Earth Observation Applications Research Articles

A qualitative study of Machine Learning practices and engineering challenges in Earth Observation

Abstract Machine Learning (ML) is ubiquitously on the advance. Like many domains, Earth Observation (EO) also increasingly relies on ML applications, where ML methods are applied to process vast amounts of heterogeneous and continuous data streams to answer socially and environmentally relevant questions. However, developing such ML- based EO systems remains challenging: Development processes and employed workflows are often barely structured and poorly reported. The application of ML methods and techniques is considered to be opaque and the lack of transparency is contradictory to the responsible development of ML-based EO applications. To improve this situation a better understanding of the current practices and engineering-related challenges in developing ML-based EO applications is required. In this paper, we report observations from an exploratory study where five experts shared their view on ML engineering in semi-structured interviews. We analysed these interviews with coding techniques as often applied in the domain of empirical software engineering. The interviews provide informative insights into the practical development of ML applications and reveal several engineering challenges. In addition, interviewees participated in a novel workflow sketching task, which provided a tangible reflection of implicit processes. Overall, the results confirm a gap between theoretical conceptions and real practices in ML development even though workflows were sketched abstractly as textbook-like. The results pave the way for a large-scale investigation on requirements for ML engineering in EO.

Reviews and syntheses: Ongoing and emerging opportunities to improve environmental science using observations from the Advanced Baseline Imager on the Geostationary Operational Environmental Satellites

Abstract. Environmental science is increasingly reliant on remotely sensed observations of the Earth's surface and atmosphere. Observations from polar-orbiting satellites have long supported investigations on land cover change, ecosystem productivity, hydrology, climate, the impacts of disturbance, and more and are critical for extrapolating (upscaling) ground-based measurements to larger areas. However, the limited temporal frequency at which polar-orbiting satellites observe the Earth limits our understanding of rapidly evolving ecosystem processes, especially in areas with frequent cloud cover. Geostationary satellites have observed the Earth's surface and atmosphere at high temporal frequency for decades, and their imagers now have spectral resolutions in the visible and near-infrared regions that are comparable to commonly used polar-orbiting sensors like the Moderate Resolution Imaging Spectroradiometer (MODIS), Visible Infrared Imaging Radiometer Suite (VIIRS), or Landsat. These advances extend applications of geostationary Earth observations from weather monitoring to multiple disciplines in ecology and environmental science. We review a number of existing applications that use data from geostationary platforms and present upcoming opportunities for observing key ecosystem properties using high-frequency observations from the Advanced Baseline Imagers (ABI) on the Geostationary Operational Environmental Satellites (GOES), which routinely observe the Western Hemisphere every 5–15 min. Many of the existing applications in environmental science from ABI are focused on estimating land surface temperature, solar radiation, evapotranspiration, and biomass burning emissions along with detecting rapid drought development and wildfire. Ongoing work in estimating vegetation properties and phenology from other geostationary platforms demonstrates the potential to expand ABI observations to estimate vegetation greenness, moisture, and productivity at a high temporal frequency across the Western Hemisphere. Finally, we present emerging opportunities to address the relatively coarse resolution of ABI observations through multisensor fusion to resolve landscape heterogeneity and to leverage observations from ABI to study the carbon cycle and ecosystem function at unprecedented temporal frequency.

Towards Streamlined Single-Image Super-Resolution: Demonstration with 10 m Sentinel-2 Colour and 10–60 m Multi-Spectral VNIR and SWIR Bands

Higher spatial resolution imaging data are considered desirable in many Earth observation applications. In this work, we propose and demonstrate the TARSGAN (learning Terrestrial image deblurring using Adaptive weighted dense Residual Super-resolution Generative Adversarial Network) system for Super-resolution Restoration (SRR) of 10 m/pixel Sentinel-2 “true” colour images as well as all the other multispectral bands. In parallel, the ELF (automated image Edge detection and measurements of edge spread function, Line spread function, and Full width at half maximum) system is proposed to achieve automated and precise assessments of the effective resolutions of the input and SRR images. Subsequent ELF measurements of the TARSGAN SRR results suggest an averaged effective resolution enhancement factor of about 2.91 times (equivalent to ~3.44 m/pixel for the 10 m/pixel bands) given a nominal SRR upscaling factor of 4 times. Several examples are provided for different types of scenes from urban landscapes to agricultural scenes and sea-ice floes.

Earth Observation and Economic Studies: A Cross-fertilization Perspective

Over the past decade, an increasing number of satellite images and other earth observation (EO) data have become available to a wide range of final users, including economists and other social scientists, boosting the amount of information they can obtain to study, analyze, and manage different research topics in their fields. In turn, economists are promoting advancement in the quantification of the benefits related to the increasing use of EO services and applications by governments, firms, and citizens. Drawing from recent literature, and our own research, we explore, on one side, how social scientists, and particularly economists, can significantly benefit from EO data, and conversely, how socio-economic impact studies of EO can take advantage from a specific set of economic methods. This article suggests that cross-fertilization and interplay between economics, social sciences, and remote sensing science are needed to advance understanding of our societies and to more rigorously evaluate the socio-economic impact of EO services and applications. The article suggests possible new research avenues.

Sea Ice Thickness Estimation Based on Regression Neural Networks Using L-Band Microwave Radiometry Data from the FSSCat Mission

Several methods have been developed to provide polar maps of sea ice thickness (SIT) from L-band brightness temperature (TB) and altimetry data. Current process-based inversion methods to yield SIT fail to address the complex surface characteristics because sea ice is subject to strong seasonal dynamics and ice-physical properties are often non-linearly related. Neural networks can be trained to find hidden links among large datasets and often perform better on convoluted problems for which traditional approaches miss out important relationships between the observations. The FSSCat mission launched on 3 September 2020, carries the Flexible Microwave Payload-2 (FMPL-2), which contains the first Reflected Global Navigation Satellite System (GNSS-R) and L-band radiometer on board a CubeSat—designed to provide TB data on global coverage for soil moisture retrieval, and sea ice applications. This work investigates a predictive regression neural network approach with the goal to infer SIT using FMPL-2 TB and ancillary data (sea ice concentration, surface temperature, and sea ice freeboard). Two models—covering thin ice up to 0.6 m and full-range thickness—were separately trained on Arctic data in a two-month period from mid-October to the beginning of December 2020, while using ground truth data derived from the Soil Moisture and Ocean Salinity (SMOS) and Cryosat-2 missions. The thin ice and the full-range models resulted in a mean absolute error of 6.5 cm and 23 cm, respectively. Both of the models allowed for one to produce weekly composites of Arctic maps, and monthly composites of Antarctic SIT were predicted based on the Arctic full-range model. This work presents the first results of the FSSCat mission over the polar regions. It reveals the benefits of neural networks for sea ice retrievals and demonstrates that moderate-cost CubeSat missions can provide valuable data for applications in Earth observation.

Distributed Deep Learning for Remote Sensing Data Interpretation

As a newly emerging technology, deep learning (DL) is a very promising field in big data applications. Remote sensing often involves huge data volumes obtained daily by numerous in-orbit satellites. This makes it a perfect target area for data-driven applications. Nowadays, technological advances in terms of software and hardware have a noticeable impact on Earth observation applications, more specifically in remote sensing techniques and procedures, allowing for the acquisition of data sets with greater quality at higher acquisition ratios. This results in the collection of huge amounts of remotely sensed data, characterized by their large spatial resolution (in terms of the number of pixels per scene), and very high spectral dimensionality, with hundreds or even thousands of spectral bands. As a result, remote sensing instruments on spaceborne and airborne platforms are now generating data cubes with extremely high dimensionality, imposing several restrictions in terms of both processing runtimes and storage capacity. In this article, we provide a comprehensive review of the state of the art in DL for remote sensing data interpretation, analyzing the strengths and weaknesses of the most widely used techniques in the literature, as well as an exhaustive description of their parallel and distributed implementations (with a particular focus on those conducted using cloud computing systems). We also provide quantitative results, offering an assessment of a DL technique in a specific case study (source code available: https://github.com/mhaut/cloud-dnn-HSI). This article concludes with some remarks and hints about future challenges in the application of DL techniques to distributed remote sensing data interpretation problems. We emphasize the role of the cloud in providing a powerful architecture that is now able to manage vast amounts of remotely sensed data due to its implementation simplicity, low cost, and high efficiency compared to other parallel and distributed architectures, such as grid computing or dedicated clusters.

Automatic land cover classification of multi-resolution dualpol data using convolutional neural network (CNN)

Synthetic Aperture Radar is an interesting topic of research for scientists & researchers as it is associated with polarimetric information which helps to detect surface & subsurface features of land, sea, and ice. Classical techniques include the use of polarimetric information to simplify SAR image interpretation and to classify it for various earth observation applications. The deep learning (DL) techniques like Convolutional Neural Network (CNN), extract useful information from an image (here dual polarimetric SAR dataset) about the land surface to segment or classify the dataset for various earth applications. In the current research paper convolutional neural network is used to automatically classify RISAT-1 dataset over the Mumbai region for land cover classification. Also impact of patch size variation was studied. In addition, the efficiency of the CNN model was tested using an approach similar to transfer learning approach on multi resolution images (different multilooked images) i.e. CNN was trained twice on different resolution images; one trained on coarser resolution and tested on comparatively higher resolution datasets and other vice versa. It was found that increasing the patch size for convolution classified the image more accurately but at the same time it smoothens the output image. Also, CNN model trained on low spatial resolution image predicted better the higher spatial resolution image as compared to the reverse scenario.

Potential for High Fidelity Global Mapping of Common Inland Water Quality Products at High Spatial and Temporal Resolutions Based on a Synthetic Data and Machine Learning Approach

There is currently a scarcity of paired in-situ aquatic optical and biogeophysical data for productive inland waters, which critically hinders our capacity to develop and validate robust retrieval models for Earth Observation applications. This study aims to address this limitation through the development of a novel synthetic dataset of top-of-atmosphere and bottom-of-atmosphere reflectances, which is the first to encompass the immense natural optical variability present in inland waters. Novel aspects of the synthetic dataset include: 1) physics-based, two-layered, size- and type-specific phytoplankton inherent optical properties (IOPs) for mixed eukaryotic/cyanobacteria assemblages; 2) calculations of mixed assemblage chlorophyll-a (chl-a) fluorescence; 3) modeled phycocyanin concentration derived from assemblage-based phycocyanin absorption; 4) and paired sensor-specific top-of-atmosphere reflectances, including optically extreme cases and the contribution of green vegetation adjacency. The synthetic bottom-of-atmosphere reflectance spectra were compiled into 13 distinct optical water types similar to those discovered using in-situ data. Inspection showed similar relationships of concentrations and IOPs to those of natural waters. This dataset was used to calculate typical surviving water-leaving signal at top-of-atmosphere, and used to train and test four state-of-the-art machine learning architectures for multi-parameter retrieval and cross-sensor capability. Initial results provide reliable estimates of water quality parameters and IOPs over a highly dynamic range of water types, at various spectral and spatial sensor resolutions. The results of this work represent a significant leap forward in our capacity for routine, global monitoring of inland water quality.

Ultracold atom interferometry in space

Bose-Einstein condensates (BECs) in free fall constitute a promising source for space-borne interferometry. Indeed, BECs enjoy a slowly expanding wave function, display a large spatial coherence and can be engineered and probed by optical techniques. Here we explore matter-wave fringes of multiple spinor components of a BEC released in free fall employing light-pulses to drive Bragg processes and induce phase imprinting on a sounding rocket. The prevailing microgravity played a crucial role in the observation of these interferences which not only reveal the spatial coherence of the condensates but also allow us to measure differential forces. Our work marks the beginning of matter-wave interferometry in space with future applications in fundamental physics, navigation and earth observation.

FPGA implementation of the JPEG XR for onboard earth-observation applications

As a more recent option for low-complexity compression, the transform-based JPEG XR standard accepts an extended range of input formats and provides good low-loss compression. Based on the performance in software of the JPEG XR compression to process remote-sensing images and foreseeing applications onboard small earth-observation satellites, the JPEG XR algorithm was implemented and evaluated in FPGA. The VHDL code was developed and evaluated in two parts, the transform/quantization/prediction and the DC/LP/HP encoding, which were then merged into the complete JPEG XR description. The results of the real-time evaluation were compared to the predictive-differential JPEG LS compression, in a similar hardware test setup, and also to other JPEG XR hardware implementations, and showed up a good trade-off between a throughput of 16 Mpix/s and a power consumption of about 180 mW, at a 50 MHz clock frequency. Better results were estimated for a more recent FPGA version at a higher frequency.

Green LAI Mapping and Cloud Gap-Filling Using Gaussian Process Regression in Google Earth Engine.

For the last decade, Gaussian process regression (GPR) proved to be a competitive machine learning regression algorithm for Earth observation applications, with attractive unique properties such as band relevance ranking and uncertainty estimates. More recently, GPR also proved to be a proficient time series processor to fill up gaps in optical imagery, typically due to cloud cover. This makes GPR perfectly suited for large-scale spatiotemporal processing of satellite imageries into cloud-free products of biophysical variables. With the advent of the Google Earth Engine (GEE) cloud platform, new opportunities emerged to process local-to-planetary scale satellite data using advanced machine learning techniques and convert them into gap-filled vegetation properties products. However, GPR is not yet part of the GEE ecosystem. To circumvent this limitation, this work proposes a general adaptation of GPR formulation to parallel processing framework and its integration into GEE. To demonstrate the functioning and utility of the developed workflow, a GPR model predicting green leaf area index (LAIG) from Sentinel-2 imagery was imported. Although by running this GPR model into GEE any corner of the world can be mapped into LAIG at a resolution of 20 m, here we show some demonstration cases over western Europe with zoom-ins over Spain. Thanks to the computational power of GEE, the mapping takes place on-the-fly. Additionally, a GPR-based gap filling strategy based on pre-optimized kernel hyperparameters is also put forward for the generation of multi-orbit cloud-free LAIG maps with an unprecedented level of detail, and the extraction of regularly-sampled LAIG time series at a pixel level. The ability to plugin a locally-trained GPR model into the GEE framework and its instant processing opens up a new paradigm of remote sensing image processing.

An Optical–Thermal Surface–Atmosphere Radiative Transfer Model Coupling Framework With Topographic Effects

Most surface–atmosphere radiative transfer models (RTMs) work only for flat surfaces, with the exception being time-consuming 3-D scene-based models. The deficiency of flat-surface RTMs that do not consider topographic effects is that their applications in earth observation and simulation studies are impaired because rugged terrains make up approximately 24% of the global land surface. Another deficiency of most surface–atmosphere RTMs is that they model reflected and emitted (i.e., solar and thermal) radiative transfer processes separately, which limits RTMs in applications, such as fire detection. This study proposes a unified optical–thermal RTM coupling framework (RTM-CF) that considers topographic effects based on the four-stream approximation theory. The framework couples surface–atmosphere RTMs and can simultaneously simulate a set of parameters at the top-of-atmosphere (TOA) and bottom-of-atmosphere (BOA) levels from optical and thermal spectral ranges. These parameters include the TOA directional radiance/reflectance, TOA exitance/albedo, TOA net radiation, surface radiance/reflectance/albedo, surface downward/upward/net radiation, and FAPAR/APAR. The RTM-CF with topographic effects is compared with the well-known 3-D discrete anisotropic radiative transfer (DART) ray-tracing model and validated by field measurements from three steep sites. The evaluation results show that the simulated reflectance, radiance, and radiation fluxes are consistent with the DART results and the field data, with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R^{2}&gt;0.93$ </tex-math></inline-formula> and scatter points close to the 1:1 line for all parameters. In this RTM-CF, atmospheric and topographic effects are simultaneously incorporated, and the surface anisotropy is also effectively considered. This framework is highly modularized, which enables it to be easily adapted to different submodels.

Framework to Create Cloud-Free Remote Sensing Data Using Passenger Aircraft as the Platform

Cloud removal in optical remote sensing imagery is essential for many Earth observation applications. To recover the cloud obscured information, some preconditions must be satisfied. For example, the cloud must be semitransparent or relationships between contaminated and cloud-free pixels must be assumed. Due to the inherent imaging geometry features in satellite remote sensing, it is impossible to observe the ground under the clouds directly; therefore, cloud removal algorithms are always not perfect owing to the loss of ground truth. Recently, the use of passenger aircraft as a platform for remote sensing has been proposed by some researchers and institutes, including Airbus and the Japan Aerospace Exploration Agency. Passenger aircraft have the advantages of short visitation frequency and low cost. Additionally, because passenger aircraft fly at lower altitudes compared to satellites, they can observe the ground under the clouds at an oblique viewing angle. In this study, we examine the possibility of creating cloud-free remote sensing data by stacking multiangle images captured by passenger aircraft. To accomplish this, a processing framework is proposed, which includes four main steps: first, multiangle image acquisition from passenger aircraft, second, cloud detection based on deep learning semantic segmentation models, third, cloud removal by image stacking, and fourth, image quality enhancement via haze removal. This method is intended to remove cloud contamination without the requirements of reference images and predetermination of cloud types. The proposed method was tested in multiple case studies, wherein the resultant cloud- and haze-free orthophotos were visualized and quantitatively analyzed in various land cover type scenes. The results of the case studies demonstrated that the proposed method could generate high quality, cloud-free orthophotos. Therefore, we conclude that this framework has great potential for creating cloud-free remote sensing images when the cloud removal of satellite imagery is difficult or inaccurate.

Object Detection in Hyperspectral Images

High spectral resolution of hyperspectral images allows the detection and classification of materials in the observed images. However, existing research on hyperspectral detection mainly focuses on pixel-level study, partially due to the low spatial resolution in typical earth observation applications. With the development of imaging techniques, high-spatial-resolution hyperspectral data can be obtained and object-level detection is necessary for many applications. In this work, the object-based hyperspectral detection problem is formulated, and a convolutional neural network is then designed based on the specific characteristics of this problem. Moreover, a hyperspectral dataset with over 400 high-quality images for object-level target detection is created. Experimental results validate the proposed framework and show its superior performance.

Challenges in Using Earth Observation (EO) Data to Support Environmental Management in Brazil

This paper presents the results of research designed to explore the challenges involved in the use of Earth Observation (EO) data to support environmental management Brazil. While much has been written about the technology and applications of EO, the perspective of end-users of EO data and their needs has been under-explored in the literature. A total of 53 key informants in Brasilia and the cities of Rio Branco and Cuiaba were interviewed regarding their current use and experience of EO data and the expressed challenges that they face. The research builds upon a conceptual model which illustrates the main steps and limitations in the flow of EO data and information for use in the management of land use and land cover (LULC) in Brazil. The current paper analyzes and ranks, by relative importance, the factors that users identify as limiting their use of EO. The most important limiting factor for the end-user was the lack of personnel, followed by political and economic context, data management, innovation, infrastructure and IT, technical capacity to use and process EO data, bureaucracy, limitations associated with access to high-resolution data, and access to ready-to-use product. In general, users expect to access a ready-to-use product, transformed from the raw EO data into usable information. Related to this is the question of whether this processing is best done within an organization or sourced from outside. Our results suggest that, despite the potential of EO data for informing environmental management in Brazil, its use remains constrained by its lack of suitably trained personnel and financial resources, as well as the poor communication between institutions.

OpenStreetMap: Challenges and Opportunities in Machine Learning and Remote Sensing

OpenStreetMap (OSM) is a community-based, freely available, editable map service created as an alternative to authoritative sources. Given that it is edited mainly by volunteers with different mapping skills, the completeness and quality of its annotations are heterogeneous across different geographical locations. Despite that, OSM has been widely used in several applications in geosciences, Earth observation, and environmental sciences. In this article, we review recent methods based on machine learning to improve and use OSM data. Such methods aim to either 1) improve the coverage and quality of OSM layers, typically by using geographic information systems (GISs) and remote sensing technologies, or 2) use the existing OSM layers to train models based on image data to serve applications such as navigation and land use classification. We believe that OSM (as well as other sources of open land maps) can change the way we interpret remote sensing data and that the synergy with machine learning can scale participatory mapmaking and its quality to the level needed for global and up-to-date land mapping. A preliminary version of this manuscript was presented in [120].

Long-Term Remote Monitoring of Ground Deformation Using Sentinel-1 Interferometric Synthetic Aperture Radar (InSAR): Applications and Insights into Geotechnical Engineering Practices

Development of synthetic aperture radar (SAR) technology and the dedicated suite of processing tools have aided the evolution of remote sensing techniques for various Earth Observation (EO) applications. Interferometric SAR (InSAR) is a relatively new geodetic technique which provides high-speed and reliable geographic, geologic, and hazards information allowing the prognosis of future environmental and urban planning. In this study, we explored the applicability of two differential interferometry techniques, conventional and advanced differential InSAR (A-DInSAR), for topographic mapping and long-term geotechnical monitoring by exploiting satellite data, particularly Sentinel-1 SAR data, which is publicly shared. We specifically used the open-source tools of SeNtinel Application Platform (SNAP) and Stanford Method for Persistent Scatterers (StaMPS) for interferometric data processing to implement A-DInSAR. This study presents various applications, which include generation of a digital elevation model (DEM), mapping of seismically induced displacement and associated damages, and detection and long-term monitoring of tunneling-induced ground deformation and rainfall-induced landslide. Geometric and temporal decorrelations posed challenges and limitations in the successful implementation of Sentinel-1 SAR interferometry specifically in vegetated areas. The presented results proved the validity and reliability of the exploited SAR data and InSAR techniques for addressing geotechnical engineering related problems.

Compact Polarimetry Response to Modeled Fast Sea Ice Thickness

Compact Polarimetric (CP) Synthetic Aperture Radar (SAR) is expected to gain more and more ground for Earth observation applications in the coming years. This comes in light of the recently launched RADARSAT Constellation Mission (RCM), which uniquely provides CP SAR imagery in operational mode. In this study, we present observations about the sensitivity of CP SAR imagery to thickness of thermodynamically-grown fast sea ice during early ice growth (September–December 2017) in the Resolute Bay area, Canadian Central Arctic. Fast ice is most suitable to use for this preliminary study since it exhibits only thermodynamic growth in absence of ice mobility and deformation. Results reveal that ice thickness up to 30 cm can be retrieved using several CP parameters from the tested set. This ice thickness corresponds to the thickness of young ice. We found the surface scattering mechanism to be dominant during the early ice growth, exposing an increasing tendency up to 30 cm thickness with a correlation coefficient with the thickness equal to 0.86. The degree of polarization was found to be the parameter with the highest correlation up to 0.95. While thickness retrieval within the same range is also possible using parameters from Full Polarimetric (FP) SAR parameters as shown in previous studies, the advantage of using CP SAR mode is the much larger swath coverage, which is an operational requirement.

Object Detection and Image Segmentation with Deep Learning on Earth Observation Data: A Review—Part II: Applications

In Earth observation (EO), large-scale land-surface dynamics are traditionally analyzed by investigating aggregated classes. The increase in data with a very high spatial resolution enables investigations on a fine-grained feature level which can help us to better understand the dynamics of land surfaces by taking object dynamics into account. To extract fine-grained features and objects, the most popular deep-learning model for image analysis is commonly used: the convolutional neural network (CNN). In this review, we provide a comprehensive overview of the impact of deep learning on EO applications by reviewing 429 studies on image segmentation and object detection with CNNs. We extensively examine the spatial distribution of study sites, employed sensors, used datasets and CNN architectures, and give a thorough overview of applications in EO which used CNNs. Our main finding is that CNNs are in an advanced transition phase from computer vision to EO. Upon this, we argue that in the near future, investigations which analyze object dynamics with CNNs will have a significant impact on EO research. With a focus on EO applications in this Part II, we complete the methodological review provided in Part I.

PROMOTING INTERNATIONAL COLLABORATION THROUGH TRAINING AND EDUCATION IN SPACE TECHNOLOGY APPLICATIONS AND ADVANCES AMONG BIMSTEC COUNTRIES – A GOVERNMENT OF INDIA INITIATIVE

Abstract. India is one of the top space-faring nations in the world, which not only to build satellites and launches them using indigenously developed launch vehicles but also spearheads the use of space technology for many applications at the national and international level. Training, education and capacity building (TECB) are important components in promoting the use of space technology, thereby benefitting the society. North Eastern Space Applications Centre (NESAC) is one of many institutions under the umbrella of Indian Space Research Organisation (ISRO), Department of Space (DOS) which is responsible for promoting TECB at National level. ISRO / DOS also supports TECB at an International level through many initiatives such as IIRS – ITC JEP, UN-CSSTEAP, CEOS, UNSPIDER, UN-FAO, WMO, Commonwealth, and BIMSTEC etc. NESAC established in 2000, initiated capacity building and spearheaded Outreach programs from 2019 in the North East Region of India. Prime Minister of India announced in the plenary meeting of BIMSTEC(Bay of Bengal Initiative for Multi-Sectoral Technical and Economic Cooperation) Summit at Kathmandu, Nepal in August 2018, recognizing NESAC to conduct space technology training programs of two weeks and 3 months for BIMSTEC Countries every year for the next five years with full financial support from Government of India. The first two week training program on “capacity building on the earth observation applications and research” was successfully conducted for 24 participants, four participants each representing Bangladesh, Bhutan, Thailand, Myanmar, Nepal, and Sri Lanka. The participants are from User Departments, Science and Technology Ministries, Space Organizations and Academia and Research Institutions. The training benefitted the participants from state of the art training facilities and experienced faculties from NESAC/ISRO/Academia from across India. The summary of feedback from the participants (excellent – 57%, good – 40%, average – 2.6% and poor – 0.4%) indicates that the course is very much useful as it has enhanced their knowledge in earth observation, geospatial analysis capabilities that will be put into use on their return. The training and capacity building has opened new vistas of cooperation and possible collaboration between BIMSTEC countries ranging from availing satellite data, expansion and access to data sharing portals, creating knowledge network to promote young researchers and expert sharing within the region, further to expand the capacity building by initiating distance learning programs, taking up joint research and application projects, etc. The rich experience of conducting a unique program, one of its kind and detailed feedback and possible expansion of BIMSTEC programs in the future is presented in the paper.