Highlights from the Literature: July to September 2020

Nov 30, 2020

Data products distributed by the Land Processes (LP) Distributed Active Archive Center (DAAC) are used in many different applications. They play an important role in modeling, help to detect changes to the landscape, and are a way to assess ecosystem variables, to name a few. Three of those applications, published between July and September 2020, are highlighted below. A more exhaustive, albeit not complete, list can be accessed via the LP DAAC Publications webpage.

Wanyama, D., Moore, N.J., and Dahlin, K.M., 2020, Persistent vegetation greening and browning trends related to natural and human activities in the Mount Elgon ecosystem: Remote Sensing, v. 12, no. 13, art. no. 2113, at https://doi.org/10.3390/rs12132113

Expanding populations lead to a higher demand for food and natural resources, which drive agricultural and other land cover/land use changes (LCLUC) in many regions of the world. In regions with high interannual climate variability and a dependence on rainfed agriculture, this can be a recipe for severe food insecurity. There is a need to better understand the long term impacts of climate change and shorter term impacts of anthropogenic activities (i.e. LCLUC) on vegetation trends (greening/browning) to inform policy decisions that can help conserve the environment and improve communities that depend on it. The Mt. Elgon ecosystem (MEE) straddles the border between Uganda and Kenya and was recently studied by Wanyama and others (2020) to characterize greening/browning trends in vegetation on a multi-temporal scale from 2001–2018.

The researchers investigated spatio-temporal patterns in the Normalized Difference Vegetation Index (NDVI) from the Terra Moderate Resolution Imaging Spectroradiometer (MODIS) Vegetation Indices 250 meter (m) 16-day composited product (MOD13Q1.006) at multi-temporal scales (10-day, 16-day, and seasonal) using an integration of Mann-Kendall, Sen’s slope, and the Breaks for Additive Season and Trend (bfast) algorithms. They included precipitation data from the Climate Hazards Group Infrared Precipitation with Stations data product to disentangle human-driven vs. natural vegetation changes. Combined Terra and Aqua MODIS Annual Land Cover Type (MCD12Q1.006) data from 2018 was also used to classify land cover types in the region. The MODIS NDVI data were extracted using NASA’s Application for Extracting and Exploring Analysis Ready Samples (AppEEARS). The authors describe performing a quality assessment of the MODIS NDVI data using the quality layer to exclude poor quality pixels from the analysis.

After applying the integrated Mann-Kendall, Sen’s slope, and bfast algorithms on the time series, the authors found that vegetation in the MEE is highly variable, with greening and browning being observed at all time scales. Using multiple algorithms helped detect different changes. The Mann-Kendall and Sen’s slope approach revealed major changes related to forest gain/loss, while the bfast approach observed more subtle changes, including vegetation degradation. The authors found that certain changes could be attributed to natural (precipitation) and anthropogenic (particularly vegetation to cropland LCLUC) factors. This work characterizes MEE vegetation dynamics, which can inform policy decisions that will both help conserve the environment and protect the livelihoods of the communities dependent on this region for agriculture.

Marino, E., Yebra, M., Guillén-Climent, M., Algeet, N., Tomé, J.L., Madrigal, J., Guijarro, M., and Hernando, C., 2020, Investigating Live Fuel Moisture Content estimation in fire-prone shrubland from remote sensing using empirical modelling and RTM simulations: Remote Sensing, v. 12, no. 14, art. no. 2251, at https://doi.org/10.3390/rs12142251.

Fuel characteristics such as vegetation moisture content substantially impact wildfire activity. Live Fuel Moisture Content (LFMC) is a key indicator of the flammability of vegetation, fire behavior, severity, and is an important parameter of wildfire models. Accurately estimating LFMC improves operational fire management activities for prevention and suppression of wildfire in addition to emergency responses. A common strategy for assessing wildfire risk is to launch field campaigns that survey fire-prone plant species (indicator species); however, these campaigns are expensive and time-consuming. Marino and others (2020) investigated the potential for using remotely sensed data to estimate LFMC for large-scale fire risk assessments in the Madrid region in Central Spain.

The researchers utilized the Combined Terra and Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) Nadir Bidirectional Reflectance Distribution Function (BRDF)-Adjusted Reflectance (NBAR) Daily L3 Global 500m product (MCD43A4.006), Terra MODIS Surface Reflectance L2G 500m/1000m product (MOD09GA.006), and Sentinel-2 Multi-Spectral Instrument (MSI) top of the atmosphere reflectance  to calibrate models for the estimation of LFMC. Spectral indices (SI), including Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Normalized Difference Infrared Index (NDII), and Normalized Difference Water Index (NDWI) were derived using the MODIS and Sentinel-2 data and used to calibrate the models. Reference data for the study area was gathered from 143 field samples where Cistus ladanifer, an indicator species, was used to determine the LFMC (the percentage of water content of vegetation on a dry-weight basis).

Multiple types of empirical models were leveraged, including linear regression (LR), multivariate linear regression (MLR), non-linear regression (NLR), and general additive models with splines (GAMs), to estimate LFMC from the derived SI. LFMC is also estimated using a radiative transfer model (RTM), and LFMC estimations from empirical and physically-based approaches were compared. LFMC derived from field observations were usually highest in the spring season and lowest in the summer season. Besides the seasonal differences, annual variations were also observed, potentially due to different weather conditions (temperature and precipitation). The results showed a high correlation between most of the remote sensing-derived SI and LFMC data. GAM had the best model fit while NLR had better results in model validation. The remotely sensed approaches generally overestimated the minimum LFMC; however, the authors believe their methods are promising, cost-effective, and provide valuable information for operational wildfire risk management.

Gang, C., Pan, S., Tian, H., Wang, Z., Xu, R., Bian, Z., Pan, N., Yao, Y., and Shi, H., 2020, Satellite observations of forest resilience to hurricanes along the northern Gulf of Mexico: Forest Ecology and Management, v. 472, art. no. 118243. [Also available at https://doi.org/10.1016/j.foreco.2020.118243.] 

When hurricanes make landfall, they endanger not only human made structures but also natural terrestrial systems. Impacts on natural systems like forests can be extensive, with damage ranging from defoliated trees and broken branches to the complete uprooting of trees. These impacts can have wide ranging detrimental implications for forest ecosystems. Understanding how forests change and respond after a hurricane is important for not only protecting and restoring the critical ecosystem services forests provide, but also in understanding how hurricane disturbances impact carbon, nitrogen, and water cycles in forest ecosystems. In this one recent study from Gang and others (2020), hurricanes that made landfall from 2001–2015 along the northern Gulf of Mexico were evaluated to determine how resilient forests are to hurricanes, and how forests recovered post-hurricane.

The authors identified thirteen hurricane tracks for which they gathered hurricane (location, wind speed, and category), forest type (evergreen forests and woody wetlands), and drought data. The researchers also derived multiple ecological indicators and used these along with the gathered data to evaluate the resilience and recovery of impacted forests. The ecological indicators were derived using the Combined Terra and Aqua MODIS Nadir Bidirectional Reflectance Distribution Function (BRDF)-Adjusted Reflectance (NBAR) Daily L3 Global 500m product (MCD43A4.006) and were used to determine the forest canopy state before and after the disturbance. These indicators included Vegetation Indices (VIs) like the Enhanced Vegetation Index (EVI) and Normalized Difference Infrared Index (NDII), among others.

The resilience of forests to hurricanes following the disturbance was determined by comparing forest types and their post-hurricane state against the pre-hurricane state. This was done in two ways: by comparing VI-based statistics (VI statistics) just before and just after the storm, and by comparing the post-hurricane VI statistics to the average of the VI statistics for all the previous years. The researchers found that among the VIs compared, NDII, a VI calculation based on the Short-Wave Infrared (SWIR) and Near-Infrared (NIR) spectral bands, was better when determining the magnitude of resilience following a hurricane. The researchers also found that wind speed was a critical component in forest damage during a hurricane and that woody wetlands were more resilient than other forest types observed in the study. Finally, post-hurricane recovery of forest canopy was variable depending on a number of factors, including the strength of the storm (stronger winds result in more damage), time since the last storm in the same area (having successive storms slows down recovery), and whether forest recovery is taking place during a drought (shown to extend the recovery period). This study brings science closer to understanding how forest ecosystems respond following a hurricane disturbance and will aid in the development of post-hurricane recovery models.

Material written by: Cole Krehbiel¹, Mahsa Jami¹, and Aaron Friesz¹

¹KBR, Inc. contractor to the U.S. Geological Survey, Earth Resources Observation and Science (EROS) Center, Sioux Falls, South Dakota, USA. Work performed under USGS contract G15PC00012 for the LP DAAC²

²LP DAAC work performed under NASA contract NNG14HH33I.