Conference Papers

Permanent URI for this collectionhttps://idr.nitk.ac.in/handle/123456789/28506

Browse

Filters

2023 - 20242

Settings

Author: search.filters.author.Shwetha, H.R.Subject: search.filters.subject.Machine Learning

Search Results

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    Crop Classification Based on Optimal Hyperspectral Narrow Bands Using Machine Learning and Hyperion Data
    (Institute of Electrical and Electronics Engineers Inc., 2023) Reddy, B.S.; Sharma, S.; Shwetha, H.R.
    In view of global climate change and the limited availability of cropland, crop classification plays a critical role in maintaining food security. Hyperspectral remote sensing has emerged as a valuable tool for classifying crops using detailed spectral information. To explore the potential of hyperspectral data for nationwide crop classification, the research uses the GHISACONUS library to identify Optimal Hyperspectral Narrow Bands (OHNBs) across seven Agricultural Experimental Zones (AEZ) in the USA. Principal Component Analysis (PCA) techniques are employed to identify 24 OHNBs from the data. OHNBs achieved notable accuracy rates, ranging from 75% to 91% when classifying different crop types and their growth stages. However, accuracy drops below 90% in significant cases, likely due to the limited selection of 24 OHNBs and the variation in crop phenology across the seven study areas. The research indicates that systematically selecting OHNBs based on crop phenological stages consistently achieves satisfactory classification accuracy. This approach effectively classifies crops in any Hyperion image. Overall, the study contributes significantly to our knowledge of using OHNBs for nationwide crop classification, highlighting the importance of considering phenological stages and data acquisition conditions to enhance accuracy. © 2023 IEEE.
  • No Thumbnail Available
    Item
    ENHANCING SPATIAL RESOLUTION OF GPM RAINFALL DATA IN UPPER CAUVERY BASIN, INDIA: MACHINE LEARNING APPROACH
    (Institute of Electrical and Electronics Engineers Inc., 2024) Kumar, P.G.; Saicharan, V.; Shwetha, H.R.
    Spatial downscaling is an effective way to obtain rainfall with sufficient spatial details. The spatial resolution of the Global Precipitation Measurement (GPM) mission (IMERG) satellite rainfall products is 0.1° × 0.1°, which is too coarse for regional-scale analysis. This study employed two Model averaging methods (Random Forest + XGBoost, Random Forest + CatBoost), Ensemble methods (Random Forest, XGBoost, CatBoost) and Stacked Random Forest + XGBoost model for downscaling GPM IMERG monthly rainfall over the Upper Cauvery Basin from 2015 to 2020 from 0.1°(~ 10 km) to 1 km resolution. Five land surface variables (auxiliary variables), NDVI, elevation, LST, slope, and aspect, were employed for this purpose. The stacked RFR+XGB model outperformed the model averaging techniques, achieving a higher R2 of 0.694 and a lower RMSE/MAE of 44.57/35.23. While the ensemble method yielded promising results, it struggled to predict extreme rainfall values. The downscaled dataset facilitates improved hydrological applications, including water footprint analysis, hydrological monitoring, and disaster warning systems. © 2024 IEEE.