Estimating soil organic carbon levels in cultivated soils from satellite image using parametric and data-driven methods DOI
Muhammed Halil Koparan, Hossein Moradi Rekabdarkolaee, Kunal Sood

et al.

International Journal of Remote Sensing, Journal Year: 2022, Volume and Issue: 43(9), P. 3429 - 3449

Published: May 3, 2022

Soil organic carbon (SOC) is one of the key soil components for cultivated soils. SOC regularly monitored and mapped to improve quality, health, productivity soil. However, traditional SOC-level monitoring expensive land managers farmers. Estimating using satellite imagery provides an easy, efficient, cost-effective way monitor surface levels. The objective this study was estimate distribution in selected soils Major Land Resource Areas (MLRA), 102A (Rolling Till Plain, Brookings County, SD), 103 (Central Iowa Minnesota Prairies, Lac qui Parle MN), with different resolutions (Landsat 8 PlanetScope). dominant area are Haplustolls, Calciustolls, Endoaquolls, which formed silty sediments, local alluvium, till. Landsat PlanetScope spectral bands were used develop prediction models. Parametric data-driven methods employed predict SOC. Multiple linear regression Linear Spatial Mixed Model (LSMM) on data. In addition parametric models, Regression Trees Random Forest also both results showed that reduced LSMM provided lowest RMSE, 0.401 0.367 PlanetScope, respectively. Furthermore, random forest has highest RPD RPIQ (RPD 2.67 2.49) 2.85 3.7). all cases, models obtained from better than those 8.

Language: Английский

A New Spatial Downscaling Method for Long-Term AVHRR NDVI by Multiscale Residual Convolutional Neural Network DOI Creative Commons
Mengmeng Sun, Xiang Zhao, Jiacheng Zhao

et al.

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Journal Year: 2024, Volume and Issue: 17, P. 7068 - 7088

Published: Jan. 1, 2024

Monitoring vegetation dynamics is essential for ecological processes, environmental changes, and natural resource protection. Fine-scale representation of indices necessary regions with complex topography high diversity species. However, the advanced very-high-resolution radiometer (AVHRR), which covers an extensive time range temporal resolution, does not provide normalized difference index (NDVI) data sufficient spatial resolutions a detailed analysis changes. The Moderate Resolution Imaging Spectroradiometer (MODIS), has higher only been limited to last few decades. To deal these issues, we propose Multi-scale Residual Convolutional Neural Net-work (MRCNN) that utilizes multi-scale structure residual convolutional neural network combine MODIS NDVI AVHRR data. MRCNN algorithm improved Mean Absolute Error (MAE) Root Squared (RMSE) by 0.026 0.032, respectively, resulting in 64.38% improvement MAE 62.79% RMSE compared NDVI. It also increased Peak-Signal-to-Noise Ratio (PSNR) 28.5% Structural Similarity (SSIM) 16.2%. method accurately captures actual state consistently tracks changing trends index. exact terrain diverse areas. This enhances resolution significantly improves accuracy monitoring nationwide changes over 30 years. findings establish solid scientific foundation implementing conservation measures promoting sustainable growth.

Language: Английский

Citations

2
Predicting Dynamics of Soil Salinity and Sodicity Using Remote Sensing Techniques: A Landscape-Scale Assessment in the Northeastern Egypt DOI Open Access
Ahmed S. Abuzaid,

Mostafa S. El-Komy,

Mohamed S. Shokr

et al.

Sustainability, Journal Year: 2023, Volume and Issue: 15(12), P. 9440 - 9440

Published: June 12, 2023

Traditional mapping of salt affected soils (SAS) is very costly and cannot precisely depict the space–time dynamics soil salts over landscapes. Therefore, we tested capacity Landsat 8 Operational Land Imager (OLI) data to retrieve salinity sodicity during wet dry seasons in an arid landscape. Seventy geo-referenced samples (0–30 cm) were collected March (wet period) September be analyzed for pH, electrical conductivity (EC), exchangeable sodium percentage (ESP). Using 70% band reflectance data, stepwise linear regression models constructed estimate EC, ESP. The validated using remaining 30% terms determination coefficient (R2) residual prediction deviation (RPD). Results revealed weak variability while EC ESP had large variabilities. three indicators (pH, ESP) increased from period. During two seasons, OLI bands associations with near-infrared (NIR) could effectively discriminate levels. predictive period developed NIR band, achieving adequate outcomes (an R2 0.65 0.61 RPD 1.44 1.43, respectively). In period, best-fitted deep blue bands, yielding 0.59 0.60 1.49 1.50, respectively. SAS covered 50% study area which 14 36% saline saline-sodic soils, extent up 59% including (12%) (47%). Our findings would facilitate precise, rapid, cost-effective monitoring areas.

Language: Английский

Citations

5
Enhancement and analysis of hyperspectral satellite images for Soil Study and Behavior DOI
Varun Malik, Ruchi Mittal, Amandeep Kaur

et al.

Multimedia Tools and Applications, Journal Year: 2023, Volume and Issue: 83(11), P. 33879 - 33902

Published: Sept. 26, 2023

Language: Английский

Citations

5
Soil parent material prediction through satellite multispectral analysis on a regional scale at the Western Paulista Plateau, Brazil DOI
Fellipe Alcântara de Oliveira Mello, Henrique Bellinaso, Danilo César de Mello

et al.

Geoderma Regional, Journal Year: 2021, Volume and Issue: 26, P. e00412 - e00412

Published: June 17, 2021

Language: Английский

Citations

12
Estimating soil organic carbon levels in cultivated soils from satellite image using parametric and data-driven methods DOI
Muhammed Halil Koparan, Hossein Moradi Rekabdarkolaee, Kunal Sood

et al.

International Journal of Remote Sensing, Journal Year: 2022, Volume and Issue: 43(9), P. 3429 - 3449

Published: May 3, 2022

Soil organic carbon (SOC) is one of the key soil components for cultivated soils. SOC regularly monitored and mapped to improve quality, health, productivity soil. However, traditional SOC-level monitoring expensive land managers farmers. Estimating using satellite imagery provides an easy, efficient, cost-effective way monitor surface levels. The objective this study was estimate distribution in selected soils Major Land Resource Areas (MLRA), 102A (Rolling Till Plain, Brookings County, SD), 103 (Central Iowa Minnesota Prairies, Lac qui Parle MN), with different resolutions (Landsat 8 PlanetScope). dominant area are Haplustolls, Calciustolls, Endoaquolls, which formed silty sediments, local alluvium, till. Landsat PlanetScope spectral bands were used develop prediction models. Parametric data-driven methods employed predict SOC. Multiple linear regression Linear Spatial Mixed Model (LSMM) on data. In addition parametric models, Regression Trees Random Forest also both results showed that reduced LSMM provided lowest RMSE, 0.401 0.367 PlanetScope, respectively. Furthermore, random forest has highest RPD RPIQ (RPD 2.67 2.49) 2.85 3.7). all cases, models obtained from better than those 8.

Language: Английский

Citations

7