USING NEURAL NETWORKS IN DETERMINING AND FORECASTING LAND SALINITY LEVELS

Qutlimurativ , Yusup, Orazbayev , Shaxmardan

Techscience.uz - техника фанлари долзарб масалалри · 2026-yil

Annotatsiya

This article presents the results of using deep learning methods to determine the level of land salinity and predict its dynamics in the Shortanbay massif of the Nukus district of the Republic of Karakalpakstan based on multispectral remote sensing data obtained during 2000–2025. In the study, multilayer perceptron (MLP), one-dimensional convolutional neural network (CNN), and long-term short-memory recurrent network (LSTM) models were built and tested based on eight feature vectors (B4, B3, B2, Salinity Index — SI, Normalized Salinity Index — NDSI, Brightness Index — BI, Normalized Vegetation Index — NDVI, Bare Soil Index — BSI) obtained from Landsat satellite images. The classification accuracy on the test set was 99.89% for the MLP and CNN models, and 98.88% for the LSTM model. The results confirm that neural networks can be used as an effective tool for remote monitoring of land and predicting its erosion.

Maqola ma’lumotlari
MualliflarQutlimurativ , Yusup, Orazbayev , Shaxmardan
JurnalTechscience.uz - техника фанлари долзарб масалалри
Nashr sanasi2026-05-14
Jild4
Son5
Betlar37-42
TilO‘zbek
DOI10.47390/ts-v4i5y2026n06

DOI: 10.47390/ts-v4i5y2026n06 · Maqolaning asl sahifasi

Kalit so‘zlar

salinity index, neural network, MLP, CNN, LSTM, remote sensing, land monitoring., sho‘rlanish indeksi, neyron tarmoqi, MLP, CNN, LSTM, masofadan zondlash, yerlar monitoringi.

Ilmiy soha

Texnika

Techscience.uz - техника фанлари долзарб масалалри jurnalidan boshqa maqolalar

Techscience.uz - техника фанлари долзарб масалалри — barcha maqolalar