Search USGS⌕ Search
Home/Archive/USGS

USGS · 70274299

Harnessing geospatial artificial intelligence and deep learning for landslide inventory mapping: Advances, challenges, and emerging directions

Xiao Chen·Wenwen Li·Chia-Yu Hsu·Samantha T. Arundel·Bretwood Higman
View original source ↗Export metadata ↓

Abstract

Recent advancements in artificial intelligence (AI) and deep learning enable more accurate, scalable, and automated mapping. This paper provides a comprehensive review of the applications of AI, particularly deep learning, in landslide inventory mapping. In addition to examining commonly used data sources and model architectures, we explore innovative strategies such as feature enhancement and fusion, attention-boosted techniques, and advanced learning approaches, including active learning and transfer learning, to enhance model adaptability and predictability. We also highlight the remaining challenges and potential research directions, including the estimation of more diverse variables in landslide mapping, multimodal data alignment, modeling regional variability and replicability, as well as issues related to data misinterpretation and model explainability. This review aims to serve as a useful resource for researchers and practitioners, promoting the integration of deep learning into landslide research and disaster management.

Explore related subjects

Remote SensingArticle

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Xiao Chen, Wenwen Li, Chia-Yu Hsu, Samantha T. Arundel, Bretwood Higman. 2025-05-26. Harnessing geospatial artificial intelligence and deep learning for landslide inventory mapping: Advances, challenges, and emerging directions. https://doi.org/10.3390/rs17111856

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related USGS reports

USGS2026

Shoreline behavior at California groin fields from satellite-based measurements

Satellite imagery has helped to provide decades of shoreline change data to coastal regions all over the world. Although shorelines around the globe have been armored with different types of coastal structures to try and protect beaches and properties from erosion, there are few studies of shoreline dynamics within structures such as groin fields using satellite images. Here we apply satellite-derived shoreline techniques to three areas of California: Ventura, Santa Monica, and Newport Beach, each site containing groin fields of varying number and length. Shoreline trends during 1984–2022 reveal that the presence of these coastal structures may alter the overall shoreline variability at these locations. Peclet numbers have been calculated for each site to better characterize the variability of longshore transport within and across the study areas and to compare with shoreline change characteristics. The satellite-derived shoreline data reveal offsets or “steps” in the shoreline position across the groins with sediment pilling up on one side more than the other due to longshore transport. The magnitudes and dynamics of these shoreline offsets were measured from satellite techniques, and we find that the average offsets are commonly related to groin length. Some shoreline offsets reveal seasonal patterns, with Newport Beach having the largest seasonal patterns of the three sites. We find that these seasonal patterns in shoreline offsets at the Newport Beach groins are related to the seasonal variability in wave direction and littoral transport. We conclude that satellite-derived shoreline techniques with Landsat and Sentinel-2 imagery are adequate to characterize shoreline behaviors within the complex coastal settings of groin fields.

California↗
USGS2026

Climate-adaptive urban planning: Quantitative assessment of drought impacts and practical strategies for climate-resilient urban green spaces

Urban green spaces (UGSs) are vital for enhancing a city’s resilience and livability; however, their functionality is increasingly jeopardized by drought, particularly in water-scarce regions. This study evaluates drought impact on UGSs in Metropolitan Adelaide, Australia, a representative semi-arid urban system, using satellite-derived Normalized Difference Vegetation Index (NDVI) time-series data spanning 2000–2020. Vegetation dynamics were analyzed through Seasonal-Trend decomposition using Loess (STL), standardized anomaly assessment, lagged Pearson correlation, Ordinary Least Squares (OLS) regression, and Mann–Kendall trend analysis. To isolate climatically sensitive signals, 29 urban lawn patches were examined separately from mixed urban canopy, given their shallow root systems and direct dependence on surface moisture. NDVI declined by approximately 0.09 units during the Millennium Drought (2001–2009), with summer greenness deficits reaching 24% below the 20-year benchmark. Temperature was the dominant driver of lawn NDVI variability (r = −0.863, R 2 = 74.5%), substantially exceeding the effect of rainfall (r = 0.156, R 2 = 2.4%). El Niño–Southern Oscillation (ENSO) cycles modulated vegetation responses, with La Niña years supporting recovery and El Niño years amplifying decline. Post-drought recovery remained incomplete, with NDVI deficits of 8–20% persisting through 2020; full recovery was observed only in 2017, coinciding with the highest recorded summer rainfall. No significant directional trend was detected over the full study period (Mann–Kendall τ = 0.005, p = 0.908). These findings demonstrate that heat, rather than water limitation alone, is the primary driver of vegetation stress in urban systems, highlighting the benefits of integrated management strategies that address both warming and moisture deficits to sustain urban green infrastructure under future climate conditions. We introduce the concept of “urban greenery drought,” referring to a form of vegetation stress in managed urban landscapes where greenness is reduced primarily by elevated temperature and atmospheric demand despite water availability.

Adelaide↗
USGS2026

Comparing DESIS hyperspectral and Landsat 10 simulated superspectral data for crop type classification in California's Central Valley

To advance crop type mapping in support of global food and water security, this study compared three spectral configurations: (A) the full 60-band DLR Earth Sensing Imaging Spectrometer (DESIS) hyperspectral narrowband (HNB) dataset, (B) a 14-band subset of DESIS-derived HNBs aligned with the planned Landsat 10 (formerly Landsat Next) spectral configuration (400–1000 nm), and (C) DESIS-based simulations of Landsat 10 superspectral broadbands. The analysis was conducted in California’s Central Valley, hereafter referred to as “the Central Valley”, during the peak growing month of August. DESIS imagery from August 2021, 2022, and 2023 was used sequentially for model development, testing, and independent validation. Over these three years, DESIS provided extensive hyperspectral coverage of much of the 4 million hectares in the Central Valley’s. Analyses were performed on Google Earth Engine using two pixel-based supervised classifiers, Random Forest (RF) and Support Vector Machine (SVM), to differentiate three major crop classes: row crops, grapes and tree crops, and winter wheat/fallow/other. The highest overall accuracy (86%) was achieved using SVM in combination with either the full DESIS hyperspectral dataset or the 14 DESIS narrowbands corresponding to Landsat 10. This finding aligns with earlier studies showing a small number of strategically positioned narrowbands can be optimal for crop type classification. Use of the narrowband datasets resulted in substantially higher accuracy (overall accuracy of 86%) compared to the simulated Landsat 10 broadbands (overall accuracy of 75%), supporting previous studies highlighting the utility of narrowbands. Despite the high accuracy using August imagery, the study indicates more granular crop type classification will require multi-temporal observations spanning the full phenological cycle (June–October), especially for a large number of crop classes. Acquiring task-based hyperspectral imagery over such large areas throughout the growing season remains operationally challenging. In contrast, Landsat 10 superspectral imagery could provide routine coverage across seasons and years that is practical and scalable for future large area crop type mapping and agricultural monitoring.

California↗