CC BY
90
Section 9. Agricultural sciences
Gafurov Zafar Asrarjonovich, researcher, International Water Management Institute E-mail: z.gafurov@cgiar.org Eltazarov Sarvarbek Baxtiyor o'g'li, consultant, International Water Management Institute E-mail: sarvarbek.eltazarov@gmail.com Akhmedova Tamara Abdurakhimovna, director, scientific research, Hydrometeorological Institute (NIGMI) E-mail: t.akhmedova@rambler.ru
CROP CLASSIFICATION IN KARSHI STEPPE USING REMOTE SENSING INFORMATION AND GOOGLE EARTH ENGINE TOOL
Abstract: Land use and land cover data are essential for water resource management, water supply planning, flood control and waste water treatment James et.al. [2] land use and crop pattern change information is the basis of better use of land and water resources. The Karshi steppe is one of the areas that is experiencing water management issues in Uzbekistan, since the economy of the region is highly dependent on agricultural fields which are irrigated from water diverted from Amudarya River that is located on the lower altitude from the irrigated area. Therefore it is a necessity to carefully study and understand the current situation and develop models for sustainable water management in the area. This study aims to learn temporal changes of main crops using satellite data for mapping crop pattern change analysis for this region. The output will be served as a basis for for further water management aspects. Crop classification using Google earth engine platform and Normalized Difference Vegetation Index (NDVI) Tangjaitrong et.al. [6] methodology was performed for the years of 1987, 2000, 2016. The overall accuracy of the classification is 81% NDVI approach.
Keywords: Karshi Steppe, NDVI, Classification, Landsat, Google earth engine.
Introduction Central Asia started during the Soviet time. After about 1980 s,
Water resources in Central Asia are very much impor- the scarcity ofwater and energy were clearly visible in the area, tant for many sectors. With the increase of Central Asian also influenced to soil, and ground water quality as well as quan-demography and potential to climate change, we as human tity has changed over time. Taking into account all above men-being should more actively act to ensure the life for tomor- tioned problems we obviously should look back and understand row's generation in this region. When we look back into the how the changes over past periods has come out in that area and second decade of 20th century, we can observe huge develop- learn crop change dynamics. The change dynamics illustrated ments of deserts into irrigated lands for the purpose of increas- in this paper will help us to better understand what the size of ing crop production in this area. The Karshi steppe is one of the study area was in fact in the past and how the dynamics de-the main examples to this point, which has already tripled veloped with time. It is of advantage to have change dynamics the irrigated area in size as a result of massive developments analysis in order to make proper decisions that should mitigate of deserts during Soviet Union period. The karshi steppe was possible changes of study area for the future if further irrigation about 100 thousand hectare in size in 1970 s but now it has extension/reduction is required. Considering all above men-turned into about 350 thousands of hectares. It has started to tioned issues, this study was carried out where the outcome can increase its size after the extension actions of cotton fields in be useful information for several sectors dealing with similar
CROP CLASSIFICATION IN KARSHI STEPPE USING REMOTE SENSING INFORMATION AND GOOGLE EARTH ENGINE TOOL
cases in another part of Central Asian countries in order to protect the nature and establish better water recourse management. Objective
The objective of this study is to conduct crop classification based on vegetation development curve over time using multiple images in a year. The innovative modern instrument so called - Google earth engine were elaborated for the entire
process of this study.
Go gte Earth Engine sh«* sIjc« ar.d djissasi
Study area
The study was carried out in Karshi Steppe south of Uzbekistan and western slopes of Pamir Mountains. The irrigation of an area of interest is fed by Talimardjan reservoir. Water is pumped up to the reservoir with special pumps through Karshi Magisterial canal and filled during winter time and used its water in summer.
Sen;)*
DMt At»»
13. Pj pw _ CijttonlWtwi LKarehl
■ Own« (1)
■ us?r¡Jfanar»ft/itc„,
> lAMO - IWMI • 10 iT_CA_lirlg fell !-10[»S..UA it?
m snmUt.
a M Wàiir h BWenLV« lljWUQlt. h í Landsil Si
í m IL
ti 1» 11 it It
■,Hr ¡lígtMifi - it .T-np-lT-llcit L^.-.
d ] T ri-B(H í ['"■■! ■■ I>1 il ¡ ' , 1 1 ; ■ ; " J ; P^^Hf S—Jílfllí];
rrcc LH1ÜLI'. J j I. ú for If JO
vir Dai liw.li - CÍ. ["Jim; var dft/llt^li * c*.It'.
óffl6*_lt • Mihwl'LM^t/lIt Vir 4»yiw_lí • «-!«■[*< UffiSWUaLUOWU
v.r - rr.l.v:
v„r J^ÍJ-l.LA - r-.l-
IMF- -
ÍHlMXe
- Ui* prinr(...J tû fit» fü thft ranvn]».
nuiUrttijjHjiiiiWLbmis )¿ .í oí J J Jiisi it.Lq«t>: ); vi':'.<.:::-coi ■■. 1 ;
W!i№¡ikW)t WHIUWVil! t¡
■' htïï L+ntilBÎ IJMRet f^r ¿ÜII6
vnr ndhiJíitlíe - 3fl;180_L6.n ■ ! : L:<c-3¡"r-:-c*( ; '
a«b<P3Da¡i
7 -
Figure 1. Interface of the developed model in Google Earth Engine Platform
Methods and materials
In the last three decades, the remote sensing technologies and methods have dramatically improved and available data for observations are increased Rogan & Chen [4]. Using remote sensing data has demonstrated and proved usefulness and effectiveness on detecting and mapping of agricultural, residential and urban areas Selguk et .al. [5]. (Figure 1) shows the interface of the developed program and a study area location, which is designed to identify cotton and winter wheat from remote sensing images, and can also be used for improved, accelerated interpretation and classification of data. The program is developed in the Java programming
language, which is customized to use the information of the Google Earth Engine database. Google Earth Engine is a new web platform for analyzing satellite images, which helps to improve the efficiency of data processing. It allows to analyze satellite images in the cloud system of the server, which allows to minimize several stages of processing satellite images and to deduce the final results for making decisions more quickly, saving labor and time resources. Landsat images are used in this model, which are recognized as the optimal source of information Rouse et.al. [3], due to open online access and an accessible time series of images with a spatial resolution of 30 meters are sufficient for determining crops. To classify
the crop, a classification method based on expert knowledge and crop phenology using NDVI was applied Hord et.al. [1]. NDVI can take values from -1 to 1. Most of the vegetation ranges from 0.1 to 0.8. In places with a large NDVI value (close to 1), usually thicker vegetation. In the places of a picture with a low value, especially less than zero, there is usually no vegetation. NDVI helps to minimize changing lighting conditions, viewing aspects, atmospheric conditions and inclination deviations. Classification of the crop using this program is based on existing knowledge of plant growth in a given territory and
phenology. NDVI layers are created for the given source images and the type of culture is determined in accordance with the change in the profile of this index during the year. In this case, the Google Earth Engine develops a decision tree and Java code for each year to identify land use in the field of research. In the proposed script, the model is developed based on the experimental plot of the Karshi region for the period 1987, 2000 and 2016. It should be noted that, in accordance with the goals and objectives of the study, time period and location of study area can be easily changed.
Figure 2. Changes of cotton and wheat over time, cotton wheat
Conclusion
Result shows crop pattern change between 1987, 2000 and 2016 land cover types of Karshi steppe for different time spans. The results gave land cover values for different years and for different crops. Through this study it was possible to demonstrate that one can easily demonstrate crop type dynamics
in agricultural fields using innovative tools and remote sensing information. The illustrations of crop type change in different years in this study can be very important in demonstrating the Karshi regions agricultural areas including different sectors and at different level. Outputs can be used for further application to find out Evapotranspiration (ET) of individual crops and etc.
References:
1. Hord R. M. Digital Image Processing of Remotely Sensed Data. - New York: Academic Press. 1982.
2. James R. A. Land Use And Land Cover Classification System. URL: http://landcover.usgs.gov/pdf/anderson.pdf
3. Rouse J. W., Haas R. H., Schell J. A., Deering D. W., and Harlan J. C., 1974. Monitoring the vernal advancement of retrogradation of natural vegetation. NASA/GSFC, Type III, Final Re-port (1974),- P. 371. Greenbelt, MD.
4. Rogan J., & Chen D. Remote sensing technology for mapping and monitoring land-cover and land-use change. Progress in planning, 61(4),2004. - P. 301-325.
5. Selguk R., Nisanci R., Uzun B., Yalcin A., Inan H., & Yomralioglu T. Monitoring land-use changes by GIS and remote sensing techniques: Case Study of Trabzon. In Proceedings of 2nd FIG Regional Conference, - Morocco 2003.- P. 1-11.
6. Tangjaitrong, Supichai. Enviremental Remote Sensing Courseware: Image Classification. URL: http://www.sc.chula.ac.th/ courseware/2309507/Lecture/remote18.htm (Last cited June 2013).
