Leaf area index estimation over the Kosi watershed in central Himalaya from remote sensing data using machine/deep learning models and validation against in situ measurements
by V. Rakesh, V. Arun Kumar & K. V. Ramesh
In this study, we present machine learning model support vector machine (SVM) and deep learning architectures, including long short-term memory (LSTM), gated recurrent unit (GRU), long- and short-term time series network (LSTNet), and temporal convolution network (TCN) to estimate leaf area index (LAI) utilizing Sentinel-2 multispectral imagery within the Kosi watershed located in the central Himalayan region is based on various vegetation indices. In situ LAI data were collected through a random sampling method using a plant canopy analyzer LI-COR over a continuous field survey period of 25 days for model training and validation. The current study evaluated four variants of the LSTM model, with the bidirectional LSTM emerging as the most effective in estimating LAI. The LAI derived using the Rendvi index was found to be superior compared to those obtained from other vegetation indices. The bidirectional LSTM (BiLSTM) model outperformed the SVM model in LAI retrieval based on all accuracy metrics computed. Specifically, the R2 and mean absolute percentage error (MAPE) values for the BiLSTM model were 0.99 and 0.29, respectively, while the SVM model achieved corresponding values of 0.93 and 0.56. Additionally, the performance of the BiLSTM in predicting LAI surpassed that of other deep learning models such as GRU (R2 = 0.99, MAPE = 0.82), LSTNet (R2 = 0.99, MAPE = 0.74), and TCN (R2 = 0.99, MAPE = 0.39). We compared the performance of the BiLSTM with the SVM model and the moderate-resolution imaging spectroradiometer (MODIS) LAI product by validating against in situ LAI measurements. The LAI retrieved using the BiLSTM model (R2 = 0.99) demonstrated superior agreement with in situ measurements compared to the MODIS LAI product (R2 = 0.0003). Furthermore, seasonal changes in vegetation and interannual variations in LAI over the Kosi watershed were more accurately depicted in the high-resolution LAI retrieved with BiLSTM from Sentinel-2 compared to the MODIS LAI product.
Impact of assimilating GNSS ground-based and Space-based observations on Extreme rainfall prediction over Karnataka, India
by Smrati Purwar, Ajay Bankar, V Rakesh, Suelynn Choy, Haobo Li, G N Mohapatra & Yuriy Kuleshov
This study delves into the assessment of the impact of assimilating two kinds of Global Navigation Satellite System (GNSS) observations (ground-based (GB) and space-based) in the prediction of extreme rainfall events (EREs). We consider seven distinct sources of GNSS radio occultation (GNSS-RO) observations as GNSS space-based observations, with six of them (COSMIC1, KOMPSAT5, MetOPA, MetOPB, TerraSARX, TanDEMX) being combined to form a group of Global Positioning System (GPS)-RO observations, while COSMIC-2 (C2), which receives signals from GPS and GLObal NAvigation Satellite System (GLONASS), is handled independently. The study is conducted over the southern state of India, Karnataka, for 16 EREs that occurred between October 2019 and December 2020. The zenith tropospheric delay (ZTD) from International GNSS Service (IGS) GB stations and refractivity data from GNSS-RO (GPS-RO and C2) are assimilated in the Weather Research and Forecasting (WRF) model for forecasting these EREs. The results indicate that the assimilation of ZTD from GNSS-GB observations does not significantly improve the forecast due to an insufficient number of observations, whereas the improvement through the assimilation of refractivity profiles from GPS-RO observations is significant even with a smaller number of observations, due to the high vertical resolution of the profiles. The assimilation of GNSS-GB and GPS-RO observations improves the humidity, temperature, and winds at different vertical levels which leads to improvement in the rainfall prediction, particularly in the interior part of Karnataka. Furthermore, the assimilation of C2 observations proves highly advantageous, resulting in significant improvements across forecasted weather parameters and rainfall when compared to GNSS-GB and GPS-RO. These improvements in weather parameters due to assimilation result in a substantial improvement in rainfall by 17% for assimilation of C2 and 21% for assimilation of all GNSS observations (GNSS-GB, GPS-RO, and C2) across most of the regions, including coastal and south interior parts of Karnataka. This study delves into the assessment of the impact of assimilating two kinds of Global Navigation Satellite System (GNSS) observations (ground-based (GB) and space-based) in the prediction of extreme rainfall events (EREs). We consider seven distinct sources of GNSS radio occultation (GNSS-RO) observations as GNSS space-based observations, with six of them (COSMIC1, KOMPSAT5, MetOPA, MetOPB, TerraSARX, TanDEMX) being combined to form a group of Global Positioning System (GPS)-RO observations, while COSMIC-2 (C2), which receives signals from GPS and GLObal NAvigation Satellite System (GLONASS), is handled independently. The study is conducted over the southern state of India, Karnataka, for 16 EREs that occurred between October 2019 and December 2020. The zenith tropospheric delay (ZTD) from International GNSS Service (IGS) GB stations and refractivity data from GNSS-RO (GPS-RO and C2) are assimilated in the Weather Research and Forecasting (WRF) model for forecasting these EREs. The results indicate that the assimilation of ZTD from GNSS-GB observations does not significantly improve the forecast due to an insufficient number of observations, whereas the improvement through the assimilation of refractivity profiles from GPS-RO observations is significant even with a smaller number of observations, due to the high vertical resolution of the profiles. The assimilation of GNSS-GB and GPS-RO observations improves the humidity, temperature, and winds at different vertical levels which leads to improvement in the rainfall prediction, particularly in the interior part of Karnataka. Furthermore, the assimilation of C2 observations proves highly advantageous, resulting in significant improvements across forecasted weather parameters and rainfall when compared to GNSS-GB and GPS-RO. These improvements in weather parameters due to assimilation result in a substantial improvement in rainfall by 17% for assimilation of C2 and 21% for assimilation of all GNSS observations (GNSS-GB, GPS-RO, and C2) across most of the regions, including coastal and south interior parts of Karnataka.
Multi-Instrument Observations of Unseasonal Post-Sunset Equatorial Plasma Bubbles During Two Moderate Geomagnetic Storms in May and June 2024 Over East/Southeast Asia
by Sampad Kumar Panda, Siva Sai Kumar Rajana, Chiranjeevi G Vivek, Ramkumar Vankadara, Punyawi Jamjareegulgarn
This study investigated an unseasonal development of post-sunset EPBs in summer solstice period over the East/Southeast Asian longitude region by using multi-instrument observations during two consecutive moderate geomagnetic storm events (16-17 May 2024 and 28-29 June 2024). The results indicate, formation of EPBs during non-climatological plasma bubble season is primarily driven by sustained southward oriented IMF-Bz in the storm main phase, which facilitated the penetration of eastward electric fields into the equatorial ionosphere. These electric fields uplifted the F-region plasma to altitudes favourable for irregularities growth. Also, noteworthy hemispheric asymmetry is noticed in the formation of EPBs, manifesting more intense occurrence in Southern Hemisphere during the geomagnetic storm of 16-17 May 2024 and extended up to ∼20oS magnetic latitude. During the 28-29 June 2024 geomagnetic storm, EPBs are more prominent in the Northern Hemisphere and reached beyond EIA region up to ∼22oN magnetic latitude. In brief, the EPBs developed initially over the 100°E longitude sector, exhibited eastward drift and thereafter extended to 120°E longitude region during the 16-17 May 2024 geomagnetic storm. During the 28-29 June 2024 geomagnetic storm, EPBs formed over the 120°E longitude region and later drifted to 140°E longitude sector. These findings highlight the dominant role of storm-time electrodynamics in triggering EPBs and emphasize the need for continuous regional monitoring of EPBs to mitigate space weather impacts on satellite-based communication and GNSS systems.
Identification of malaria vulnerable zones in North East Region of India: Spatiotemporal assessment through geographical information system and self-organizing maps
by Rajasekhar Mopuri, Madhusudhan Rao Kadiri, Kantha Rao Bhimala and Srinivasa Rao Mutheneni
WRF Model Forecast Evaluation of Extreme Rainfall Events over Karnataka during the Indian Summer Monsoon
by Ajay Bankar, Rakesh V, and Smrati Purwar
The skill of the Weather Research and Forecasting (WRF) model in predicting extreme rainfall events (EREs) is investigated for two different initialization times: 1200 UTC (12-hour lead time) and 0000 UTC (zero lead time). A high resolution (3-km) WRF model configuration is used to simulate 38 EREs in Karnataka, a state in southwest India. These EREs were identified by analyzing daily accumulated rainfall recorded by telemetric rain-gauge (TRG) stations over the region, specifically during the prominent rainy season in the state (June to November) for the period 2015–2019. EREs were defined using the 99.993rd percentile threshold for each subdivision and year. The model forecasts of rainfall and other meteorological variables were verified against high density TRG (6480 stations) and telemetric weather station (860 stations) observations throughout the Karnataka. The model performance was assessed through computation of quantitative error measures and statistical skill scores. Consistently, both sets of experiments reveal a propensity of WRF model to overestimate rainfall in the coastal region, with simulations initialized at 0000 UTC displaying a more pronounced wet bias compared to those at 1200 UTC. However, simulations initialized at 0000 UTC are more skillful than 1200 UTC in predicting the frequency of rainfall occurrences. Validation of other meteorological variables indicates that the model tends to underpredict temperature and overpredict relative humidity, especially during night hours, while it overpredicts wind speed over the region. Quantification of the model errors against in-situ measurements are crucial before deploying these models for operational applications and a deeper understanding of these errors will aid in further improvements to the model.
- Regional assessment of rainfall indices for drought conditions during the monsoon season in Odisha
- Influence of large-scale climate modes (ENSO, IOD and PJ) on marine heatwave characteristics in the Indian ocean region
- Impact of assimilation of microwave imager radiance data on simulation of tropical cyclones over Bay of Bengal using National Centre for Medium Range Weather Forecasting Unified Model modelling system
- Quantifying CMIP6 models’ performance in simulating marine heatwaves over the Indian Ocean region: A comprehensive analysis of key MHW properties and future projections
Page 5 of 27
