skip to main content

Title: Assessing Wildfire Burn Severity and Its Relationship with Environmental Factors: A Case Study in Interior Alaska Boreal Forest
In recent years, there have been rapid improvements in both remote sensing methods and satellite image availability that have the potential to massively improve burn severity assessments of the Alaskan boreal forest. In this study, we utilized recent pre- and post-fire Sentinel-2 satellite imagery of the 2019 Nugget Creek and Shovel Creek burn scars located in Interior Alaska to both assess burn severity across the burn scars and test the effectiveness of several remote sensing methods for generating accurate map products: Normalized Difference Vegetation Index (NDVI), Normalized Burn Ratio (NBR), and Random Forest (RF) and Support Vector Machine (SVM) supervised classification. We used 52 Composite Burn Index (CBI) plots from the Shovel Creek burn scar and 28 from the Nugget Creek burn scar for training classifiers and product validation. For the Shovel Creek burn scar, the RF and SVM machine learning (ML) classification methods outperformed the traditional spectral indices that use linear regression to separate burn severity classes (RF and SVM accuracy, 83.33%, versus NBR accuracy, 73.08%). However, for the Nugget Creek burn scar, the NDVI product (accuracy: 96%) outperformed the other indices and ML classifiers. In this study, we demonstrated that when sufficient ground truth data is available, the ML classifiers can be very effective for reliable mapping of burn severity in the Alaskan boreal forest. Since the performance of ML classifiers are dependent on the quantity of ground truth data, when sufficient ground truth data is available, the ML classification methods would be better at assessing burn severity, whereas with limited ground truth data the traditional spectral indices would be better suited. We also looked at the relationship between burn severity, fuel type, and topography (aspect and slope) and found that the relationship is site-dependent.  more » « less
Award ID(s):
Author(s) / Creator(s):
; ; ; ; ;
Date Published:
Journal Name:
Remote Sensing
Page Range / eLocation ID:
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract

    In post‐fire Siberian larch forests, where tree density can vary within a burn perimeter, shrubs constitute a substantial portion of the vegetation canopy. Leaf area index (LAI), defined as the one‐sided total green leaf area per unit ground surface area, is useful for characterizing variation in plant canopies. We estimated LAI with allometry for trees and tall shrubs (>0.5 and <1.5 m) across 26 sites with varying tree stem density (0.05–3.3 stems/m2) and canopy cover (4.6%–76.9%) in a uniformly‐aged mature Siberian larch forest that regenerated following a fire ∼75 years ago. We investigated relationships between tree density, tree LAI, and tall shrub LAI, and between LAI and satellite observations of Normalized Difference and Enhanced Vegetation Indices (NDVI and EVI). Across the density gradient, tree LAI increases with increasing tree density, while tall shrub LAI decreases, exhibiting no patterns in combined tree‐shrub LAI. We also found significant positive relationships between tall shrub LAI and NDVI/EVI from PlanetScope and Landsat imagery. These findings suggest that tall shrubs compensate for lower tree LAI in tree canopy gaps, forming a canopy with contiguous combined tree‐shrub LAI across the density gradient. Our findings suggest that NDVI and EVI are more sensitive to variation in tall shrub canopies than variation in tree canopies or combined tree‐shrub canopies in these ecosystems. The results improve our understanding of the relationships between forest density and tree and shrub leaf area and have implications for interpreting spatial variability in LAI, NDVI, and EVI in Siberian boreal forests.

    more » « less
  2. Abstract Because the manual counting of soybean ( Glycine max ) plants, pods, and seeds/pods is unsuitable for soybean yield predictions, alternative methods are desired. Therefore, the objective was to determine if satellite remote sensing − based artificial intelligence (AI) models could be used to predict soybean yield. In the study, multiple remote sensing − based AI models were developed for soybean growth stage ranging from VE/VC (plant emergence) to R6/R7 (full seed to beginning maturity). The ability of the Deep Neural Network (DNN), Support Vector Machine (SVM), Random Forest (RF), Least Absolute Shrinkage and Selection Operator (LASSO), and AdaBoost to predict soybean yield, based on blue, green, red, and near infrared reflectance data collected by the PlanetScope satellite at 6 growth stages, was determined. Remote sensing and soybean yield monitor data from 3 different fields in two years (2019 and 2021) were aggregated into 24,282 grid cells that had the dimensions of 10 by 10m. A comparison across models showed that the DNN outperformed the other models. Moreover, as crops matured from VE/VC to R4/R5, the R 2 value of the models increased from 0.26 to over 0.70. These findings indicate that remote sensing data collected at different growth stages can be combined for soybean yield predictions. Moreover, additional work needs to be conducted to assess the model's ability to predict soybean yield with vegetation indices (VI) data for fields not used to train the model. This article is protected by copyright. All rights reserved 
    more » « less
  3. null (Ed.)
    A combination of satellite image indices and in-field observations was used to investigate the impact of fuel conditions, fire behavior, and vegetation regrowth patterns, altered by invasive riparian vegetation. Satellite image metrics, differenced normalized burn severity (dNBR) and differenced normalized difference vegetation index (dNDVI), were approximated for non-native, riparian, or upland vegetation for traditional timeframes (0-, 1-, and 3-years) after eleven urban fires across a spectrum of invasive vegetation cover. Larger burn severity and loss of green canopy (NDVI) was detected for riparian areas compared to the uplands. The presence of invasive vegetation affected the distribution of burn severity and canopy loss detected within each fire. Fires with native vegetation cover had a higher severity and resulted in larger immediate loss of canopy than fires with substantial amounts of non-native vegetation. The lower burn severity observed 1–3 years after the fires with non-native vegetation suggests a rapid regrowth of non-native grasses, resulting in a smaller measured canopy loss relative to native vegetation immediately after fire. This observed fire pattern favors the life cycle and perpetuation of many opportunistic grasses within urban riparian areas. This research builds upon our current knowledge of wildfire recovery processes and highlights the unique challenges of remotely assessing vegetation biophysical status within urban Mediterranean riverine systems. 
    more » « less
  4. For farmers, policymakers, and government agencies, it is critical to accurately define agricultural crop phenology and its spatial-temporal variability. At the moment, two approaches are utilized to report crop phenology. On one hand, land surface phenology provides information about the overall trend, whereas weekly reports from USDA-NASS provide information about the development of particular crops at the regional level. High-cadence earth observations might help to improve the accuracy of these estimations and bring more precise crop phenology classifications closer to what farmers demand. The second component of the proposed solution requires the use of robust classifiers (e.g., random forest, RF) capable of successfully managing large data sets. To evaluate this solution, this study compared the output of a RF classifier model using weather, two different satellite sources (Planet Fusion; PF and Sentinel-2; S-2), and ground truth data to improve maize (Zea mays L.) crop phenology classification using two regions of Kansas (Southwest and Central) as a testbed during the 2017 growing season. Our findings suggests that high temporal resolution (PF) data can significantly improve crop classification metrics (f1-score = 0.94) relative to S-2 (f1-score = 0.86). Additionally, a decline in the f1-score between 0.74 and 0.60 was obtained when we assessed the ability of S-2 to extend the temporal forecast for crop phenology. This research highlights the critical nature of very high temporal resolution (daily) earth observation data for crop monitoring and decision making in agriculture. 
    more » « less
  5. Burn severity, which can be reliably estimated by validated spectral indices, is a key element for understanding ecosystem dynamics and informing management strategies. However, in North Patagonian forests, where wildfires are a major disturbance agent, studies aimed at the field validation of spectral indices of burn severity are scarce. The aim of this work was to develop a field validated methodology for burn-severity mapping by studying two large fires that burned in the summer of 2013–2014 in forests of Araucaria araucana and other tree species. We explored the relation between widely used spectral indices and a field burn-severity index, and we evaluated index performance by examining index sensitivity in discriminating burn-severity classes in different vegetation types. For those indices that proved to be suitable, we adjusted the class thresholds and constructed confusion matrices to assess their accuracy. Burn severity maps of the studied fires were generated using the two most accurate methods and were compared to evaluate their level of agreement. Our results confirm that reliable burn severity estimates can be derived from spectral indices for these forests. Two severity indices, the delta normalized burn ratio (dNBR) and delta normalized difference vegetation index (dNDVI), were highly related to the fire-induced changes observed in the field, but the strength of these associations varied across the five different vegetation types defined by tree heights and tree and tall shrub species regeneration strategies. The thresholds proposed in this study for these indices generated classifications with global accuracies of 82% and Kappa indices of 70%. Both the dNBR and dNDVI classification approaches were more accurate in detecting high severity, but to a lesser degree for detecting low severity burns. Moderate severity was poorly classified, with producer and user errors reaching 50%. These constraints, along with detected differences in separability, need to be considered when interpreting burn severity maps generated using these methods. 
    more » « less