Fine Spatial Resolution Research Articles

Performance study of imageries from MODIS, Landsat 8 and Sentinel-2 on measuring shoreline change at a regional scale

In monitoring shoreline change, remote sensing data plays a pivotal role. Although different freely accessible satellite data are used in studies of shoreline change at various scales, it is imperative to identify their effectiveness at a regional scale. This study aims to determine how effectively MODIS, Landsat 8, and Sentinel-2 satellite images measure regional shoreline change. Digital Shoreline Analysis System (DSAS), an ArcGIS extension, was utilized to analyze the changes in the shoreline of Ganges deltaic coast in Bangladesh. The Shoreline Change Envelope (SCE) technique was applied to compare the shorelines extracted from MODIS, Landsat 8, and Sentinel-2 for the year 2016 with high-resolution Google Earth image. Next, using the End Point Rate (EPR) method, shoreline change rates between 2016 and 2019 were calculated separately from each satellite data. The study points out which satellite data was more accurate in delineating shoreline. Due to a higher spatial resolution, Sentinel-2 data were more effective than Landsat and MODIS data at a regional scale. In addition, MODIS, Landsat 8, and Sentinel-2 data indicate that the Ganges deltaic coast experiences landward movements of 54.65 m/yr, 12.56 m/yr, and 1.79 m/yr, respectively. Hence, it can be concluded that if the spectral resolution is coarse, shoreline change is overestimated. Therefore, in order to analyze or predict shoreline change (even at a regional scale), priority should be given to using images with fine spatial resolution. The spectral analysis of MODIS, Landsat and Sentinel showed a similar pattern of separability among the land cover classes observed along the Ganges deltaic coast. The separability is higher in NIR (Near Infrared) and SWIR (Short Wave Infrared) bands. This study will direct researchers in the selection of remote sensing data sources for shoreline change detection within a regional scale.

Temporal Sequence Object-based CNN (TS-OCNN) for crop classification from fine resolution remote sensing image time-series

Accurate crop distribution mapping is required for crop yield prediction and field management. Due to rapid progress in remote sensing technology, fine spatial resolution (FSR) remotely sensed imagery now offers great opportunities for mapping crop types in great detail. However, within-class variance can hamper attempts to discriminate crop classes at fine resolutions. Multi-temporal FSR remotely sensed imagery provides a means of increasing crop classification from FSR imagery, although current methods do not exploit the available information fully. In this research, a novel Temporal Sequence Object-based Convolutional Neural Network (TS-OCNN) was proposed to classify agricultural crop type from FSR image time-series. An object-based CNN (OCNN) model was adopted in the TS-OCNN to classify images at the object level (i.e., segmented objects or crop parcels), thus, maintaining the precise boundary information of crop parcels. The combination of image time-series was first utilized as the input to the OCNN model to produce an ‘original’ or baseline classification. Then the single-date images were fed automatically into the deep learning model scene-by-scene in order of image acquisition date to increase successively the crop classification accuracy. By doing so, the joint information in the FSR multi-temporal observations and the unique individual information from the single-date images were exploited comprehensively for crop classification. The effectiveness of the proposed approach was investigated using multitemporal SAR and optical imagery, respectively, over two heterogeneous agricultural areas. The experimental results demonstrated that the newly proposed TS-OCNN approach consistently increased crop classification accuracy, and achieved the greatest accuracies (82.68% and 87.40%) in comparison with state-of-the-art benchmark methods, including the object-based CNN (OCNN) (81.63% and 85.88%), object-based image analysis (OBIA) (78.21% and 84.83%), and standard pixel-wise CNN (79.18% and 82.90%). The proposed approach is the first known attempt to explore simultaneously the joint information from image time-series with the unique information from single-date images for crop classification using a deep learning framework. The TS-OCNN, therefore, represents a new approach for agricultural landscape classification from multi-temporal FSR imagery. Besides, it is readily generalizable to other landscapes (e.g., forest landscapes), with a wide application prospect.

Errors in global cloud climatology due to transect sampling with the CALIPSO satellite lidar mission

Although cloud profiling lidars provide the most accurate data on cloud amount (CA), cloud top height (CTH) and cloud optical thickness (COT), they only sample cloud fields along a one-dimensional transect. This approach introduces uncertainty into studies that interpret cloud layers as two- or three-dimensional. The present study verified the accuracy of mean annual CA, COT, and CTH when calculated based on transect sampling, and when spatiotemporally averaged over grid cells measuring 1°, 2.5°, and 5° in latitude and longitude. The study evaluates the sampling scheme used by the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) lidar mission. In this study for the very first time, transect observations are investigated globally, at the level of climatological means, and for actual cloud regimes. Lidar data were simulated based on MODIS observations for 2005, resulting in 3,762,678 individual laser transects, and 58,647 annual means for each parameter under study. The results of this study demonstrate that, for most locations on Earth, CALIPSO sampling does report ‘true’ mean annual CA, COT, and CTH. At 1° resolution, statistically significant (α ≤ 0.05) differences in CA were only noted for 11% of cells, and for 2–3% of cells in the case of COT and CTH. Polar regions were identified as the most affected by transect sampling. Mean global errors (RMSE) observed at the annual timescale were within or close to the accuracy expected for satellite data used in climate change studies: 1.9% (CA), 2.4% (COT), and 187 m (CTH). This study also found that the spatial resolution of the grid impacts the magnitude of errors: an increase in cell size increased the underestimation of cloud parameters, increased the extent of areas where statistically significant errors/ differences occurred, and increased uncertainty in the climatological mean estimation (wider confidence intervals). However, at the cell level, errors were up to ±5% (CA), ±25% (COT) and ± 400 m (CTH); moreover, the magnitude of error was difficult to estimate a priori due to the noisy spatial distribution. These findings suggest that a dedicated uncertainty analysis should be made a requirement in CALIPSO lidar-based studies at fine spatial resolution.

Elongatable Gripper Fingers With Integrated Stretchable Tactile Sensors for Underactuated Grasping and Dexterous Manipulation

The ability to grasp a wider range of objects in size and shape directly relates to the performance of robotic grippers. Adapting to complex geometries of objects requires large degrees of freedom to allow complex configurations. However, complexity in controlling many individual joints leads to introduction of underactuated mechanisms, in which traditional finger designs composed of revolute joints allow only flexion/extension motions. In this article, we propose a length-adjustable linkage mechanism in the underactuated finger controlled by an antagonistic tendon pair. The resulting gripper can elongate the fingers for an increased task space or shorten them for a finer spatial resolution. For tactile sensing, hyperelastic soft sensors are used to stretch with finger elongation. Contact pressures measured by the soft sensors are used in force-feedback control for which either the joint angles or the link lengths are adjusted. Lastly, a multimodal control scheme that combines elongation and flexion modes is demonstrated with tasks of dexterous manipulation.

Mapping Plant Species in a Former Industrial Site Using Airborne Hyperspectral and Time Series of Sentinel-2 Data Sets

Industrial activities induce various impacts on ecosystems that influence species richness and distribution. An effective way to assess the resulting impacts on biodiversity lies in vegetation mapping. Species classification achieved through supervised machine learning algorithms at the pixel level has shown promising results using hyperspectral images and multispectral, multitemporal images. This study aims to determine whether airborne hyperspectral images with a high spatial resolution or phenological information obtained by spaceborne multispectral time series (Sentinel-2) are suitable to discriminate species and assess biodiversity in a complex impacted context. The industrial heritage of the study site has indeed induced high spatial heterogeneity in terms of stressors and species over a reduced scale. First, vegetation indices, derivative spectra, continuum removed spectra, and components provided by three feature extraction techniques, namely, Principal Component Analysis, Minimal Noise Fraction, and Independent Component Analysis, were calculated from reflectance spectra. These features were then analyzed through Sequential Floating Feature Selection. Supervised classification was finally performed using various machine learning algorithms (Random Forest, Support Vector Machines, and Regularized Logistic Regression) considering a probability-based rejection approach. Biodiversity metrics were derived from resulted maps and analyzed considering the impacts. Average Overall Accuracy (AOA) reached up to 94% using the hyperspectral image and Regularized Logistic Regression algorithm, whereas the time series of multispectral images never exceeded 72% AOA. From all tested spectral transformations, only vegetation indices applied to the time series of multispectral images increased the performance. The results obtained with the hyperspectral image degraded to the specifications of Sentinel-2 emphasize the importance of fine spatial and spectral resolutions to achieve accurate mapping in this complex context. While no significant difference was found between impacted and reference sites through biodiversity metrics, vegetation mapping highlighted some differences in species distribution.

Aerial Imagery-Based Building Footprint Detection with an Integrated Deep Learning Framework: Applications for Fine Scale Wildland–Urban Interface Mapping

Human encroachment into wildlands has resulted in a rapid increase in wildland–urban interface (WUI) expansion, exposing more buildings and population to wildfire risks. More frequent mapping of structures and WUIs at a finer spatial resolution is needed for WUI characterization and hazard assessment. However, most approaches rely on high-resolution commercial satellite data with a particular focus on urban areas. We developed a deep learning framework tailored for building footprint detection in the transitional wildland–urban areas. We leveraged meter scale aerial imageries publicly available from the National Agriculture Imagery Program (NAIP) every 2 years. Our approach integrated Mobile-UNet and generative adversarial network. The deep learning models trained over three counties in California performed well in detecting building footprints across diverse landscapes, with an F1 score of 0.62, 0.67, and 0.75 in the interface WUI, intermix WUI, and rural regions, respectively. The bi-annual mapping captured both housing expansion and wildfire-caused building damages. The 30 m WUI maps generated from these finer footprints showed more granularity than the existing census tract-based maps and captured the transition of WUI dynamics well. More frequent updates of building footprint and improved WUI mapping will improve our understanding of WUI dynamics and provide guidance for adaptive strategies on community planning and wildfire hazard reduction.

Long-Term Variability of Wind Speed and Direction in the Mediterranean Basin

In this work, the ERA5 reanalysis dataset, with its fine spatial and temporal resolution, is used to study the wind speed and direction characteristics in the Mediterranean basin from 1979 to 2020. Their variability, trend and mean values, as well as their association, are examined in the monthly, annual and interannual/decadal time scales. The long-term variability of the wind direction is assessed using the angular variance, while the mean annual and interannual variability are used for the wind speed. The most characteristic and constant flow systems appear in the Gulf of Lion (Mistral) and the Aegean Sea during the summer (Etesian winds). The mean annual variability maximizes in the northern part of the basin, while the western part appears to have the most variable wind directions, while the fastest increasing wind speeds appear in the southwestern Levantine sub-basin. The long-term linear trend for the mean, the 95th and 99th quantiles of annual wind speed and the number of occurrences of extreme wind events have been also assessed. For particular areas, there is an increasing tendency in both intensity and frequency of occurrence of extreme wind events. Τhe linear-circular correlation coefficient has been implemented in order to quantify the linear association between the wind direction and the wind speed. This parameter demonstrates higher annual values for the Mistral wind in the Gulf of Lion and the Etesian winds in the Aegean Sea during the summer. Finally, the comparison of the results of ERA5 with the results of ERA-Interim highlighted significant differences in the Mediterranean wind speed and direction characteristics.

The Interplay between Spatial Urban Expansion and Morphologic Landscapes East of Cairo, Egypt Using Time Series Satellite Imagery

This study aims to monitor the magnitudes, rates, and directions of spatial urban expansions east of Cairo and their interactions with the initial morphologic landscapes. The approach relies on using CORONA satellite images acquired in 1969 with fine spatial resolution and time series images of Landsat and ASTER from 1984 to 2020. The CORONA images enable retrieval of the initial morphologic components, whereas the Landsat and ASTER images enable the spatial urban expansions to be mapped. The magnitudes of spatial urban expansions have been massive, in the order of 165 km2. These expansions have occurred through four main temporal phases with different spatial extents, rates, and directions in response to common urban policies and socioeconomic settings. Assessing the interactions between urban expansions and the morphology of watersheds in the study area indicates that the directions of urban expansion have been opposite to the geospatial orientations of the watersheds. In addition, significant urban areas in the order of ~8 km2 are under the direct threat of flash floods because they are misplaced within the valley floors of the studied watersheds. The study concludes that successful spatial urban expansion should consider the morphologic characteristics of the initial landscape for the purpose of maximizing interests and avoiding or reducing potential hazards against urban settlements.

Towards a novel high-spatial-resolution satellite rainfall product from C-band SAR Sentinel-1 over Central South Korea based on a bottom-up approach

ABSTRACT Finer spatial resolution (roughly 1 km) of terrestrial rainfall is necessary for urban hydrology and microscale agriculture demands but still limited in various ‘top-down’ satellite rainfall products (SRPs). This study aims to evaluate the potential application of an emerged ‘bottom-up’ approach, the parameterized SM2RAIN model, to the C-band Synthetic Aperture Radar Sentinel-1 (S1) satellites for a high-spatial-resolution SRP (SM2RAIN-S1, 0.01º/6-day) over Central South Korea. By comparing its performance with the Integrated Multi-satellitE Retrieval for Global Precipitation Measurement (IMERG) Early Run product at a similar resolution against a common reanalysis rainfall product and rain gauges network during a 1-year period, we explore whether this novel bottom-up SRP can add improvements to the conventional top-down SRP regarding differences in climate areas (coast and inland), vegetation covers (cropland and mixed forest), and seasonality (rainy and non-rainy seasons). The SM2RAIN-S1 demonstrates its feasibility in capturing general spatiotemporal rainfall patterns in the study area compared to other rainfall products; however, noises associated with the S1 images still occurred. For the comparison over climate-vegetation areas, the extended triple collocation revealed that the SM2RAIN-S1 is relatively superior to the IMERG mainly over inland and forest regions, suggesting the suitability of 6-day SM2RAIN-S1 data to the stable and lower temporal rainfall variation in such regions, where the IMERG data are limited. The comparison against rain gauge data over rainfall seasons indicated that the SM2RAIN-S1 generally outperformed IMERG for bias and Root-Mean-Square Error, particularly during the rainy season. Moreover, the SM2RAIN-S1 shows its feasibility for moderate and heavy rain detection, while it is less effective in detecting light rain events compared to the IMERG. This study encourages the potential use of the bottom-up SM2RAIN-S1 rainfall product to complement, downscale, and correct the biases in top-down SRPs; together with supporting their integration for improving high spatio-temporal resolution rainfall products.

The Impact of Exposure Measurement Error on the Estimated Concentration-Response Relationship between Long-Term Exposure to and Mortality.

Background:Exposure measurement error is a central concern in air pollution epidemiology. Given that studies have been using ambient air pollution predictions as proxy exposure measures, the potential impact of exposure error on health effect estimates needs to be comprehensively assessed.Objectives:We aimed to generate wide-ranging scenarios to assess direction and magnitude of bias caused by exposure errors under plausible concentration–response relationships between annual exposure to fine particulate matter [PM in aerodynamic diameter ()] and all-cause mortality.Methods:In this simulation study, we use daily predictions at spatial resolution to estimate annual exposures and their uncertainties for ZIP Codes of residence across the contiguous United States between 2000 and 2016. We consider scenarios in which we vary the error type (classical or Berkson) and the true concentration–response relationship between exposure and mortality (linear, quadratic, or soft-threshold—i.e., a smooth approximation to the hard-threshold model). In each scenario, we generate numbers of deaths using error-free exposures and confounders of concurrent air pollutants and neighborhood-level covariates and perform epidemiological analyses using error-prone exposures under correct specification or misspecification of the concentration–response relationship between exposure and mortality, adjusting for the confounders.Results:We simulate 1,000 replicates of each of 162 scenarios investigated. In general, both classical and Berkson errors can bias the concentration–response curve toward the null. The biases remain small even when using three times the predicted uncertainty to generate errors and are relatively larger at higher exposure levels.Discussion:Our findings suggest that the causal determination for long-term exposure and mortality is unlikely to be undermined when using high-resolution ambient predictions given that the estimated effect is generally smaller than the truth. The small magnitude of bias suggests that epidemiological findings are relatively robust against the exposure error. In practice, the use of ambient predictions with a finer spatial resolution will result in smaller bias. https://doi.org/10.1289/EHP10389

Approximating Three‐Dimensional (3‐D) Transport of Atmospheric Pollutants via Deep Learning

AbstractThe physical transport process is the bottleneck of the computational efficiency in regional chemical transport modeling. The issue will be worse with the smaller time step due to increased iterations required with finer spatial resolution at scale. Reported surrogates of the transport process are usually unfeasible according to integrated assessment of efficiency promotion, long‐term consistency, and spatial dimensions. This study intended to approximate the three‐dimensional (3‐D) transport process (including advection and diffusion) of a state‐of‐the‐art chemical transport model, that is, Models 3/Community Multiscale Air Quality (CMAQ), via the U‐Net structure of deep learning. Two temporal resolutions of models with 1‐hr and 5‐min were developed. Validation results indicated that single‐step R squared of both models were higher than 0.9, and the lifetime for continuous running was 400 and 1,000 steps for 1‐hr and 5‐min model, respectively. Meanwhile, the computational efficiency can be promoted with the maximum of 164 times for 1‐hr and 14 times for 5‐min resolution on one GPU. The 1‐hr deep learning surrogate could still achieve 12 times acceleration on the same CPU configurations of CMAQ, mainly through the end‐to‐end direct inferring rather than time step iterations. This study preliminarily proves the feasibility of the data‐driven approach in approximating the 3‐D complex transport process of atmospheric pollutants. Furthermore, computational efficiency can be efficiently improved while maintaining consistency and accuracy. Rapid transport simulation of different pollutants with wide concentration range can be expected, which will finally benefit the acceleration of whole chemical transport modeling.

Advancing early warning capabilities with CHIRPS-compatible NCEP GEFS precipitation forecasts

CHIRPS-GEFS is an operational data set that provides daily bias-corrected forecasts for next 1-day to ~15-day precipitation totals and anomalies at a quasi-global 50-deg N to 50-deg S extent and 0.05-degree resolution. These are based on National Centers for Environmental Prediction (NCEP) Global Ensemble Forecast System version 12 (GEFS v12) precipitation forecasts. CHIRPS-GEFS forecasts are compatible with Climate Hazards center InfraRed Precipitation with Stations (CHIRPS) data, which is actively used for drought monitoring, early warning, and near real-time impact assessments. A rank-based quantile matching procedure is used to transform GEFS v12 “reforecast” and “real-time” forecast ensemble means to CHIRPS spatial-temporal characteristics. Matching distributions to CHIRPS makes forecasts better reflect local climatology at finer spatial resolution and reduces moderate-to-large forecast errors. As shown in this study, having a CHIRPS-compatible version of the latest generation of NCEP GEFS forecasts enables rapid assessment of current forecasts and local historical context. CHIRPS-GEFS effectively bridges the gap between observations and weather predictions, increasing the value of both by connecting monitoring resources (CHIRPS) with interoperable forecasts.

Effect of aggregation and disaggregation of land surface temperature imagery on evapotranspiration estimation

Understanding the spatial distribution of evapotranspiration (ET) and other surface energy fluxes in tree crops is vital for optimizing precision irrigation management practices. The land surface temperature (LST) is the primary input to estimate ET at field scale, but currently, available satellite data have a coarse resolution, which is not suitable for estimating ET at the required spatiotemporal scale at the field level. The development of unmanned aerial vehicles (UAVs) or airborne platforms during the last two decades has enabled the acquisition of LST and subsequent derivation of ET at the finer spatial resolution of up to a few centimeters due to the advances in low-weight thermal cameras onboard these platforms. In this study, the disaggregation methodology was applied to Landsat imagery based on aerial imagery collected at 0.5 m resolution over almond orchards in the Sacramento Valley of California, USA, to estimate ET at the orchard scale. The TsHARP algorithm was applied to sharpen the Landsat LST images (30 m spatial resolution) to target resolutions of 0.5, 10, 20 m. Also, the simple average aggregation technique was used to aggregate the aerial imagery (0.5 resolution) to 10, 20, and 30 m scales. Two Source Energy Balance (TSEB) model was used to estimate spatially distributed ET fluxes by combining the ground-based meteorological data for aggregated and disaggregated LST images. The estimated ET fluxes were compared to the measured eddy covariance fluxes, and then the aggregation and disaggregation processes were evaluated at 0.5, 10, 20, and 30 m spatial scales. At four spatial scales (0.5, 10, 20, and 30 m) aggregation and disaggregation derived ET values were highly correlated with the measured eddy covariance fluxes within the source footprint area. Furthermore, both aggregation and disaggregation approaches demonstrated similar ET spatial patterns within the study area, with slight variations observed under the Landsat 30 m resolution. The results indicated that TsHARP downscaled high-resolution imagery provides optimal spatial resolution for mapping and analyzing ET at the plant to orchard scale, with a 95% confidence interval and an RMSE of 0.43 mm/day. Overall, the results suggest that high spatial resolution imagery disaggregated from Landsat was the best alternative for estimating ET fluxes from the plant to the orchard scale, which is beneficial for precision irrigation management in orchards, but the framework can also be applied to other precision agriculture applications.

Exploring Solar Energy Potential using Solar Radiation GIS Toolset - A Case of College of Natural Resources, Lobesa, Punakha

This study aimed to explore solar energy potential on a small landscape (campus of the College of Natural Resources) using topography of fine spatial resolution that can detect number of rooftops feasible for harnessing solar energy. Digital elevation model was derived by extracting elevation values from the Google Earth Pro which was then used as input for solar radiation calculation. The total radiation that the campus received was 2,100 kWh over 259 days and the result was validated using power access viewer and global solar atlas databases. The monthly average radiation indicated that October, November, December and January are the radiation peaking months at the college campus. Twenty two rooftops were identified to receive more than or equal to 3.08 kWh/m2/day, while the remaining rooftops receiving less than or equal to 3.08 kWh/m2/day. The college has potential of supplementing power shortage during lean hydro season using the rooftop solar panels.

Evaluation of Vulnerability Status of the Infection Risk to COVID-19 Using Geographic Information Systems (GIS) and Multi-Criteria Decision Analysis (MCDA): A Case Study of Addis Ababa City, Ethiopia.

COVID-19 is a disease caused by a new coronavirus called SARS-CoV-2 and is an accidental global public health threat. Because of this, WHO declared the COVID-19 outbreak a pandemic. The pandemic is spreading unprecedently in Addis Ababa, which results in extraordinary logistical and management challenges in response to the novel coronavirus in the city. Thus, management strategies and resource allocation need to be vulnerability-oriented. Though various studies have been carried out on COVID-19, only a few studies have been conducted on vulnerability from a geospatial/location-based perspective but at a wider spatial resolution. This puts the results of those studies under question while their findings are projected to the finer spatial resolution. To overcome such problems, the integration of Geographic Information Systems (GIS) and Multi-Criteria Decision Analysis (MCDA) has been developed as a framework to evaluate and map the susceptibility status of the infection risk to COVID-19. To achieve the objective of the study, data like land use, population density, and distance from roads, hospitals, bus stations, the bank, markets, COVID-19 cases, health care units, and government offices are used. The weighted overlay method was used; to evaluate and map the susceptibility status of the infection risk to COVID-19. The result revealed that out of the total study area, 32.62% (169.91 km2) falls under the low vulnerable category (1), and the area covering 40.9% (213.04 km2) under the moderate vulnerable class (2) for infection risk of COVID-19. The highly vulnerable category (3) covers an area of 25.31% (132.85 km2), and the remaining 1.17% (6.12 km2) is under an extremely high vulnerable class (4). Thus, these priority areas could address pandemic control mechanisms like disinfection regularly. Health sector professionals, local authorities, the scientific community, and the general public will benefit from the study as a tool to better understand pandemic transmission centers and identify areas where more protective measures and response actions are needed at a finer spatial resolution.

A 30 m annual maize phenology dataset from 1985 to 2020 in China

Abstract. Crop phenology indicators provide essential information on crop growth phases, which are highly required for agroecosystem management and yield estimation. Previous crop phenology studies were mainly conducted using coarse-resolution (e.g., 500 m) satellite data, such as the moderate resolution imaging spectroradiometer (MODIS) data. However, precision agriculture requires higher resolution phenology information of crops for better agroecosystem management, and this requirement can be met by long-term and fine-resolution Landsat observations. In this study, we generated the first national maize phenology product with a fine spatial resolution (30 m) and a long temporal span (1985–2020) in China, using all available Landsat images on the Google Earth Engine (GEE) platform. First, we extracted long-term mean phenological indicators using the harmonic model, including the v3 (i.e., the date when the third leaf is fully expanded) and the maturity phases (i.e., when the dry weight of maize grains first reaches the maximum). Second, we identified the annual dynamics of phenological indicators by measuring the difference in dates when the vegetation index in a specific year reaches the same magnitude as its long-term mean. The derived maize phenology datasets are consistent with in situ observations from the agricultural meteorological stations and the PhenoCam network. Besides, the derived fine-resolution phenology dataset agrees well with the MODIS phenology product regarding the spatial patterns and temporal dynamics. Furthermore, we observed a noticeable difference in maize phenology temporal trends before and after 2000, which is likely attributable to the changes in temperature and precipitation, which further altered the farming activities. The extracted maize phenology dataset can support precise yield estimation and deepen our understanding of the future agroecosystem response to global warming. The data are available at https://doi.org/10.6084/m9.figshare.16437054 (Niu et al., 2021).

Multi-Temporal LiDAR and Hyperspectral Data Fusion for Classification of Semi-Arid Woody Cover Species

Mapping the spatial distribution of woody vegetation is important for monitoring, managing, and studying woody encroachment in grasslands. However, in semi-arid regions, remotely sensed discrimination of tree species is difficult primarily due to the tree similarities, small and sparse canopy cover, but may also be due to overlapping woody canopies as well as seasonal leaf retention (deciduous versus evergreen) characteristics. Similar studies in different biomes have achieved low accuracies using coarse spatial resolution image data. The objective of this study was to investigate the use of multi-temporal, airborne hyperspectral imagery and light detection and ranging (LiDAR) derived data for tree species classification in a semi-arid desert region. This study produces highly accurate classifications by combining multi-temporal fine spatial resolution hyperspectral and LiDAR data (~1 m) through a reproducible scripting and machine learning approach that can be applied to larger areas and similar datasets. Combining multi-temporal vegetation indices and canopy height models led to an overall accuracy of 95.28% and kappa of 94.17%. Five woody species were discriminated resulting in producer accuracies ranging from 86.12% to 98.38%. The influence of fusing spectral and structural information in a random forest classifier for tree identification is evident. Additionally, a multi-temporal dataset slightly increases classification accuracies over a single data collection. Our results show a promising methodology for tree species classification in a semi-arid region using multi-temporal hyperspectral and LiDAR remote sensing data.

Spatial Disaggregation of Population Subgroups Leveraging Self-Trained Multi-Output Gradient Boosting Regression Trees

Abstract. Accurate and consistent estimations on the present and future population distribution, at fine spatial resolution, are fundamental to support a variety of activities. However, the sampling regime, sample size, and methods used to collect census data are heterogeneous across temporal periods and/or geographic regions. Moreover, the data is usually only made available in aggregated form, to ensure privacy. In an attempt to address these issues, several previous initiatives have addressed the use of spatial disaggregation methods to produce high-resolution gridded datasets describing the human population distribution, although these projects have usually not addressed specific population subgroups. This paper describes a spatial disaggregation method based on self-training regression models, innovating over previous studies in the simultaneous prediction of disaggregated counts for multiple inter-related variables, by leveraging multi-output models based on gradient tree boosting. We report on experiments for two case studies, using high-resolution data (i.e., counts for different subgroups available at a resolution of 100 meters) for the municipality of Amsterdam and the region of Greater Copenhagen. Results show that the proposed approach can capture spatial heterogeneity and the dependency on local factors, outperforming alternatives (e.g., seminal disaggregation algorithms, or approaches leveraging individual regression models for each variable) in terms of averaged error metrics, and also upon visual inspection of spatial variation in the resulting maps.

Identification of temporary livestock enclosures in Kenya from multi-temporal PlanetScope imagery

The use of night-time livestock enclosures, often referred to as “bomas”, “corrals”, or “kraals”, is a common practice across African rangelands. Bomas protect livestock from predation by wildlife and potential theft. Due to the concentration of animal faeces inside bomas, they not only become nutrient-rich patches that can add to biodiversity, but also hotspots for the emission of nitrous oxide (N2O), an important greenhouse gas, especially when animals are kept inside for long periods. To provide an accurate estimate of such emissions for wider landscapes, bomas need to be accounted for. Moreover, initial experiments indicated that more frequent shifts in the boma locations could help to reduce N2O emissions. This stresses the need for better understanding where bomas are located, their numbers, as well as when they are actively used. Given the recent advances in satellite technology, resulting in high-frequent spectral measurements at fine spatial resolution, solutions to address these needs are now within reach. This study is a first effort to map and monitor the appearance of bomas with the use of satellite image time series. Our main dataset was a dense times series of 3 m resolution PlanetScope multispectral imagery. In addition, a reference dataset of boma and non-boma locations was created using GPS-collar tracking data and 0.5 m resolution Pléiades imagery. The reduction of vegetation cover and increase of organic material following boma installation result in typical spectral changes when contrasted against its surroundings. Based on these spectral changes we devised an empirical approach to infer approximate boma installation dates from PlanetScope's near infrared (NIR) band and used our reference dataset for setting optimal parameter values. A NIR spatial difference index resulted in clear temporal patterns, which were more apparent during the wet season. At landscape scale our approach reveals clear spatio-temporal patterns of boma installation, which could not be revealed from less frequent sub-meter resolution imagery alone. While further improvements are possible, we show that small-sized (150–500 m2) temporary surface changes, such as those that occur when pastoralists use mobile bomas, can be detected with dense image time series like those offered by the PlanetScope constellation. In future, this could lead to better assessment of a) spatio-temporal livestock distribution, b) the contribution of bomas to N2O emissions and soil fertility at landscape scale, and c) the uptake of enclosure rotation practices at large spatial scales.

Tradeoffs between UAS Spatial Resolution and Accuracy for Deep Learning Semantic Segmentation Applied to Wetland Vegetation Species Mapping

Recent advances in image classification of fine spatial resolution imagery from unoccupied aircraft systems (UASs) have allowed for mapping vegetation based on both multispectral reflectance and fine textural details. Convolutional neural network (CNN)-based models can take advantage of the spatial detail present in UAS imagery by implicitly learning shapes and textures associated with classes to produce highly accurate maps. However, the spatial resolution of UAS data is infrequently examined in CNN classification, and there are important tradeoffs between spatial resolution and classification accuracy. To improve the understanding of the relationship between spatial resolution and classification accuracy for a CNN-based model, we captured 7.6 cm imagery with a UAS in a wetland environment containing graminoid (grass-like) plant species and simulated a range of spatial resolutions up to 76.0 cm. We evaluated two methods for the simulation of coarser spatial resolution imagery, averaging before and after orthomosaic stitching, and then trained and applied a U-Net CNN model for each resolution and method. We found untuned overall accuracies exceeding 70% at the finest spatial resolutions, but classification accuracy decreased as spatial resolution coarsened, particularly beyond a 22.8 cm resolution. Coarsening the spatial resolution from 7.6 cm to 22.8 cm could permit a ninefold increase in survey area, with only a moderate reduction in classification accuracy. This study provides insight into the impact of the spatial resolution on deep learning semantic segmentation performance and information that can potentially be useful for optimizing precise UAS-based mapping projects.