Search

This work builds on the original semi-automated land cover mapping algorithm and quantifies improvements to class accuracy, analyzes the results, and conducts a more in-depth accuracy assessment in conjunction with test sites and the National Land Cover Database (NLCD). This algorithm uses support vector machines trained on data collected across the continental United States to generate a pre-trained model for inclusion into a decision support tool within ArcGIS Pro. Version 2 includes an additional snow cover class and accounts for snow cover effects within the other land cover classes. Overall accuracy across the continental United States for Version 2 is 75% on snow-covered pixels and 69% on snow-free pixels, versus 16% and 66% for Version 1. However, combining the “crop” and “low vegetation” classes improves these values to 86% for snow and 83% for snow-free, compared to 19% and 83% for Version 1. This merging is justified by their spectral similarity, the difference between crop and low vegetation falling closer to land use than land cover. The Version 2 tool is built into a Python-based ArcGIS toolbox, allowing users to leverage the pre-trained model—along with image splitting and parallel processing techniques—for their land cover type map generation needs.

This study presents an automated methodology to generate training data for surface water mapping from a single Sentinel-2 granule at 10 m (4 band, VIS/NIR) or 20 m (9 band, VIS/NIR/SWIR) resolution without the need for ancillary training data layers. The 20 m method incorporates an ensemble of three spectral indexes with optimal band thresholds, whereas the 10 m method achieves similar results using fewer bands and a single spectral index. A spectrally balanced and randomly generated set of training data based on the index values and optimal thresholds is used to fit machine learning classifiers. Statistical validation compares the 20 m ensemble-only method to the 20 m ensemble method with a random forest classifier. Results show the 20 m ensemble-only method had an overall accuracy of 89.5% (±1.7%), whereas the ensemble method combined with the random forest classifier performed better, with a ~4.8% higher overall accuracy: 20 m method (94.3% (±1.3%)) with optimal spectral index and SWIR thresholds of −0.03 and 800, respectively, and 10 m method (93.4% (±1.5%)) with optimal spectral index and NIR thresholds of −0.01 and 800, respectively. Comparison of other supervised classifiers trained automatically with the framework typically resulted in less than 1% accuracy improvement compared with the random forest, suggesting that training data quality is more important than classifier type. This straightforward framework enables accurate surface water classification across diverse geographies, making it ideal for development into a decision support tool for water resource managers.

Wildland fires result in a unique signal detectable by multispectral remote sensing and synthetic aperture radar (SAR). However, in many regions, such as Southeast Asia, persistent cloud cover and aerosols temporarily obstruct multispectral satellite observations of burned area, including the MODIS MCD64A1 Burned Area Product (BAP). Multiple days between cloud free pre- and postburn MODIS observations result in burn date uncertainty. We incorporate cloud-penetrating, C-band SAR-with the MODIS MCD64A BAP in Southeast Asia, to exploit the strengths of each dataset to better estimate the burn date and reduce the potential burn date uncertainty range. We incorporate built-in quality control using MCD64A1 to reduce erroneous pixel updating. We test the method over part of Laos and Thailand during April 2016 and found average uncertainty reduction of 4.5 d, improving 15% of MCD64A1 pixels. A new BAP could improve monitoring temporal trends of wildland fires, air quality studies and monitoring post-fire vegetation dynamics.

Wildland fires result in a unique signal detectable by multispectral remote sensing and synthetic aperture radar (SAR). However, in many regions, such as Southeast Asia, persistent cloud cover and ...

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.

Monitoring the impacts of ecosystem restoration strategies requires both short-term and long-term land surface monitoring. The combined use of unmanned aerial systems (UAS) and satellite imagery enable effective landscape and natural resource management. However, processing, analyzing, and creating derivative imagery products can be time consuming, manually intensive, and cost prohibitive. In order to provide fast, accurate, and standardized UAS and satellite imagery processing, we have developed a suite of easy-to-use tools integrated into the graphical user interface (GUI) of ArcMap and ArcGIS Pro as well as open-source solutions using NodeOpenDroneMap. We built the Monitoring Ecological Restoration with Imagery Tools (MERIT) using Python and leveraging third-party libraries and open-source software capabilities typically unavailable within ArcGIS. MERIT will save US Army Corps of Engineers (USACE) districts significant time in data acquisition, processing, and analysis by allowing a user to move from image acquisition and preprocessing to a final output for decision-making with one application. Although we designed MERIT for use in wetlands research, many tools have regional or global relevancy for a variety of environmental monitoring initiatives.

Feature extraction algorithms are routinely leveraged to extract building footprints and road networks into vector format. When used in conjunction with high resolution remotely sensed imagery, machine learning enables the automation of such feature extraction workflows. However, many of the feature extraction algorithms currently available have not been thoroughly evaluated in a scientific manner within complex terrain such as the cities of developing countries. This report details the performance of three automated feature extraction (AFE) datasets: Ecopia, Tier 1, and Tier 2, at extracting building footprints and roads from high resolution satellite imagery as compared to manual digitization of the same areas. To avoid environmental bias, this assessment was done in two different regions of the world: Maracay, Venezuela and Niamey, Niger. High, medium, and low urban density sites are compared between regions. We quantify the accuracy of the data and time needed to correct the three AFE datasets against hand digitized reference data across ninety tiles in each city, selected by stratified random sampling. Within each tile, the reference data was compared against the three AFE datasets, both before and after analyst editing, using the accuracy assessment metrics of Intersection over Union and F1 Score for buildings and roads, as well as Average Path Length Similarity (APLS) to measure road network connectivity. It was found that of the three AFE tested, the Ecopia data most frequently outperformed the other AFE in accuracy and reduced the time needed for editing.

In this study, we characterize the impacts of COVID-19 on air pollution using NO2 and Aerosol Optical Depth (AOD) from TROPOMI and MODIS satellite datasets for 41 cities in India. Specifically, our results suggested a 13% NO2 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 NO2 was observed during the 2020-lockdown as compared to the same period during 2019. The top cities where NO2 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 NO2 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 NO2 levels decreased exponentially. In contrast, to the decreased NO2 observed for most of the cities, we observed an increase in NO2 for cities in Northeast India during the 2020 lockdown period and attribute it to vegetation fires. The NO2 temporal patterns matched the AOD signal; however, the correlations were poor. Overall, our results highlight COVID-19 impacts on NO2, and the results can inform pollution mitigation efforts across different cities of India.

Paddy Rice is the prevalent land cover in the mosaicked landscape of the Hanoi Capital Region, Vietnam. In this study, we map double and single crop rice in Hanoi using a random forest algorithm and a time-series of Sentinel-1 SAR imagery at 10 and 20 m resolution using VV-only, VH-only, and both polarizations. We compare spatial and areal variation and quantify input band importance, estimate crop growth stages, estimate rice field/collective metrics using Fragstats with image segmentation, and highlight the importance of the results for land use and land cover. Results suggest double crop rice ranged from 208 000 to 220 000 ha with 20-m resolution imagery accounting for the most area in all polarizations. Based on accuracy assessment, we found 10 m data for VV/VH to have highest overall accuracy (93.5%, ±1.33%), while VV at 10 and 20 m had lowest overall accuracies (90.9%, ±1.57; 91.0%, ±2.75). Mean decrease in accuracy suggests for all but VV at 10 m, data from harvest and flooding stages are most critical for classification. Results suggest 20 m data for both VV and VH overestimates rice land cover, however 20 m data may be indicative of rice land use. Analysis of growing season suggests average estimated length of 93–104 days for each season. Commune-level results suggest up to 20% coefficient of variation between VV10m and VH10m with significant spatial variation in rice area. Landscape metrics show rice fields are typically planted in groups of 3–4 fields with over 796 000 collectives and 2.69 million fields estimated in the study area.

Recent abundance of moderate-to-high spatial resolution satellite imagery has facilitated land-cover map production. However, in cloud-prone areas, building high-resolution land-cover maps is still...

Can Tho City has quickly become a modernized and industrialized city undergoing rapid population growth affecting the local environment, especially air quality and human health. In 2015, Can Tho had 1,251,809 inhabitants with a total of 566,593 motorcycles and 15,105 automobiles. There are about 1000 factories in the city. The top polluters are the industries of textile and dyeing, food processing, cement, and steel mill and rice processing. The aims of this research are to (i) conduct a detailed air pollution emission inventory (ii) study the formation of the air pollution plume over the city, and (iii) study different pollution abatement strategies for the city. We employ a combination of bottom-up and top-down approaches to conduct air pollution emission inventory, then, the finite volume model-transport and photochemistry mesoscale model is applied for studying the formation of the pollution plume. The results showed that transportation and industrial activities are the two main emission sources responsible for 80% of total NO x , 90% of total SO2, 75% of CO, 60% of total suspended particles, and 60% of non-methane volatile organic compounds. Modeling results showed that the highest average—1 h—of O3 is 206 μg/m3 which is higher than the Vietnam ambient air quality standard. The pollution plume is developed in the northeastern part of the city. Finally, abatement measures were proposed. This is the first comprehensive study on air pollution emissions and air quality modeling in the Mekong Delta, yielding insight to support government authorities to promulgate plans and actions to reduce emissions, protecting human health and the environment while leading towards sustainable development.

Satellite observations for regional air quality assessment rely on comprehensive spatial coverage, and daily monitoring with reliable, cloud-free data quality. We investigated spatiotemporal variation and data quality of two global satellite Aerosol Optical Depth (AOD) products derived from MODIS and VIIRS imagery. AOD is considered an essential atmospheric parameter strongly related to ground Particulate Matter (PM) in Southeast Asia (SEA). We analyze seasonal variation, urban/rural area influence, and biomass burning effects on atmospheric pollution. Validation indicated a strong relationship between AERONET ground AOD and both MODIS AOD (R2 = 0.81) and VIIRS AOD (R2 = 0.68). The monthly variation of satellite AOD and AERONET AOD reflects two seasonal trends of air quality separately for mainland countries including Myanmar, Laos, Cambodia, Thailand, Vietnam, and Taiwan, Hong Kong, and for maritime countries consisting of Indonesia, Philippines, Malaysia, Brunei, Singapore, and Timor Leste. The mainland SEA has a pattern of monthly AOD variation in which AODs peak in March/April, decreasing during wet season from May–September, and increasing to the second peak in October. However, in maritime SEA, AOD concentration peaks in October. The three countries with the highest annual satellite AODs are Singapore, Hong Kong, and Vietnam. High urban population proportions in Singapore (40.7%) and Hong Kong (21.6%) were associated with high AOD concentrations as expected. AOD values in SEA urban areas were a factor of 1.4 higher than in rural areas, with respective averages of 0.477 and 0.336. The AOD values varied proportionately to the frequency of biomass burning in which both active fires and AOD peak in March/April and September/October. Peak AOD in September/October in some countries could be related to pollutant transport of Indonesia forest fires. This study analyzed satellite aerosol product quality in relation to AERONET in SEA countries and highlighted framework of air quality assessment over a large, complicated region.

Biomass burning emissions variation was calculated based on differences in satellite-derived products: MODIS and MERIS-based land cover, MODIS and MERIS-based burned area (BA), as well as global and regionally - averaged emission factors. The products and resulting emissions were compared for 3 years (2006–2008) in Vietnam. They were compared at four spatial scales including: (1) country level; (2) region level; (3) land cover; and (4) grid-cell level. For the different products, we especially focused on BA as it is the major input for emission calculations. The BA products were tested for differences using the mean absolute deviation (MAD), Average Absolute Deviation (AAD), and Student’s t-test for significance. Of the different regions, Central Highlands showed the highest AAD in BAs. At a country level, the MERIS BA amounts were relatively higher than the MODIS BAs and especially during peak biomass burning months. We also found that emissions calculations using MERIS LC were relatively lower than those from MODIS LC. While over croplands, the regional emission factors yielded notably higher emissions compared with the global emission factors suggesting that current large-scale studies may be underestimating the biomass burning emissions. We further addressed the potential impact of emissions on urban air quality in Hanoi City through HYSPLIT trajectory modeling.

In this study, we quantified the relationship between fires and vegetation disturbance at varied spatial scales using moderate resolution imaging spectroradiometer (MODIS) datasets for the period 2003–2012. We report satellite-derived fire characteristics (frequency, extent, seasonality, and type of vegetation burnt) in Myanmar, the extent of fire disturbance, and the impact of the fires on gross primary productivity (GPP) at multiple scales. Results suggested March as the peak fire season with burnt areas (BAs) of $\textbf{12\,900}\;\textbf{km}^\textbf{2}$ and 95 000 fire counts. Forests accounted for 41.3% of the total BAs followed by shrub lands (33.6%) and agriculture (24.7%). The “low” vegetation disturbance category accounted for 9.2% of total fires, whereas the medium and high categories accounted for about 89.7%. We found relatively higher negative correlation between BA and GPP for deciduous forests ( $\textbf{r} = \textbf{0.49}, \textbf{p} \sim \textbf{0}$ ) than for evergreen forests ( $\textbf{r} = \textbf{0.36}, \textbf{p} \sim \textbf{0}$ ). A maximum decrease in 29% of original GPP (2007–2012) was observed in the evergreen forest patches. The scale-dependent correlation analysis suggested significant BA–GPP correlation at $\textbf{1} \times \textbf{1}$ degree compared to finer resolutions. Our results highlight the impact of fire disturbance on vegetation greenness and GPP in tropical forests of Myanmar.

Agricultural fires in the Indo-Gangetic Plains (IGP) are a major cause of air pollution. In this study, we evaluate fire regimes and quantify the potential of agricultural residues in generating bioenergy that otherwise are subject to burning by local farmers in the region. For characterizing the fire regimes, we used MODIS satellite datasets in conjunction with IRS-AWiFS classified data. We collected crop statistical data for area, production, and yield for 31 different crops and mapped the bioenergy potential of agricultural residues. We also tested the MODIS net primary production (NPP) dataset potential for crop yield estimation and thereby bioenergy calculations. Results from land use-fire analysis suggested that 88.13% of fires occurred in agricultural areas. Relatively more fires and burnt areas were recorded during the winter rice residue burning season than the summer wheat residue burning season. Monte Carlo analysis suggested that nearly 16.5 Tg of crop residues are burned at 60% probability. MODIS NPP data could explain 62% of variation in field-level crop yield estimates. Our analysis revealed that in the IGP nearly 73.28 Tg of crop residue biomass is available for recycling. The energy equivalent from these residues is estimated to be 1110.77 PJ. From the residues, the biogas potential production is estimated to be 1165.1098 million m 3 , the electric power potential at 20% efficiency is estimated at 61698.9 kWh, and the total bioethanol production potential at 21.0 billion liters. Results also highlight geographic locations of bioenergy resources in the IGP useful for energy planning. Controlling agricultural residue burning and promoting the bioenergy sector is an attractive “win–win” strategy in the IGP.

In Southeast Asia and Vietnam, rice residues are routinely burned after the harvest to prepare fields for the next season. Specific to Vietnam, the two prevalent burning practices include: a). piling the residues after hand harvesting; b). burning the residues without piling, after machine harvesting. In this study, we synthesized field and laboratory studies from the literature on rice residue burning emission factors for Particulate Matter less than 2.5 microns (PM2.5). We found significant differences in the resulting burning-practice specific emission factors, with 16.9 grams per square kilogram (plus or minus 6.9) for pile burning and 8.8 grams per square kilogram (plus or minus 3.5) for non-pile burning. We calculated burning practice specific emissions based on rice area data, region-specific fuel-loading factors, combined emission factors, and estimates of burning from the literature. Our results for year 2015 estimate 180 gigagrams of PM2.5 result from the pile burning method and 130 gigagrams result from non-pile burning method, with the most-likely current emission scenario of 150 gigagrams PM2.5 emissions for Vietnam. For comparison purposes, we calculated emissions using generalized agricultural emission factors employed in global biomass burning studies. These results estimate 80 gigagrams PM2.5, which is only 44 percent of the pile burning-based estimates, suggesting underestimation in previous studies. We compare our emissions to an existing all-combustion sources inventory, results show emissions account for 14-18 percent of Vietnam's total PM2.5 depending on burning practice. Within the highly-urbanized and cloud-covered Hanoi Capital region (HCR), we use rice area from Sentinel-1A to derive spatially-explicit emissions and indirectly estimate residue burning dates. Results from HYSPLIT (Hybrid Single-Particle Lagrangian Integrated Trajectory) back-trajectory analysis stratified by season show autumn has most emission trajectories originating in the North, while spring has most originating in the South, suggesting the latter may have bigger impact on air quality. From these results, we highlight locations where emission mitigation efforts could be focused and suggest measures for pollutant mitigation. Our study demonstrates the need to account for emissions variation due to different burning practices.

Forest fires are a significant source of air pollution in Asia. In this study, we integrate satellite remote sensing data and ground-based measurements to infer fire-air pollution relationships in selected regions of Vietnam. We first characterized the active fires and burnt areas at a regional scale from MODIS satellite data. We then used satellite-derived active fire data to correlate the resulting atmospheric pollution. Further, we analyzed the relationship between satellite atmospheric variables and ground-based air pollutant parameters. Our results show peak fire activity during March in Vietnam, with hotspots in the Northwest and Central Highlands. Active fires were significantly correlated with UV Aerosol Index (UVAI), aerosol extinction absorption optical depth (AAOD), and Carbon Monoxide. The use of satellite aerosol optical thickness improved the prediction of Particulate Matter (PM) concentration significantly.

In this study, we assess the intense pollution episode of June 2013, in Riau province, Indonesia from land clearing. We relied on satellite retrievals of aerosols and Carbon monoxide (CO) due to lack of ground measurements. We used both the yearly and daily data for aerosol optical depth (AOD), fine mode fraction (FMF), aerosol absorption optical depth (AAOD) and UV aerosol index (UVAI) for characterizing variations. We found significant enhancement in aerosols and CO during the pollution episode. Compared to mean (2008–2012) June AOD of 0.40, FMF-0.39, AAOD-0.45, UVAI-1.77 and CO of 200 ppbv, June 2013 values reached 0.8, 0.573, 0.672, 1.77 and 978 ppbv respectively. Correlations of fire counts with AAOD and UVAI were stronger compared to AOD and FMF. Results from a trajectory model suggested transport of air masses from Indonesia towards Malaysia, Singapore and southern Thailand. Our results highlight satellite-based mapping and monitoring of pollution episodes in Southeast Asia.

Vegetation fires are an important source of air pollution in several regions of the world including Asia. An important question with respect to satellite retrievals of air pollutants is “how well do they capture temporal and spatial variations and how well do they relate to episodic events such as fires?” We addressed this question using MOPITT surface CO and MODIS fire retrievals. We also evaluated MODIS aerosol optical depth (AOD) as well as small mode aerosol fraction (SMAF) variations in relation to fire seasonality. Results from temporal analysis (2003–2012) of fires in Asia suggested 22 % of all fires occurring in Myanmar, followed by India (20.91 %), Indonesia (18.31 %), Thailand (9.42 %), etc. Fire frequencies were highest in northeast India and Southeast Asia countries. Further, we observed significant spatial variation and seasonality in fires in Southeast Asia. In the northern Southeast Asia, the peak fire season was during January–March whereas in the south, the fires peak is from August through October. AOD followed a similar trend as that of fires, however, small mode aerosol fraction showed some discrepancies. Locally weighted regression yielded good results between vegetation fires and CO emissions. Results showed that areas with high vegetation fires were also areas of high CO emissions, with highest spatial correlation during the month of March. Among the fire counts and FRP, the correlations varied for individual months, however, both showed significant (P < 0.001) positive correlations suggesting that either of them can be used as predictor of CO concentrations. Locally weighted regression maps revealed how the relationship between fire counts versus CO and FRP versus CO change across time and space. The study captures the influence of vegetation fires on CO pollution in Asia using satellite data.