Search for: All records

Integrating Artificial Intelligence (AI) techniques with remote sensing holds great potential for revolutionizing data analysis and applications in many domains of Earth sciences. This review paper synthesizes the existing literature on AI applications in remote sensing, consolidating and analyzing AI methodologies, outcomes, and limitations. The primary objectives are to identify research gaps, assess the effectiveness of AI approaches in practice, and highlight emerging trends and challenges. We explore diverse applications of AI in remote sensing, including image classification, land cover mapping, object detection, change detection, hyperspectral and radar data analysis, and data fusion. We present an overview of the remote sensing technologies, methods employed, and relevant use cases. We further explore challenges associated with practical AI in remote sensing, such as data quality and availability, model uncertainty and interpretability, and integration with domain expertise as well as potential solutions, advancements, and future directions. We provide a comprehensive overview for researchers, practitioners, and decision makers, informing future research and applications at the exciting intersection of AI and remote sensing. 

Computational workflows are widely used in data analysis, enabling automated tracking of steps and storage of provenance information, leading to innovation and decision-making in the scientific community. However, the growing popularity of workflows has raised concerns about reproducibility and reusability which can hinder collaboration between institutions and users. In order to address these concerns, it is important to standardize workflows or provide tools that offer a framework for describing workflows and enabling computational reusability. One such set of standards that has recently emerged is the Common Workflow Language (CWL), which offers a robust and flexible framework for data analysis tools and workflows. To promote portability, reproducibility, and interoperability of AI/ML workflows, we developedgeoweaver_cwl, a Python package that automatically describes AI/ML workflows from a workflow management system (WfMS) named Geoweaver into CWL. In this paper, we test our Python package on multiple use cases from different domains. Our objective is to demonstrate and verify the utility of this package. We make all the code and dataset open online and briefly describe the experimental implementation of the package in this paper, confirming thatgeoweaver_cwlcan lead to a well-versed AI process while disclosing opportunities for further extensions. Thegeoweaver_cwlpackage is publicly released online athttps://pypi.org/project/geoweaver-cwl/0.0.1/and exemplar results are accessible at:https://github.com/amrutakale08/geoweaver_cwl-usecases.

 

Mountain snowpack provides critical water resources for forest and meadow ecosystems that are experiencing rapid change due to global warming. An accurate characterization of snowpack heterogeneity in these ecosystems requires snow cover observations at high spatial resolutions, yet most existing snow cover datasets have a coarse resolution. To advance our observation capabilities of snow cover in meadows and forests, we developed a machine learning model to generate snow-covered area (SCA) maps from PlanetScope imagery at about 3-m spatial resolution. The model achieves a median F1 score of 0.75 for 103 cloud-free images across four different sites in the Western United States and Switzerland. It is more accurate (F1 score = 0.82) when forest areas are excluded from the evaluation. We further tested the model performance across 7,741 mountain meadows at the two study sites in the Sierra Nevada, California. It achieved a median F1 score of 0.83, with higher accuracy for larger and simpler geometry meadows than for smaller and more complexly shaped meadows. While mapping SCA in regions close to or under forest canopy is still challenging, the model can accurately identify SCA for relatively large forest gaps (i.e., 15m < DCE < 27m), with a median F1 score of 0.87 across the four study sites, and shows promising accuracy for areas very close (>10m) to forest edges. Our study highlights the potential of high-resolution satellite imagery for mapping mountain snow cover in forested areas and meadows, with implications for advancing ecohydrological research in a world expecting significant changes in snow. 

Crop growth depends on the root-zone soil moisture (RZSM) (~top 1m). Accurate estimation of RZSM is vital to optimize irrigation management for saving water and energy while sustaining crop yield. The High-Resolution Land Assimilation System (HRLDAS) from NCAR can generate RZSM at field scales for irrigation management. The soil moisture data from various agriculture sites in the AmeriFlux network, U.S. Climate Reference Network (USCRN), and Soil Climate Analysis Network (SCAN) are used to verify the soil moisture products generated by HRLDAS. Although the HRLDAS products is not location specific and could be applied nationwide, this study will focus on Nebraska for evaluation, validation, and further calibration. We also compared NASA’s SMAP surface soil moisture products to HRLDAS surface layer soil moisture. Since the accuracy of the SMAP product is known, this comparison directly validates the HRLDAS surface soil moisture product and indirectly validate its RZSM products. Results from these two validation methods show a good accuracy of HRLDAS soil moisture products. The conspicuous differences between HRLDAS and SMAP products indicate that HRLDAS omits the irrigation activities as its simulation is based on weather variables and energy balance. It’s hard for HRLDAS to consider and include the irrigation actions in its results, while as SMAP products remotely sense the soil moisture as it is, the changes caused by irrigation are clearly reflected. Therefore, a simple calibration is applied to the HRLDAS products by including irrigation amount as its variables. 

Irrigation is the primary consumer of freshwater by humans and accounts for over 70% of all annual water use. However, due to the shortage of open critical information in agriculture such as soil, precipitation, and crop status, farmers heavily rely on empirical knowledge to schedule irrigation and tend to excessive irrigation to ensure crop yields. This paper presents WaterSmart-GIS, a web-based geographic information system (GIS), to collect and disseminate near-real-time information critical for irrigation scheduling, such as soil moisture, evapotranspiration, precipitation, and humidity, to stakeholders. The disseminated datasets include both numerical model results of reanalysis and forecasting from HRLDAS (High-Resolution Land Data Assimilation System), and the remote sensing datasets from NASA SMAP (Soil Moisture Active Passive) and MODIS (Moderate-Resolution Imaging Spectroradiometer). The system aims to quickly and easily create a smart, customized irrigation scheduler for individual fields to relieve the burden on farmers and to significantly reduce wasted water, energy, and equipment due to excessive irrigation. The system is prototyped here with an application in Nebraska, demonstrating its ability to collect and deliver information to end-users via the web application, which provides online analytic functionality such as point-based query, spatial statistics, and timeseries query. Systems such as this will play a critical role in the next few decades to sustain agriculture, which faces great challenges from climate change and increased natural disasters. 

To effectively disseminate location-linked information despite the existence of digital walls across institutions, this study developed a cross-institution mobile App, named GeoFairy2, to overcome the virtual gaps among multi-source datasets and aid the general users to make thorough accurate in-situ decisions. The app provides a one-stop service with relevant information to assist with instant decision making. It was tested and proven to be capable of on-demand coupling and delivering location-based information from multiple sources. The app can help general users to crack down the digital walls among information pools and serve as a one-stop retrieval place for all information. GeoFairy2 was experimented with to gather real-time and historical information about crops, soil, water, and climate. Instead of a one-way data portal, GeoFairy2 allows general users to submit photos and observations to support citizen science projects and derive new insights, and further refine the future service. The two-directional mechanism makes GeoFairy2 a useful mobile gateway to access and contribute to the rapidly growing, heterogeneous, multisource, and location-linked datasets, and pave a way to drive us into a new mobile web with more links and less digital walls across data providers and institutions.