Corina Cerovski-Darriau

Landscapes in humid climates often experience rapid chemical and mechanical weathering that can change groundwater fluxes, drainage pathways, and frictional strength via alterations in subsurface mineral composition and matrix porosity. These changes to hillslope critical zone structure and strength should control landslide susceptibility, size, and failure style. Here we test the impact of differential weathering on landslide failure processes by exploiting a natural experiment along the dioritic Utuado batholith in Puerto Rico's Cordillera Central, where a sharp weathering gradient follows an upstream‐propagating knickzone into an uplifted shoreline platform that was affected by Hurricane Maria in 2017. Through a combination of landslide mapping, field geophysical and geotechnical surveys, topographic analysis, calibrated stream profile inversion, 3D back‐analysis of hillslope strength, and debris‐flow inundation assessments, we show that differential weathering induces changes to critical zone structure and consequently landslide susceptibility and hazard. Our mapping reveals that the hurricane‐induced landslides preferentially cluster along less weathered hillslopes despite a marked increase in frictional strength. Jointly inverted seismic refraction and electrical resistivity profiles show that the frictionally weak, heavily weathered hillslopes are highly porous and dry, which we posit acts to stabilize slopes through limiting deeper hydrologic connectivity that can produce shallow water tables with destabilizing lateral seepage forces during storms. Debris‐flow inundation mapping and analysis reveals that major contributors to debris‐flow growth, channel steepness and landslide coalescence, scale linearly with hillslope friction angle. This leads to higher‐mobility debris flows along knickzone reaches, which disproportionately affects communities at the base of the plateau.

Background: Wildfires abruptly change landscapes by altering soil properties and vegetation cover. These changes are thought to reduce soil infiltration capacity, making landscapes susceptible to runoff and erosion. However, post-fire soil response is complex and likely varies across locations and time. Aims: Here, we aim to understand regional post-fire soil response and recovery by tracking changes across different northern California (USA) lithology and vegetation types. Methods: We conducted repeat in situ soil infiltration tests for 3 years post-fire at 31 burned and 10 unburned sites spanning the 2021 Dixie, 2020 LNU Lightning Complex, 2020 Walbridge and 2020 Glass fires. Key results: Our two main findings are: (1) burned chaparral soils have increased hydraulic conductivity compared with unburned sites, and (2) infiltration rates return to pre-fire conditions within 3 years across most lithologies and vegetations. Conclusions: Recovery might be generalizable by vegetation and lithology but differ regionally, making it important to identify meaningful hydrologic response units (HRUs). Multi-year studies with paired burned and unburned measurements can constrain the recovery timeline and provide information missed by observations solely of burned soils. Implications: Understanding where, and for how long, soil remains susceptible to runoff and erosion can help prioritize areas and time periods most in need of mitigation.

Landslide runout and debris-flow inundation can disrupt areas well beyond their initial sources, causing widespread damage and extensive fatalities. Understanding where they start and how far they might travel is essential in many locations worldwide. However, most landslide susceptibility maps focus on initiation areas and fail to incorporate runout areas. Furthermore, hazards vary greatly with the degree of landslide mobility, and debris flows can grow as they travel beyond their initial source. Here, we use the USGS Grfin Tools software suite to map susceptible areas from start to stop in the Federated States of Micronesia where landslide and debris-flows runout onto gentle ground is a deadly threat. The DEM-based models in Grfin Tools use simple, empirical, and well-documented approaches that require minimal parameters and, thus, can be used in areas that lack detailed physical properties. We illustrate methods to estimate the required parameters using direct landslide observations, and we suggest alternate methods if direct observations are absent. Our simple parameter estimations in the Federated States of Micronesia successfully predict areas that match past landslide runout and debris-flow inundation, as well as zones with no inundation. In addition to creating preliminary regional assessments, Grfin Tools can be used for comparing multiple scenarios and/or identifying areas for further investigations.

This Review synthesizes progress and outlines a new framework for understanding how land surface hazards interact and propagate as sediment cascades across Earth’s surface, influenced by interactions among the atmosphere, biosphere, hydrosphere, and solid Earth. Recent research highlights a gap in understanding these interactions on human timescales, given rapid climatic change and urban expansion into hazard-prone zones. We review how surface processes such as coseismic landslides and post-fire debris flows form a complex sequence of events that exacerbate hazard susceptibility. Moreover, innovations in modeling, remote sensing, and critical zone science can offer new opportunities for quantifying cascading hazards. Looking forward, societal resilience can increase by transforming our understanding of cascading hazards through advances in integrating data into comprehensive models that link across Earth systems.

Accurate, high-resolution maps of bedrock outcrops can be valuable for applications such as models of land–atmosphere interactions, mineral assessments, ecosystem mapping, and hazard mapping. The increasing availability of high-resolution imagery can be coupled with machine learning techniques to improve regional bedrock outcrop maps. In the United States, the existing 30 m U.S. Geological Survey (USGS) National Land Cover Database (NLCD) tends to misestimate extents of barren land, which includes bedrock outcrops. This impacts many calculations beyond bedrock mapping, including soil carbon storage, hydrologic modeling, and erosion susceptibility. Here, we tested if a machine learning (ML) model could more accurately map exposed bedrock than NLCD across the entire Sierra Nevada Mountains (California, USA). The ML model was trained to identify pixels that are likely bedrock from 0.6 m imagery from the National Agriculture Imagery Program (NAIP). First, we labeled exposed bedrock at twenty sites covering more than 83 km² (0.13%) of the Sierra Nevada region. These labels were then used to train and test the model, which gave 83% precision and 78% recall, with a 90% overall accuracy of correctly predicting bedrock. We used the trained model to map bedrock outcrops across the entire Sierra Nevada region and compared the ML map with the NLCD map. At the twenty labeled sites, we found the NLCD barren land class, even though it includes more than just bedrock outcrops, accounted for only 41% and 40% of mapped bedrock from our labels and ML predictions, respectively. This substantial difference illustrates that ML bedrock models can have a role in improving land-cover maps, like NLCD, for a range of science applications.

Wildfire can impact soil‐hydraulic properties by reducing saturated hydraulic conductivity and sorptivity, making recently burned landscapes prone to debris flows and flash floods. The post‐fire hazard window can range from years to decades. In Northern California, where wildfire frequency is steadily increasing, the impact and soil‐hydraulic recovery from wildfires is unknown. Following the October 2017 Nuns and Tubbs fires in the Northern Bay Area of California, we established 41 monitoring sites for repeat tension‐disc infiltrometer measurements of field‐saturated hydraulic conductivity (K_fs) over 3.5 years. Our site arrays, which encompass grasslands, chaparral, and oak and conifer forests across a range in lithology, show a marked decrease in K_fs following the wildfires and a swift partial recovery following the initial post‐fire rainy season. Our time series reveals a complex path to soil‐hydraulic recovery marked by distinct seasonal stages. Analysis of changing K_fs, sorptivity, and infiltration model residuals collectively suggests that these stages are related to transitions between soil‐hydraulic processes like structural soil sealing from rainsplash, thermal cracking of bare soil, and vegetation regrowth. While soil infiltration rates were strongly impacted by the 2017 fires, dry ravel estimates are an order of magnitude less for similar slopes than the 2009 Station fire in the San Gabriel mountains of Southern California, suggesting that limited ravel flux may insufficiently load channels for debris flows that initiate from within‐channel failure. Our analysis suggests that burned landscapes in the Northern Bay Area of California may experience rapid soil‐hydraulic recovery and limited pathways toward post‐fire debris flow initiation.

Bedrock weathering and soil production set the rate that hillslope colluvium is produced, while material properties determine whether sediment can fail under a given set of conditions. Together these factors generally control both the frequency and magnitude of landslides. In 2017, Puerto Rico (USA) experienced widespread landsliding across a range of lithologies due to Hurricane Maria, making it an ideal setting to explore the role of sediment generation in landslide response to storms. Based on an inventory of >70,000 landslides island-wide and detailed field mapping from a subset of source areas, we estimate that 0.01-0.1 km3 of material was evacuated from the hillslopes (approximately 1-10 mm of lowering). Focusing on the high-density landslide area of Utuado, we estimate an average lowering of 5-50 mm. From past inventories and records of storm events, a watershed is impacted by a hurricane every ~25 years with enough rain for widespread landslides every ~5 years. Assuming a similar density, the landslide contribution to hillslope lowering could be on the order of 1-10 mm/yr. In a humid-tropical environment, where weathering rates are likely high, can hillslopes continue to produce material at this pace? Elsewhere on the island, soil production rates are on the order of 0.1 mm/yr leading to soil residence times of approximately 10 ky. However, to keep pace with landslide events like Hurricane Maria, soil production likely needs to be at least an order of magnitude faster to maintain soil-mantled hillslopes in this study area. For our study area, we ask: has the large-magnitude rainfall from Hurricane Maria caused an abnormally high density of landsliding, resetting the clock on the material availability for areas like Utuado? Here we relate measures of material properties, bedrock weathering intensity, denudation, and land-use history to begin answering this question.

Episodic runoff brings suspended sediment to West Maui’s nearshore waters, turning them from blue to brown. This pollution degrades the ecological, cultural, and recreational value of these iconic nearshore waters. We used mapping, monitoring, and modeling to identify and quantify the watershed sources for fine sediment that pollutes the nearshore each year. These results focus strategies to reduce pollution on the outstanding sources for this sediment.Terrestrial runoff causing coastal plumes now occurs when two or more hours of rain falls at rates greater than 10–20 millimeter (mm) per hour in source watersheds. Analysis of recent and historical rainfall indicates that West Maui communities can expect rainfalls to bring coastal plumes at least 3–5 times per year. Former agricultural fields and some unimproved roads are possible sources for the fine sediment of these plumes. We found, however, that these obvious sources do not produce plumes during small annual storms, because they drain water at rates that far exceed most annual rainfalls and because there is no evidence for runoff from rains that caused recent plumes. Streambanks now eroding into historic fill terraces of sands, silts, and clays are a more plausible source. These terraces are found only downstream of historical agricultural fields and are composed of silt and fine sand. Surveys show that the fill terraces occupy ~40 percent of streambank length, making them extensive. During 2015–2016, these deposits eroded at median rates of 5–24 mm per year. Summed over West Maui’s watersheds, these rates imply sediment loads carried to the coast that can be ten times or more than modeled pre-human values. A sediment budget indicates that bank erosion of fill terraces from a few watersheds likely dominates the current annual fine-sediment load to the nearshore, with Kahana Stream watershed producing the largest annual input of 285 metric tons, the equivalent of 29 dump-truck loads every year.Although past large storms have contributed to sediment loading, annual plume generation is now caused by smaller rainfalls eroding these near-stream terrace deposits, a legacy of historic agriculture. Treatments of former agricultural fields, roads, and reserve forests are consequently not likely to measurably effect sediment pollution from smaller, more frequent storms. Increased runoff from residential and commercial development of West Maui has the potential to exacerbate sediment plumes from such storms.

Mobility is an important element of landslide hazard and risk assessments yet has been seldom studied for shallow landslides and debris flows in tropical environments. In September 2017, Hurricane Maria triggered > 70,000 landslides across Puerto Rico. Using aerial imagery and a lidar digital elevation model (DEM), we mapped and characterized the mobility of debris slides and flows in four different geologic materials: (1) mudstone, siltstone, and sandstone; (2) submarine basalt and chert; (3) marine volcaniclastics; and (4) granodiorite. We used the ratio of landslide-fall height (H) to travel length (L), H/L, to assess the mobility of landslides in each material. Additionally, we differentiated between landslides with single and multiple source areas and landslides that either did or did not enter drainages. Overall, extreme rainfall contributed to the mobility of landslides during Hurricane Maria, and our results showed that the mobility of debris slides and flows in Puerto Rico increased linearly as a function of the number of source areas that coalesced. Additionally, landslides that entered drainages were more mobile than those that did not. We found that landslides in soils developed on marine volcaniclastics were the most mobile and landslides in soils on submarine basalt and chert were the least mobile. While landslides were generally small (< 100 m²) and displayed a wide range of H/L values (0.1–2), coalescence increased the mobility of landslides that transitioned to debris flows. The high but variable mobility of landslides that occurred during Hurricane Maria and the associated hazards highlight the importance of characterizing and understanding the factors influencing landslide mobility in Puerto Rico and other tropical environments.