A Multi-Source Data Fusion Method to Improve the Accuracy of Precipitation Products: A Machine Learning Algorithm

Remote Sensing, Год журнала: 2022, Номер 14(24), С. 6389 - 6389

Опубликована: Дек. 17, 2022

In recent decades, several products have been proposed for estimating precipitation amounts. However, due to the complexity of climatic conditions, topography, etc., providing more accurate and stable is great importance. Therefore, purpose this study was develop a multi-source data fusion method improve accuracy products. study, from 14 existing products, digital elevation model (DEM), land surface temperature (LST) soil water index (SWI) recorded at 256 gauge stations in Saudi Arabia were used. first step, assessed. second importance degree various independent variables, such as interpolation maps obtained stations, elevation, LST SWI improving modelling, evaluated. Finally, produce product with higher accuracy, information variables combined using machine learning algorithm. Random forest regression 150 trees used The highest lowest production based on characteristics, respectively. properties including SWI, DEM 65%, 22% 13%, IMERGFinal (9.7), TRMM3B43 (10.6), PRECL (11.5), GSMaP-Gauge (12.5), CHIRPS (13.0 mm/mo) had RMSE values. KGE values these estimation 0.56, 0.48, 0.52, 0.44 0.37, 6.6 mm/mo 0.75, respectively, which indicated compared results showed that different improved estimation.

Язык: Английский

---

A Novel Approach to Mapping the Spatial Distribution of Fruit Trees Using Phenological Characteristics

Agronomy, Год журнала: 2024, Номер 14(1), С. 150 - 150

Опубликована: Янв. 9, 2024

The lack of high-spectral and high-resolution remote sensing data is impeding the differentiation various fruit tree species that share comparable spectral spatial features, especially for evergreen broadleaf trees in tropical subtropical areas. Here, we propose a novel decision approach to map distribution at 10 m resolution based on growth stage features extracted from Sentinel-1A (S-1A) time-series synthetic aperture radar (SAR) data. This method was applied Maoming City, which known its vast cultivation trees, such as litchi, citrus, longan. results showed key extracting information lies fact ripening expansion period attenuates vegetation characteristic reproductive period. Under VH polarization, different traits were more separable easier distinguish. optimal Hv (high valley value 14 May, 26 7 June SAR data), Tb (difference between January Cr 13 July, 25 6 August Lo 23 September, 17 October, 11 November constructed window. thresholds these set 1, 1.5, respectively. classification model can effectively distinguish extract with overall accuracy (OA) 90.34% Kappa coefficient 0.84. proposed extracts accurately provides reference extraction species.

Язык: Английский

---

Improved method for cropland extraction of seasonal crops from multi-sensor satellite data

International Journal of Remote Sensing, Год журнала: 2024, Номер 45(18), С. 6249 - 6284

Опубликована: Авг. 26, 2024

Monitoring agricultural land over vast geographical areas presents challenges due to the absence of accurate, comprehensive and precise data, which has become a complex process that is difficult do in terms both timespans consistency. Hence, this study an improved approach for identification by utilizing capabilities Sentinel-1 Sentinel-2 satellites with variety vegetation non-vegetation indices machine learning algorithms. The Multispectral Correlation Mapper (MCM) Random Forest (RF) algorithms are adopted train different lands, crop types sowing cultivation seasons. 45-bands mega-file data cube (MFDC) fusion each season incorporates essential features derived from datasets seasons, i.e. Rabi (winter-spring season) Kharif (summer-autumn season). proposed method demonstrated resilience when applied satellite while effectively reducing impact non-agricultural elements such as shrubs, grass, bare soil orchards. results demonstrate notable ability differentiate between resulting high level precision measuring extent cultivated during seasons area 626,947 acres 590,858 acres, respectively. total area, ascertained observation cropping pattern modifications entire year (June 2021–May 2022) 635,655 acres. validation exercise shows higher accuracy cropland, overall 98.8%, kappa 0.97, user 98.69% producer 99.13%. Additionally, it was spatially compared ESRI, ESA MODIS cropland layers government statistical data. Furthermore, research investigates temporal dynamics growth phases using spectral bands indices. This improves provides useful insights into phenology.

Язык: Английский

---

Identifying Changes and Their Drivers in Paddy Fields of Northeast China: Past and Future

Agriculture, Год журнала: 2024, Номер 14(11), С. 1956 - 1956

Опубликована: Окт. 31, 2024

Northeast China plays a crucial role as major grain-producing region, and attention to its land use cover changes (LUCC), especially farmland changes, are ensure food security promote sustainable development. Based on the Moderate Resolution Imaging Spectroradiometer (MODIS) data decision tree model, types, those of paddy fields in from 2000 2020, were extracted, spatiotemporal their drivers analyzed. The development trends under different future scenarios explored alongside Coupled Model Intercomparison Project Phase 6 (CMIP6) data. findings revealed that kappa coefficients classification 2020 reached 0.761–0.825, with an overall accuracy 80.5–87.3%. proposed method can be used for long-term field monitoring China. LUCC is dominated by expansion fields. centroids gradually shifted toward northeast distance 292 km, climate warming being main reason shift. Under various scenarios, temperature surrounding regions projected rise. Each scenario anticipated meet conditions necessary northeastward This study provides support ensuring agricultural

Язык: Английский

---

Developing a Semi-Supervised Strategy in Time Series Mapping of Wetland Covers: A Case Study of Zrebar Wetland, Iran

Earth Systems and Environment, Год журнала: 2024, Номер 8(3), С. 815 - 830

Опубликована: Авг. 12, 2024

Abstract Wetlands, essential for Earth’s health, ecological balance, and local economies, require accurate monitoring assessment effective conservation. Data-driven models based on remote sensing are highly capable of the status classification wetlands. This study developed a semi-supervised framework mapping wetland covers in Zrebar, Iran, using Landsat time series data from 1984 to 2022. A pixel purification technique was applied temporal candidate images refine initial training (conventional scenario) generate purified (proposed scenario). The Support Vector Machine (SVM) algorithm utilized classify land cover within wetland, accuracy two scenarios evaluated compared. Over period, analysis changes Zrebar Wetland revealed significant spatial soil farmland, reed, water omission error rates classes were decreased 0.14, 0.12 scenario 1 0.03, 0.05, 0.05 2, respectively. In addition, commission these 0.13, 0.18, 0.09 0.04, 0.06, 0.04 after applying filtered 2. Finally, overall (scenario 1) 2) 0.86 0.94, These results underscore effectiveness proposed strategy enhancing over time, highlighting its potential future conservation efforts.

Язык: Английский

---

A New Risk-Based Method in Decision Making to Create Dust Sources Maps: A Case Study of Saudi Arabia

Remote Sensing, Год журнала: 2023, Номер 15(21), С. 5193 - 5193

Опубликована: Окт. 31, 2023

Dust storms are one of the major causes destruction natural ecosystems and human infrastructure worldwide. Therefore, identification mapping susceptible regions to dust storm formation (SRDSFs) is great importance. Determining SRDSFs by considering concept risk in decision-making process kind manager’s attitude planning can be very valuable dedicating financial resources time identifying controlling negative impacts SRDSFs. The purpose this study was present a new risk-based method decision making create SRDSF maps pessimistic optimistic scenarios. To achieve research, effective criteria obtained from various sources were used, including simulated surface data, satellite products, soil data Saudi Arabia. These included vegetation cover, moisture, erodibility, wind speed, precipitation, absolute air humidity. For purpose, ordered weighted averaging (OWA) model employed generate existing different results showed that speed precipitation had highest lowest impact centers, respectively. areas identified as pessimistic, neutral, optimistic, scenarios 85,950, 168,275, 255,225, 410,000, 596,500 km2, overall accuracy 84.1, 83.3, 81.6, 78.2, 73.2%, scenario identify area with higher accuracy. these compared previous studies 92.7, 94.2, 95.1, 88.4, 79.7% have agreement neutral scenario. proposed has high flexibility producing wide range suitable for risk-averse managers limited time, risk-taking sufficient time.

Язык: Английский

---

Enhanced Corn Mapping with Height-Spectral Gaussian Mixture Modeling

Опубликована: Янв. 1, 2024

Язык: Английский

---

Uncertainty Analysis of Remote Sensing Underlying Surface in Land–Atmosphere Interaction Simulated Using Land Surface Models

Atmosphere, Год журнала: 2023, Номер 14(2), С. 370 - 370

Опубликована: Фев. 13, 2023

This paper reports a comparative experiment using remote sensing underlying surface data (ESACCI) and Community Land Model (CLM_LS) to analyze the uncertainty of land types in land–atmosphere interaction. The results showed that global distribution ESACCI cropland is larger than CLM_LS, there great degree difference some regions, which can reach more 50% regionally. Furthermore, changes conditions be transmitted model through itself, resulting energy balance, micro-meteorological elements, water balance simulated by model, further affects climate simulation effect.

Язык: Английский

---

Multimodal crop cover identification using deep learning and remote sensing

Multimedia Tools and Applications, Год журнала: 2023, Номер 83(11), С. 33141 - 33159

Опубликована: Сен. 26, 2023

Язык: Английский

---