Explore the vast range of titles available.

Multispectral imagery provides unprecedented information on Earth system processes: however, data gaps due to clouds and shadows are a major limitation. Normalized-Difference Vegetation Index (NDVI) and Normalized-Difference Water Index (NDWI) are two spectral indexes employed for monitoring vegetation phenology, land-cover change and more. Synthetic Aperture Radar (SAR) with its cloud-penetrating abilities can fill data gaps using coincident imagery. In this study, we evaluated C-band Sentinel-1, L-band Uninhabited Aerial Vehicle SAR (UAVSAR) and texture for gap filling using efficient machine learning regression algorithms across three seasons. Multiple models were evaluated including Support Vector Machine, Random Forest, Gradient Boosted Trees and an ensemble of models. The Gap filling ability of SAR was evaluated with Sentinel-2 imagery from the same date, 3 days and 8 days later than both SAR sensors in September. Sentinel-1 and Sentinel-2 imagery from winter and spring seasons were also evaluated. Because SAR imagery contains noise, we compared two robust de-noising methods and evaluated performance against a refined lee speckle filter. Mean Absolute Error (MAE) rates of the cloud gap-filling model were assessed across different dataset combinations and land covers. The results indicated de-noised Sentinel-1 SAR and UAVSAR with GLCM texture provided the highest predictive abilities with random forest R2 = 0.91 (±0.014), MAE = 0.078 (±0.003) (NDWI) and R2 = 0.868 (±0.015), MAE = 0.094 (±0.003) (NDVI) during September. The highest errors were observed across bare ground and forest, while the lowest errors were on herbaceous and woody wetland. Results on January and June imagery without UAVSAR were less strong at R2 = 0.60 (±0.036), MAE = 0.211 (±0.005) (NDVI), R2 = 0.61 (±0.043), MAE = 0.209 (±0.005) (NDWI) for January and R2 = 0.72 (±0.018), MAE = 0.142 (±0.004) (NDVI), R2 = 0.77 (±0.022), MAE = 0.125 (±0.004) (NDWI) for June. Ultimately, the results suggest de-noised C-band SAR with texture metrics can accurately predict NDVI and NDWI for gap-filling clouds during most seasons. These shallow machine learning models are rapidly trained and applied faster than intensive deep learning or time series methods.

We assessed the fire trends from Moderate Resolution Imaging Spectroradiometer (MODIS) (2003–2016) and Visible Infrared Imaging Radiometer Suite (VIIRS) (2012–2016) in South/Southeast Asia (S/SEA) at a country level and vegetation types. We also quantified the fire frequencies, anomalies and climate drivers. MODIS data suggested India, Pakistan, Indonesia and Myanmar as having the most fires. Also, the VIIRS-detected fires were higher than MODIS (AQUA and TERRA) by a factor of 7 and 5 in S/SEA. Thirty percent of S/SEA had recurrent fires with the most in Laos, Cambodia, Thailand, and Myanmar. Statistically-significant increasing fire trends were found for India (p = 0.004), Cambodia (p = 0.001), and Vietnam (p = 0.050) whereas Timor Leste (p = 0.004) had a decreasing trend. An increasing trend in fire radiative power (FRP) were found for Cambodia (p = 0.005), India (0.039), and Pakistan (0.06) and declining trend in Afghanistan (0.041). Fire trends from VIIRS were not significant due to limited duration of data. In S/SEA, fires in croplands were equally frequent as in forests, with increasing fires in India, Pakistan, and Vietnam. Specific to climate drivers, precipitation could explain more variations in fires than the temperature with stronger correlations in Southeast Asia than South Asia. Our results on fire statistics including spatial geography, variations, frequencies, anomalies, trends, and climate drivers can be useful for fire management in S/SEA countries.

Quantifying emissions from crop residue burning is crucial as it is a significant source of air pollution. In this study, we first compared the fire products from two different sensors, the Visible Infrared Imaging Radiometer Suite (VIIRS) 375 m active fire product (VNP14IMG) and Moderate Resolution Imaging Spectroradiometer (MODIS) 1 km fire product (MCD14ML) in an agricultural landscape, Punjab, India. We then performed an intercomparison of three different approaches for estimating total particulate matter (TPM) emissions which includes the fire radiative power (FRP) based approach using VIIRS and MODIS data, the Global Fire Emissions Database (GFED) burnt area emissions and a bottom-up emissions approach involving agricultural census data. Results revealed that VIIRS detected fires were higher by a factor of 4.8 compared to MODIS Aqua and Terra sensors. Further, VIIRS detected fires were higher by a factor of 6.5 than Aqua. The mean monthly MODIS Aqua FRP was found to be higher than the VIIRS FRP; however, the sum of FRP from VIIRS was higher than MODIS data due to the large number of fires detected by the VIIRS. Besides, the VIIRS sum of FRP was 2.5 times more than the MODIS sum of FRP. MODIS and VIIRS monthly FRP data were found to be strongly correlated (r2 = 0.98). The bottom-up approach suggested TPM emissions in the range of 88.19-91.19 Gg compared to 42.0-61.71 Gg, 42.59-58.75 Gg and 93.98-111.72 Gg using the GFED, MODIS FRP, and VIIRS FRP based approaches, respectively. Of the different approaches, VIIRS FRP TPM emissions were highest. Since VIIRS data are only available since 2012 compared to MODIS Aqua data which have been available since May 2002, a prediction model combining MODIS and VIIRS FRP was derived to obtain potential TPM emissions from 2003-2016. The results suggested a range of 2.56-63.66 (Gg) TPM emissions per month, with the highest crop residue emissions during November of each year. Our results on TPM emissions for seasonality matched the ground-based data from the literature. As a mitigation option, stringent policy measures are recommended to curtail agricultural residue burning in the study area.

Remote sensing is essential for mapping and monitoring burnt areas. Integrating Very High-Resolution (VHR) data with medium-resolution datasets like Landsat and deep learning algorithms can enhance mapping accuracy. This study employs two deep learning algorithms, UNET and Gated Recurrent Unit (GRU), to classify burnt areas in the Bandipur Forest, Karnataka, India. We explore using VHR imagery with limited samples to train models on Landsat imagery for burnt area delineation. Four models were analyzed: (a) custom UNET with Landsat labels, (b) custom UNET with PlanetScope-labeled data on Landsat, (c) custom UNET-GRU with Landsat labels, and (d) custom UNET-GRU with PlanetScope-labeled data on Landsat. Custom UNET with Landsat labels achieved the best performance, excelling in precision (0.89), accuracy (0.98), and segmentation quality (Mean IOU: 0.65, Dice Coefficient: 0.78). Using PlanetScope labels resulted in slightly lower performance, but its high recall (0.87 for UNET-GRU) demonstrating its potential for identifying positive instances. In the study, we highlight the potential and limitations of integrating VHR with medium-resolution satellite data for burnt area delineation using deep learning.

Air pollution is one of the major environmental concerns in Vietnam. In this study, we assess the current status of air pollution over Hanoi, Vietnam using multiple different satellite datasets and weather information, and assess the potential to capture rice residue burning emissions with satellite data in a cloud-covered region. We used a timeseries of Ozone Monitoring Instrument (OMI) Ultraviolet Aerosol Index (UVAI) satellite data to characterize absorbing aerosols related to biomass burning. We also tested a timeseries of 3-hourly MERRA-2 reanalysis Black Carbon (BC) concentration data for 5 years from 2012-2016 and explored pollution trends over time. We then used MODIS active fires, and synoptic wind patterns to attribute variability in Hanoi pollution to different sources. Because Hanoi is within the Red River Delta where rice residue burning is prominent, we explored trends to see if the residue burning signal is evident in the UVAI or BC data. Further, as the region experiences monsoon-influenced rainfall patterns, we adjusted the BC data based on daily rainfall amounts. Results indicated forest biomass burning from Northwest Vietnam and Laos impacts Hanoi air quality during the peak UVAI months of March and April. Whereas, during local rice residue burning months of June and October, no increase in UVAI is observed, with slight BC increase in October only. During the peak BC months of December and January, wind patterns indicated pollutant transport from southern China megacity areas. Results also indicated severe pollution episodes during December 2013 and January 2014. We observed significantly higher BC concentrations during nighttime than daytime with peaks generally between 2130 and 0030 local time. Our results highlight the need for better air pollution monitoring systems to capture episodic pollution events and their surface-level impacts, such as rice residue burning in cloud-prone regions in general and Hanoi, Vietnam in particular.

Burnt area (BA) mapping is crucial for assessing wildfire impact, guiding restoration efforts, and improving fire management strategies. Accurate BA data helps estimate carbon emissions, biodiversity loss, and land surface properties post-fire changes. In this study, we designed and evaluated two deep learning-based architectures, a Custom UNET and a novel UNET-Gated Recurrent Unit (GRU), for burnt area classification using PlanetScope data over Bandipur, India. Both models demonstrated high accuracy in classifying burnt and unburnt areas. Performance metrics, including Precision, Recall, F1-Score, Accuracy, Mean Intersection over Union (IoU), and Dice Coefficient, revealed that the UNET-GRU hybrid consistently outperformed the Custom UNET, particularly in Recall and spatial overlap metrics. The Receiver Operating Characteristic (ROC) curve indicated excellent classification performance for both models, with the UNET-GRU achieving a higher AUC (0.98) compared to the Custom UNET (0.96). These findings highlight the UNET-GRU’s enhanced capacity to handle finer distinctions and capture spatial and contextual features, making it a robust choice for burnt area classification in the study area. While both models avoided overfitting and maintained generalizability, integrating GRU into the UNET architecture proved particularly effective for precise classification and spatial accuracy. Our results highlight the potential of the novel UNET-GRU for burnt area mapping using very high-resolution data.

In this study, we characterize the impacts of COVID-19 on air pollution using NO 2 and Aerosol Optical Depth (AOD) from TROPOMI and MODIS satellite datasets for 41 cities in India. Specifically, our results suggested a 13% NO 2 reduction during the lockdown (March 25–May 3rd, 2020) compared to the pre-lockdown (January 1st–March 24th, 2020) period. Also, a 19% reduction in NO 2 was observed during the 2020-lockdown as compared to the same period during 2019. The top cities where NO 2 reduction occurred were New Delhi (61.74%), Delhi (60.37%), Bangalore (48.25%), Ahmedabad (46.20%), Nagpur (46.13%), Gandhinagar (45.64) and Mumbai (43.08%) with less reduction in coastal cities. The temporal analysis revealed a progressive decrease in NO 2 for all seven cities during the 2020 lockdown period. Results also suggested spatial differences, i.e., as the distance from the city center increased, the NO 2 levels decreased exponentially. In contrast, to the decreased NO 2 observed for most of the cities, we observed an increase in NO 2 for cities in Northeast India during the 2020 lockdown period and attribute it to vegetation fires. The NO 2 temporal patterns matched the AOD signal; however, the correlations were poor. Overall, our results highlight COVID-19 impacts on NO 2 , and the results can inform pollution mitigation efforts across different cities of India.

In this study, we explored the relationships between the satellite-retrieved fire counts (FC), fire radiative power (FRP) and aerosol indices using multi-satellite datasets at a daily time-step covering ten different biomass burning regions in Asia. We first assessed the variations in MODIS-retrieved aerosol optical depths (AOD's) in agriculture, forests, plantation and peat land burning regions and then used MODIS FC and FRP (hereafter FC FRP) to explain the variations in AOD characteristics. Results suggest that tropical broadleaf forests in Laos burn more intensively than the other vegetation fires. FC FRP-AOD correlations in different agricultural residue burning regions did not exceed 20% whereas in forest regions they reached 40%. To specifically account for absorbing aerosols, we used Ozone Monitoring Instrument-derived aerosol absorption optical depth (AAOD) and UV aerosol index (UVAI). Results suggest relatively high AAOD and UVAI values in forest fires compared with peat and agriculture fires. Further, FC FRP could explain a maximum of 29% and 53% of AAOD variations, whereas FC FRP could explain at most 33% and 51% of the variation in agricultural and forest biomass burning regions, respectively. Relatively, UVAI was found to be a better indicator than AOD and AAOD in both agriculture and forest biomass burning plumes. Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations data showed vertically elevated aerosol profiles greater than 3.2-5.3 km altitude in the forest fire plumes compared to 2.2-3.9 km and less than 1 km in agriculture and peat-land fires, respectively. We infer the need to assimilate smoke plume height information for effective characterization of pollutants from different sources.

A near-global framework for automated training data generation and land cover classification using shallow machine learning with low-density time series imagery does not exist. This study presents a methodology to map nine-class, six-class, and five-class land cover using two dates (winter and non-winter) of a Sentinel-2 granule across seven international sites. The approach uses a series of spectral, textural, and distance decision functions combined with modified ancillary layers (such as global impervious surface and global tree cover) to create binary masks from which to generate a balanced set of training data applied to a random forest classifier. For the land cover masks, stepwise threshold adjustments were applied to reflectance, spectral index values, and Euclidean distance layers, with 62 combinations evaluated. Global (all seven scenes) and regional (arid, tropics, and temperate) adaptive thresholds were computed. An annual 95th and 5th percentile NDVI composite was used to provide temporal corrections to the decision functions, and these corrections were compared against the original model. The accuracy assessment found that the regional adaptive thresholds for both the two-date land cover and the temporally corrected land cover could accurately map land cover type within nine-class (68.4% vs. 73.1%), six-class (79.8% vs. 82.8%), and five-class (80.1% vs. 85.1%) schemes. Lastly, the five-class and six-class models were compared with a manually labeled deep learning model (Esri), where they performed with similar accuracies (five classes: Esri 80.0 ± 3.4%, region corrected 85.1 ± 2.9%). The results highlight not only performance in line with an intensive deep learning approach, but also that reasonably accurate models can be created without a full annual time series of imagery.

Vegetation fires are common in South/Southeast Asian (SA/SEA) countries. However, very few studies focused on vegetation fires and the changes during the COVID as compared to pre-pandemic. This study fills an information gap and reports total fire incidences, total burnt area, type of vegetation burnt, and total particulate matter emission variations in SA/SEA during COVID-2020 and pre-pandemic (2012–2019). Results from the short-term 2020-COVID versus 2019-non-COVID year showed a decline in fire counts varying from − 2.88 to 79.43% in S/SEA. The exceptions in South Asia include Afghanistan and Sri Lanka, with a 152% and 4.9% increase, and Cambodia and Myanmar in Southeast Asia, with an 11.1% and 8.5% increase in fire counts in the 2020-COVID year. The burnt area decline for 2020 compared to 2019 varied from − 0.8% to 92% for South/Southeast Asian countries, with most burning in agricultural landscapes than forests. Several patches in S/SEA showed a decrease in fires for the 2020 pandemic year compared to long term 2012–2020 pre-pandemic record, with Z scores greater or less than two denoting statistical significance. However, on a country scale, the results were not statistically significant in both S/SEA, with Z scores ranging from − 0.24 to − 1, although most countries experienced a decrease in fire counts. The associated mean TPM emissions declined from ~ 2.31 Tg (0.73stdev) during 2012–2019 to 2.0 (0.65stdev)Tg in 2020 in South Asia and 6.83 (0.70stdev)Tg during 2012–2019 to 5.71 (0.69 stdev)Tg in 2020 for South East Asian countries. The study highlights variations in fires and emissions useful for fire management and mitigation.