Projects
Our Research Projects & Works
Observations and modeling permafrost in High Mountain Asia
Improved Assessments of Permafrost and seasonally frozen ground in High Mountain Asia by Integrating Satellite Observations with Physics-Based Models and In-Situ Observation
It is estimated that permafrost covers more than 1.2 million km2 of the area in High Mountain Asia, exceeding the glaciated area by more than an order of magnitude. About 75% of this area is in the Tibetan plateau. Even outside of China, the permafrost covered area exceeds glaciated areas significantly in almost all countries of the HMA. Besides permafrost, seasonal ground freezing (SFG) occurs extensively in the HMA. The area covered by SFG is poorly constrained. Although there are many previous studies of permafrost and SFG in the Tibetan plateau, there are almost no studies in the rest of the HMA, and there is an acute lack of research capacity for detailed measurements and field studies. Unlike on the Tibetan plateau, permafrost in the mountain ranges of the HMA is largely discontinuous. The overall objective of the proposed research is to improve quantitative estimates of permafrost and seasonally frozen ground in HMA over the period from 1980-present, by integrating remote sensing observations with physics-based thermo-hydrologic models for snow and the subsurface, and in-situ observations. The models will account for the important influence of snow cover on permafrost and SFG, and rigorous representations of freeze-thaw processes in porous media. This project will make extensive use of HMA datasets for forcing thermo-hydrologic models and for developing improved permafrost zonation maps for the HMA.
The modeling strategy will interrogate the wide climatic variability across the HMA, ranging from winter precipitation dominated by westerlies in the western margins to the strong control exerted by the summer monsoon in large parts of the HMA. The role of the monsoon, extremely high elevations and aridity distinguish the mountain permafrost regions of HMA from those in other mountain ranges. We consider six sub-regions in the HMA across this climatic gradient. Within each sub-region, factors controlling permafrost occurrence vary significantly due to the mountainous terrain. Patches for high-resolution physics-based simulations will be selected based on clustering analysis of differentiation in major topographic variables and land surface characteristics that control permafrost and SFG occurrence. This approach avoids the need to model the entire area, facilitating computational efficiency. Thermo-hydrologic modeling will produce estimates of ground temperature profiles; active layer thickness; extent, duration and depth of seasonally frozen ground over the period from 1980-present. Modeling results will be verified against the HMA ASCAT freeze-thaw data after aggregation, previously reported local-scale field observations from the Tibetan plateau and proxy datasets (rock glacier maps). To address the paucity of data on permafrost outside Tibet, we will carry out field investigations at three sites in Nepal across a climatic gradient that captures the influence of both the westerlies and monsoon, and variations in annual precipitation.
We are committed to participating in and developing synergies with the HMA team. We will work together with other HMA investigators to compile high-resolution forcing datasets for 1980-present in HMA. Our proposed research will produce the following products for HMA’s GMELT toolbox: (i) Improved high-resolution permafrost occurrence index maps, (ii) Model predictions of active layer thickness in permafrost covered regions, (iii) duration and depth of SFG, (iii) Hydrologic consequences of changes in active layer thickness and SFG dynamics – partitioning into lateral and vertical fluxes at selected depths, (iv) Trends in all relevant variables between 1980-present.
Sponsor: NASA
Development of a long-term consistent multi-satellite soil moisture data record
Development of a long-term consistent multi-satellite soil moisture data record with Uncertainty information
In this project we improve the utility and science quality of a long-term soil moisture data record derived from 20 years of passive microwave satellite observations. This investigation is sponsored by the 2017 NASA ROSES MEaSUREs (Making Earth System Data Records for Use in Research Environments)
We will focus on the following research activities that would address the overarching objective of this collaboration – to improve the utility and science quality of our MEaSUREs soil moisture earth system data record (ESDR):
- Provide quality control of and usage feedback throughout the ESDR development process
- Evaluate geographical trend detectability as a function of available data record length, data natural variability, geophysical parameter (primarily soil moisture in this investigation), confidence level, trend magnitude, and to some extent, sensing frequencies.
- Explore science questions and practical applications that can be more fully addressed with longer soil moisture data records compared with shorter ones.
- Identify historically active drought regions over CONUS (or other geographical regions as applicable) and apply spatial downscaling techniques in conjunction with knowledge from 2) above to evaluate potential hindcast/forecast benefits on drought risk management.
Sponsor: NASA Jet Propulsion Laboratory
AI/ML for flash drought and water security
Connection between flash drought and water security using AI/ML algorithms for certain African and Asian Watersheds
Flash droughts can occur very quickly between 2-6 weeks and impact large areas in a watershed including a sudden reduction in streamflow and thereby a loss of water supply to a region dependent on it. Some of the watersheds in Asia and Africa that have national security implications to United States are studied in this proposal. In the past two decades (since 2000), there have been numerous satellite missions that have been launched that make routine observations of land surface and atmospheric variables that play a role in the terrestrial water balance. These include precipitation, soil moisture, evapotranspiration, vegetation and total water. In addition to these earth observations, there are global models that simulate these hydrological variables. In this proposal we will develop the “training” using deep learning for the Mississippi River Basin where we have in-situ observations of daily streamflow. We will use transfer learning to “transfer” the relationship between flash drought and streamflow to eight watersheds in Asia and Africa that do not provide streamflow observations or lack daily streamflow that is needed to track a flash drought.
Sponsor: Army Research Office
Machine learning for groundwater studies
High-Resolution Estimation of Groundwater Withdrawals using Machine Learning Integration of Satellite Datasets
Globally, groundwater withdrawals drive changes in storage. However, very few regions of the U.S. and world monitor their groundwater withdrawals at the local scale necessary to implement sustainable management solutions. In this proposal, we seek to develop methodologies that estimate both unconfined and confined groundwater withdrawals at the local scale (~1 km). We will accomplish this by integrating data from active and passive satellite sensors at a wide range of spatial and temporal scales that are sensitive to different aspects of the water balance.
We will use data from a variety of disparate satellite sensors including Landsat, Sentinel-1, GRACE, GRACE-FO, MODIS, TRMM, GPM, and SMAP, which measure different components of the water balance at resolutions ranging from 10 m to 100s of km. We propose to integrate these inherently different remote sensing datasets using a hybrid water balance/machine learning approach, which can integrate datasets with a wide range of spatio-temporal resolutions.
After estimating total groundwater withdrawals, we will couple these estimates with InSAR- derived groundwater storage change. InSAR measures the storage change in confined aquifers, so integrating the two measurements will enable us to determine the likely origin of the groundwater pumping - from confined or unconfined aquifers, providing estimates of the ability of these aquifers to be recharged and further highlighting water security issues. We will calibrate and validate the performance of our approach initially in two areas: Arizona and Kansas. Both regions are pumping large volumes of groundwater and are actively monitoring this pumping
with associated water level drawdowns, but are in very different hydrologic settings, providing excellent existing datasets to calibrate and validate our models.
The result of this research will be a new algorithm that combines active and passive satellite data to estimate withdrawals in unconfined and confined aquifers at high (~1 km) spatial resolution. This information will allow water managers to prioritize aquifer management, as confined aquifers are harder to recharge and can lose storage capacity.
Sponsor: NASA
Multi sensor satellite data fusion for flood studies
Mapping flood impacts using multi-sensor satellite data fusion in urban areas
Floods cause more damage than any other disaster. Today floods account for almost half of all weather-related disasters over the last two decades, affecting 2.3 billion people (UNISDR 2015). This high cost of natural disasters pushes 26 million people into poverty every year, causing setbacks to development as government budgets are stretched and people without financial protection are forced to sell assets (Hallegatte et al. 2016). This project will contribute to hydrology and remote sensing, which have identified urban flooding mapping as a primary challenge and concern (Ward et al. 2015, Schumann et al. 2018).
Mapping urban floods with satellites has been, until recently, challenging. Moderate resolution satellites, like MODIS, image the globe daily. However, flood detail in narrow channels or small features such as streets cannot be detected due to low spatial resolution and large view angles (Ticehurst et al. 2014). Higher resolution images like Landsat (30m resolution) and SAR (Synthetic Aperture Radar) sensors have been used for urban flood detection (Horritt et al. 2001, Martinis et al. 2009, 2013, Feyisa et al. 2014, Mason et al. 2014). SAR sensors can be very important for flood detection due to their ability to see through clouds, which often partially or completely obfuscate the ability to map peak flood extent. Yet, these sensors are limited by temporal resolution (Landsat passes ~ 16 days and Sentinel-1 ~10 days). Existing lower resolution SAR satellites have limitations in dense urban areas due to its side-looking nature that can obfuscate urban ground surface from radar shadows (Soergel et al. 2003). In some cases, however, loss of coherence in SAR sensors has been proven to located damaged areas from floods (Chini et al. 2019). A rise in the production of “microsatellites” in the last decade (e.g., from the company Planet) have been globally deployed to attain high spatial (0.8-5m) and temporal (daily) resolution optical imagery since 2016. New companies such as ICEYE have launched daily radar satellites, which could prove key for mapping flooding in cloudy, urban areas, as was shown in mapping flood exposure after hurricane Dorian in the Bahamas. The rise of UAV imagery has also been tested for flood mapping, with even higher resolution (0.2m) shown to map urban flooding with high accuracy (Feng et al. 2015).
This project will provide the foundational science needed to leverage data to map the spatial extent of urban floods, which can aid responder and relief, insurance efforts, urban planning, and help calibrate and reduce uncertainty in urban flood models critical for climate change scenario planning for cities around the world. We will leverage both public and commercial optical and radar satellite data to identify areas inundated or potentially damaged by flood events. We will use data fusion techniques to leverage observations during flood events by each satellite and predict potential flood damage in areas obscured by clouds in optical imagery. The data fusion model will leverage elevation, precipitation, and past flood observations from satellites to infill missing areas. We will test and develop this technique for a variety of urban areas in different climatic zones and improve the utility of satellite based urban flood maps for urban decision makers.
Sponsor: NASA
High resolution satellite data to map tickborne disease in Virginia
Identifying environmental determinants of transmission of tickborne disease in Virginia using high resolution remote sensing data
Climate is a major driver of infectious diseases transmission in the US and globally. The survival and dispersal of microorganisms and the distribution of their intermediary hosts and vectors is determined and constrained by environmental and hydrometeorological variability. Blood-feeding ticks act as vectors for numerous infectious diseases in the US, most commonly the bacterial diseases Lyme disease, anaplasmosis, ehrlichiosis, and spotted fever rickettsiosis. From 2004-2016, nearly half a million tick-borne disease cases were reported in the US – 75% of the total national vector-borne disease burden – with Lyme disease accounting for 82%. This disease group costs the US economy an estimated US$1.3 billion per year and due to climate change, the geographical ranges of disease-transmitting insects, such as the soil-dwelling Lyme disease vector Ixodid ticks, are expanding. This climate sensitivity is partially mediated by the effect of prevailing meteorological conditions on small-scale soil ecosystems, however, there is considerable uncertainty surrounding these relationships due to the scarcity of empirical data linking climate and health data spatiotemporally and lack of collaboration between discipline. The increased availability of historic meteorological data offers great potential to environmental epidemiology that is only just being explored. Climate data products derived from earth observation (EO) and model-based reanalysis can be matched to health outcomes and used to make inferences about transmission mechanisms of specific infectious pathogens. Many health outcomes of international public health importance can now be mapped at global or regional scale and at high spatial resolution using emerging geostatistical methods combined with publicly available georeferenced data on climate and other covariates.
The expected outcomes from completing these aims will be high-resolution (1km2) maps of environmental variables (precipitation, soil moisture, vegetation and air temperature and humidity) and transmission risk for selected diseases in the Commonwealth of Virginia and nationally.
Sponsor: University of Virginia
Remote sensing and model estimation of local to regional water and carbon budgets in Virginia
Remote sensing and model estimation of local to regional water and carbon budgets in Virginia
This proposal addresses the linkages between water and carbon leveraging of the studies on (a) hydrological modeling in watersheds (b) remote sensing and downscaling of soil moisture (c) coupled water and carbon cycling at landscape to regional levels. We will use the progress accomplished by that links satellite data to well-established biogeochemical models for applications to watersheds in Virginia to understand and scale the connections between water and carbon.
We will integrate top-down, satellite driven methods with bottom-up high-resolution process modeling of water and carbon storage and cycling over the full state of Virginia. The top-down methods will provide estimates of coupled water and carbon cycling at 1 km resolution, while the bottom-up approach will provide resolutions of 10-30m. The coarser resolution is designed to provide regional scale patterns in response to changing climate and land use, while the finer resolutions provide landscape level estimates appropriate for planning and evaluation of local land management efforts to promote freshwater sustainability and enhance carbon sequestration. The project will cover the full land area of Virginia, including rural and urban regions, and the Mountain, Piedmont, and Coastal Plain.
Sponsor: University of Virginia
Environmental justice in connected coastal communities
Supporting Environmental Justice in Connected Coastal Communities through a Regional Approach to Collaborative Community Science
Does engaging diverse communities in regional science partnerships make them more resilient to coastal hazards and less susceptible to environmental injustices? This project will investigate how the co-production of scientific knowledge between community members, regional stakeholders, and academic researchers contributes to understanding socioenvironmental drivers that impact resilience to coastal hazards and the adoption of solutions to overcome them, particularly for marginalized populations that are being disproportionately affected by poor water quality, hurricanes, floods, droughts, and sea level rise. Divergent economic interests, significant racial inequities, and differing degrees of flood and water quality risk for communities throughout the region around North Carolina’s Pamlico Sound, along with the ecologic and economic importance of these coastal waters, make this estuary an ideal study area for a CoPe hub. Stakeholders in the project represent fishers, farmers, local and state government, tourists, and residents at multiple nested scales of decision making reflecting different social contexts
built on environmental attitudes, economic incentives and inequities, and propensities for social cooperation. These contexts are themselves dependent on individual perceptions of identity, trust, norms, and control over the environmental system. Our team of academic researchers takes a transdisciplinary approach to understanding this complex system by pursuing four objectives: (1) mapping key natural, built, and socioeconomic resources and interdependencies that define the regional socio-engineered-environmental system (SEES); (2) understanding how coastal hazards enhance vulnerabilities in the region; (3) identifying opportunities for locally appropriate adaptation and mitigation strategies to build community and regional resilience; and (4) establishing a Coastal Environmental Justice Institute as a long-term mechanism to promote and support collaboration among stakeholder groups from diverse communities throughout the region and beyond.
Sponsor: NSF
Building Capacity for Data-driven Adaptation in Rural Coastal Communities
Coastal Futures: Building Capacity for Data-driven Adaptation in Rural Coastal Communities
Rural coastal communities are strongly dependent on natural resources that are affected by climate change, particularly by saltwater intrusion from accelerated sea-level rise and more severe storm flooding. At the same time, they face especially difficult challenges in responding to climate risks, including lack of access to scientific information and expertise, lack of coordination among communities, geographical isolation, social inequities, unstructured governance, and limited institutional capacity. The decentralized nature of decisions in rural regions often results in a focus on short-term and local benefits and the inequitable use of public resources rather than on longer-term climate adaptation strategies. We propose a focused Coastal Futures Hub to address this inequity and information gap for rural coasts by bringing together scholars and stakeholders to co-produce an open-source and interactive data, modeling, and visualization platform to enable sustained collaboration and support equitable decision making and solution adoption.
Sponsor: NSF