Space monitoring and GIS for quantifying soil salinity and degradation of agricultural land in southern Kazakhstan (BR 10965172)

According to the technical assignment of the Customer in this program, the object and subject of research are saline agricultural lands of South Kazakhstan, in which the arable soil layer is at the level (0-30 cm). The scale of the required salinity maps is M 1: 300000. The relationship between the map scale and information saturation is determined by the minimum object that must be mapped (normative documentation: GOST R 51608).

Purpose of the study: The goal of the program is formulated by the Science Committee of the Republic of Kazakhstan as: “Creation of a web-based geoinformation service for operational monitoring of a quantitative assessment of the degree of soil salinity and degradation of agricultural land in the South of Kazakhstan based on data from remote sensing of the Earth. ”

Research methods: Expert interpretation, standard statistical methods of correlation analysis of remotely recorded parameters (NDVI, LST, etc.) and ground information. Methods for clustering and classifying satellite data.

Scope and implementation of the results: Salinization of irrigated arable land and degradation of agricultural lands in the South of Kazakhstan.

Current results

Methodological scheme of research:

  • The methodological scheme of the study includes the following main elements (figure):
  • Model A. Estimation of salinity based on spectral indices. As a result of this model, based on expert assessment, a map is formed with five levels of salinity, which are then converted into binary values ​​(there is salinity and there is no salinity).
  • Model B. A classification model where the inputs are radar images and the targets are binary salinity estimates generated by Model A.

Model A results

Decesion tree classification

Decesion tree classificationModel B results

Learning and applying flowchart of model B

Learning and applying flowchart of model B 

The obtained result indicates that model B based on the GPC algorithm shows good results: a recall  0.92 for class 0 (no salinity) and recall 0.86 for class 1 (there is salinity). The considered process of salinity classification can increase the efficiency of solving problems for automating  large areas salinity digitization.

Soil sampling routes

Soil sampling routes

Publications

  1. Mukhamediev, R. I., Popova, Y., Kuchin, Y., Zaitseva, E., Kalimoldayev, A., Levashenko V., Symagulov, A., Abdoldina F., Gopejenko V., Yakunin K., Muhamedijeva E., Yelis, M. Review of Artificial Intelligence and Machine Learning Technologies: Classification, Restrictions, Opportunities and Challenges ////Mathematics. – 2022. – V. 10. – №. 15. – P. 2552. https://doi.org/10.3390/math10152552
  2. Mukhamediev, R. I., Kuchin, Y., Amirgaliyev, Y., Yunicheva, N., Muhamedijeva, E. Estimation of Filtration Properties of Host Rocks in Sandstone-Type Uranium Deposits Using Machine Learning Methods //IEEE Access. – 2022. – Т. 10. – С. 18855-18872. (SCOPUS, Q1, процентиль 87, Web of Science Impact Factor=3,05) https://doi.org/10.1109/ACCESS.2022.3149625
  3. Mukhamediev R. I. et al. Review of Some Applications of Unmanned Aerial Vehicles Technology in the Resource-Rich Country //Applied Sciences. – 2021. – Т. 11. – №. 21. – С. 10171. https://doi.org/10.3390/app112110171 (CiteScore highest quartile = Q2, JCR – Q2, CiteScore =3.0, CiteScore highest percentile=71%, Web of Science Impact Factor==832) https://www.mdpi.com/2076-3417/11/21/10171

Conferences thesises

  1. Yan Kuchin, Jan Rabcan , A Symagulov , Ravil I. Mukhamediev , Bayangali Abdygalym, Nadiya Yunicheva and Elena Mukhamedieva. Calculation filtration coefficient using regression models//The 20th INTERNATIONAL SCIENTIFIC CONFERENCE INFORMATION TECHNOLOGIES AND MANAGEMENT 2022, April 21-22, 2022, ISMA University of Applied Science, Riga, Latvia.-C.27-28. https://www.dropbox.com/s/plnloy6tktyte61/11_ITM2022_Kuchin_Calculation%20filtration%20coefficient.pdf?dl=0
  2. Symagulov, Y.Kuchin, Jan Rabcan, Ye.Kulakova, I. Assanov , R. Mukhamediev. Pretrained Deep Neural Network Models for Image Change Detection//The 20th INTERNATIONAL SCIENTIFIC CONFERENCE INFORMATION TECHNOLOGIES AND MANAGEMENT 2022, April 21-22, 2022, ISMA University of Applied Science, Riga, Latvia.-C.20-21. https://www.dropbox.com/s/4xuh5ytjvm5x78u/08_ITM2022_Symagulov_Pretrained%20Deep%20Neural%20Network%20Models.pdf?dl=0
  3. Symagulov, I. Assanov, Y.Kuchin, Jan Rabcan, Ye.Kulakova, Bayangali Abdygalym. Video pre-processing for computer vision tasks using UAVs //The 20th INTERNATIONAL SCIENTIFIC CONFERENCE INFORMATION TECHNOLOGIES AND MANAGEMENT 2022, April 21-22, 2022, ISMA University of Applied Science, Riga, Latvia.-C.25-26. https://www.dropbox.com/s/isivvdkd7lo7fp5/10_ITM2022_Symagulov_Video%20pre-processing%20for%20computer%20vision.pdf?dl=0

Author’s certificates

Mukhamediev R.I., Amirgaliev E.N. Introduction to Machine Learning Tutorial. Certificate of information into the state register of rights to objects protected by copyright No. 28386 dated August 19, 2022.

(Мухамедиев Р.И., Амиргалиев Е.Н. Введение в машинное обучению. Учебникю.

Свидетельство о внесении сведений в государственный реестр прав на объекты охраняемые авторским правом № 28386 от 19 августа 2022 года. https://www.dropbox.com/s/xav2wuywqigt2m2/ML_Book_%D0%A1%D0%B2%D0%B8%D0%B4%D0%B5%D1%82%D0%B5%D0%BB%D1%8C%D1%81%D1%82%D0%B2%D0%BE.pdf?dl=0)

The results of 2021