68.729 1
Segmentation analysis. After numbers of segmentation analyses, we have chosen the following optimal parameters for segmentation of WV images (Tab. 2):
Table 2.- Segmentation parameters for WV2 data
Satellite name Image layers Image Layer weights Scale parameter Shape Compactness
Pan 3
Costal 3
Blue 2
WV2-2011 (Beshbulak village) WV2-2012 (Yangiobod village) Green 2
Red 2 50.75 0.2 0.9
NIR 2
NIR2 2
Red edge 1
Yellow 1
Water extraction methods. For water extraction used different indexes which created in different years by different scientists and different bands of satellites. They are:
McFeeters' [5] index - NDWIM=
=(G-NIR)//(G+NIR) (1)
There: G-green band of satellite images, NIR-Near infrared band of satellite images.
Wolf et al., [7]; [8] index - NDWIW=
=(C-NIR2)//(C+NIR2) (2)
There: C - coastal band of satellite images, NIR2-2nd near infrared band of satellite images. Chen et al. [1] - NDWICh=(G-NIR2)/(G+NIR2) (3) Jawak & Luis [3] [4] index - NDWIJ1=
= (C-NIR1)/(C+NIR1) (4)
NDWI =(B-NIR1)/ (B+NIR1) (5)
J2
NDWI = (B-NIR2) / (B+NIR2) (6)
J3
There: B-blue band of satellite images.
Homogeneous objects nearby the river are larger in segmentation of the image, so we have set SP=250 for Quandtree based segmentation. After the segmentation, small segment objects (< = 256 p x l) and objects with NIR < = 120 have been classified into "watery objects", and the others into "other objects".
In the next step, watery objects were segmented with the algorithm "multiresolution segmentation region grow", the scale parameter has been increased until the creation of a homogeneous object and optimal parameter scale above had been chosen (Tab. 2). The reason of choosing a small scale parameter for the growth algorithm is the covering of water bodies with strong coastal vegetation in this area. And it becomes difficult to separate the coastal vegetation in segmentation. That is why
Table 3.- Classification value
we have chosen a small parameter. And now, the second step is the classification process.
Classification. In this step we used assign class and merge region algorithms. In analysis it is seen that high reflectance objects prevent the determination of water objects when using NDWI. For this reason, the first step must be the classification of these objects into "high reflectance". Most high reflectance objects are artificial areas. This situation we can see in a lot of similar analyses. In other words, in our analysis the most effective layer was coastal bands to determine "high reflectance" objects. We used DN indices of this band for determining open ground, ground ways, buildings and asphalt roads, as well. Consequently, coastal band is the most suitable band for classifications of asphalt ground roads and population centers (Tab. 3).
of urban areas for WV2 data
Satellite name IF THEN ELSE
Coastal
WV2-2011 (Beshbulak village) > = 417 Roads and urban areas Unclassified
WV2-2012 (Yangiobod village) > = 349
Water bodies and irrigation sets. The use of the above mentioned formulas for NDWI calculation in WV2 images analysis in different conditions has given positive results. We used "water object classification with WV2
Table 4.- Classification value
image analysis" from all these formulas and we evaluated the effectiveness of this formula (l)-(6). Besides, we used different bands of images for water extraction (Tab. 4).
of water objects for WV2 data.
Satellite name IF THEN ELSE
NIR NIR2 Pan RedEdge
WV2-2011 < = 260 < = 430 < = 278 < = 360 Water bodies and irrigation sets Unclassified
WV2-2012 < = 440 < = 420 < = 237 < = 420
Brightness NDVI NDWIw NDWICh NDWIJ3
WV2-2011 < = 292 < = 0,09 > = 0 > = 0 > = -0.1
WV2-2012 < = 325 < = 0,18 > = -0.1 > = -0.05 > = -0.18
NDWIj, NDWIJ2 NDWT
WV2-2011 > = 0,13 > = 0 > = 0.1
WV2-2012 > = 0,12 > = 0.2 > = -0.06
The accuracy of the results have been tested by classification accuracy
3. Results
Classification accuracy. For the classification accuracy checks were created in water objects shape file by using GIS and GPS in the research area (Fig.l).
Observations showed that the desert area consists of two water objects (lakes) and 19 irrigation sets. Only two of the investigated irrigation branches are
more than 10 meters width (main channels), the rest of the canals' width is 4-5 meters and a lot of other channels have a width of less than four meters. On the coastal area of the river we have found four water objects, one of them is a river that divides 21 irrigation sets. Four branches out of those are of10 meters width
(Fig. 1).
Figure 1. Created water sets and object map for Yangiobod and Beshbulak villages (source Akmalov)
Classification results. The smallest value of the near infrared band of the WV2 satellite gave us water objects. Because of this band we could extract large water obj ects and big water branches. In both analyses large water objects are fully determined. As Beshbulak village is located on the coastal area and as its main agricultural planting is rice growing, water full rice fields are also classified into water objects. That is why the map contains a lot of water objects.
The water objects were classified with different methods and evaluated accuracy of those methods. The conclusion was that NIR2 band is the most efficient in determining water branches in Yangiobod and Beshbulak villages of Sirdarya region then other indexes and layers. Investigations demonstrated that, NIR2 layer showed the highest accuracy. According to this formula we could extract 19 water objects from desert area and 26 water objects from the river area (Fig. 2).
Figure 2. NIR2 Water extraction analysis with using NIR2 band result for Yangiobod And Beshbulak villages
By the results of this analysis we can recommend the ing small water objects in Beshbulak village. It was impos-
optimal index for calculating water objects. The estab- sible to separate rice fields from water objects as they were
lished new and rich map for water sets and resources is of filled with water. Moreover, groundwater has also mixed
benefit for the region. We faced some difficulties in detect- with water objects as they come to the surface. This issue
could be observed in Dubovyk et al., [2] 's work. They classified water objects with the OBIA method and classified above mentioned waters as agricultural lands. In conclusion, they marked those areas as flood areas.
4. Conclusions
For analysing irrigation sets and controlling their exploitation period, reconstructing and calculating its water share, irrigation sets extraction methods were created with the help of WV2 data. It is a valuable information base for the region's agriculture. This information has very important consequences for the region.
Analysis brought the following general results:
- OBIA is the most reliable and advanced method for analyzing RS images, it gives the possibility of analyzing water objects conditions;
- The NIR2 band gives the best results in identifying water objects in Syr-Darya Province;
Our research work concerns a scientific analysis of Syr-Darya region's water resources and RS, preventing the following problems and introducing the following novelties.
Acknowledgements
The author is grateful to Erasmus Mundus Eur-asian-CEA scholarship, Lillel University Science and technology, Doctoral department SESAM and Laboratory TVES, head and personnel of departments of UFR "Geography and urban planning" and "Department of International affairs": Prof. Philippe Deboudt, Prof. Francois-Olivier Seys, Christine Vandenbosch, Mojca Maher Pirc, Veronique Level, Antoine Sugita, Klavdija, Emeline Marchesse and others, "Tashkent Institute of Irrigation and Melioration" in Uzbekistan head and personnel of departments of Hydromelioration: Prof. Mukhamadkhon Khamidov, Dr. Bakhodir Mirzaev, Dr. Toxir Sultanov and others, head of UzGi-droMet centre Dr. Bakhriddin Nishanov and others for providing financial and material assistance to carry out this research work, who supported with materials and finance during the writing of this article. The author is especially grateful to his supervisors Prof. Olivier Blan-pain and Dr. Eric Masson.
References:
1. Chen G., Hay G. J., Carvalho L. M.T. & Wulder M. A. Object-Based Change Detection. International Journal of Remote Sensing, 33 (14).- 2012.- P. 4434-4457.
2. Dubovyk O., Menz G., Conrad C., Thonfeld F. & Khamzina A. Object-Based Identification of Vegetation Cover Decline in Irrigated Agro-Ecosystems in Uzbekistan. Quaternary International, Hydrological and Ecological Responses to Climatic Change and to Land-use/land-cover changes in Central Asia, 311.- 2013.- P. 163-174.
3. Jawak S. D., & Luis A.J. A Spectral Index Ratio-Based Antarctic Land-Cover Mapping Using Hyperspatial 8-Band WorldView-2 Imagery. Polar Science, 7 (1).- 2013.- P. 18-38.
4. Jawak S. D., & Luis A. J. A Rapid Extraction of Water Body Features from Antarctic Coastal Oasis Using Very High-Resolution Satellite Remote Sensing Data. Aquatic Procedia, 4.- 2015.- P. 125-132.
5. Mc. Feeters S. K. The Use of the Normalized Difference Water Index (NDWI) in the Delineation of Open Water Features. International Journal of Remote Sensing, 17 (7).- 1996.- P. 1425-1432.
6. Navulur K., Pacifici F. & Baugh B. Trends in Optical Commercial Remote Sensing Industry [Industrial Profiles]. Geoscience and Remote Sensing Magazine, 1 (4).- 2013.- P. 57-64.
7. Wolf A. F. (2012). Using Worldview-2 Vis-NIR Multispectral Imagery to Support Land Mapping and Feature Extraction Using Normalized Difference Index Ratios. In SPIE Defense, Security, and Sensing. DOI: 10.1117/12.917717.
8. Wolf A. T., Natharius J. A., Danielson J. J., Ward B. S. & Pender J. K. International River Basins of the World. International Journal ofWater Resources Development, 15 (4).- 1999.- P. 387-427.
