Remote sensing of vegetation conditions after post-fire mulch treatments

Remote sensing of vegetation conditions after post-fire mulch treatments

Journal of Environmental Management 260 (2020) 109993 Contents lists available at ScienceDirect Journal of Environmental Management journal homepage...

2MB Sizes 0 Downloads 9 Views

Journal of Environmental Management 260 (2020) 109993

Contents lists available at ScienceDirect

Journal of Environmental Management journal homepage: http://www.elsevier.com/locate/jenvman

Research article

Remote sensing of vegetation conditions after post-fire mulch treatments Viet D. Vo **, Alicia M. Kinoshita * Department of Civil, Construction, & Environmental Engineering, 5500, Campanile Drive, San Diego, CA, USA

A R T I C L E I N F O

A B S T R A C T

Keywords: Post-fire rehabilitation Mulch Vegetation Treatment EVI

Wildfires are becoming more prevalent and are impacting forests, watersheds and important resources. Hy­ drologic and geomorphic processes following wildfires can include erosion flooding, and degraded water quality. To mitigate these secondary impacts, post-fire restoration treatments can be applied to a burned area to stabilize the land surface or promote vegetative regrowth. This research focuses on wood and straw mulch treatment implemented after the 2012 Waldo Canyon Fire in Colorado (United States) and estimates the spatial and temporal changes in annual and seasonal vegetation after a fire with respect to geomorphic factors. This study highlights the use of satellite-based remote sensing products to investigate the impacts of post-fire rehabilitation treatments on vegetation. Using Enhanced Vegetation Index as a proxy for vegetative growth, vegetation con­ ditions are evaluated with respect to slope, slope aspect, and burn severity to understand the impact of the ground cover treatments on vegetation for five years before and after the fire (2007–2016). Sixty-three burned and untreated sites, forty-nine burned sites treated with wood mulch, and twenty-eight burned sites treated with straw mulch were analyzed. These sites were also compared to two control sites that were unburned and un­ treated, Hunter’s Run and Fountain Creek. Generally, post-fire conditions did not return to pre-fire levels, where average vegetation levels were lower. By the end of the study, burned and untreated sites had larger vegetative levels than burned and treated sites. The vegetation levels of the burned sites were statistically different (α ¼ 0.05) from pre-fire conditions in all areas of treatment. Burned sites treated with wood and straw recovered to 69% and 73% of pre-fire conditions, respectively. This work demonstrates the novel use of remote sensing to observe vegetation after post-fire treatment applications to augment the number of sites and length of time that can be analyzed. The observed change in vegetation conditions also contributes to furthering our understanding of the impacts of post-fire restoration, which is important for post-fire management.

1. Introduction Wildfires are becoming more prevalent and are impacting forests, watersheds, and critical water resources (Hallema et al., 2018; Scasta et al., 2016). Climate change and anthropogenic factors contribute to the increase of frequency and severity of these fire events (Higuera et al., 2015; Kinoshita et al., 2016). The damage caused during a wildfire event can be significant. However, many secondary hydrologic and geomor­ phic processes can immediately follow a wildfire. Hazards include erosion, flooding, and degraded water quality (Moody et al., 2013; Neary et al., 2005; Smith et al., 2011). The majority of these natural landscape processes recover within the first three years post-fire, when natural vegetative recovery tends to supersede the effects of treatment (Parson et al., 2010). In response, managers utilize post-fire rehabilitation to mitigate the

adverse effects that follow a wildfire. Typically post-fire restoration is implemented immediately following the end of the fire and is fully established within the first post-fire year (Napper, 2006). Treating land affected by fire has been conceptualized for a long time and its effec­ tiveness can have varying responses (Robichaud et al., 2000; Wagen­ brenner et al., 2006). Application of post-fire rehabilitation treatment can promote vegetative regrowth, reduce erosion, and control excessive runoff. Typical methods of treatment include erosion barriers, seeding, and mulching, each with a specific treatment objective (Badia and Marti, 2000; Robichaud et al., 2010). Mulching, especially with straw and wood materials, has increasingly become more widespread (Cerda and Robichaud 2009; Robichaud et al., 2014). These methods are econom­ ical and can provide ground cover, reducing the potential for down­ stream flooding and water quality degradation (Napper, 2006; Robichaud et al., 2013; Wagenbrenner et al., 2006).

* Corresponding author. Department of Civil, Construction, & Environmental Engineering, San Diego State University, 5500, Campanile Drive, San Diego, CA, USA. ** Corresponding author. E-mail addresses: [email protected] (V.D. Vo), [email protected] (A.M. Kinoshita). https://doi.org/10.1016/j.jenvman.2019.109993 Received 29 May 2019; Received in revised form 8 December 2019; Accepted 11 December 2019 Available online 21 January 2020 0301-4797/© 2019 Elsevier Ltd. All rights reserved.

V.D. Vo and A.M. Kinoshita

Journal of Environmental Management 260 (2020) 109993

vegetation trajectories (Kinoshita and Hogue, 2011; Rachels et al., 2016; Storey et al., 2016; Wittenberg et al., 2007). Wittenberg et al. (2007) found that the Mediterranean ecosystem in Mt. Carmel, Israel stabilized within three years, but only on north slope aspects. Other studies indi­ rectly reference the impact of vegetative recovery on erosion or flooding processes due to decreased ground cover and increased soil exposure (Badía et al., 2008; Kinoshita and Hogue, 2011; Moody et al., 2013; Shvetsov et al., 2019; Viana-Soto et al., 2017). In the Greater Hinggan Mountain, China, the 5.6 Fire was observed using Landsat TM and ETMþ (Chen et al., 2014). Chen et al. (2014) examined three restoration types (listed in order of increasing intervention): natural regeneration, artificial regeneration, and artificial promotion. It was found that

While literature on post-fire rehabilitation is widely available, the focus is typically on the practical use and response of plant life to fire. Pyke et al. (2013) examined the effectiveness of post-fire seeding treatment relative to invasive species and found that 67% of cases were neutral, neither increasing nor decreasing in invasive plants. In the favorable cases, the treatments that lasted the longest had the most potential to stop invasive plants and establish native vegetation. The best vegetation recovery was observed in regions where plants were able to establish, typically due to favorable climatic conditions. This research investigates the effects of post-fire treatment on vegetation. While in situ observations can be limited, the use of satellitebased remote sensing is becoming a common method to analyze post-fire

Fig. 1. The burn severity and treatment type are shown for the Waldo Canyon Fire in Colorado, USA. The burned and untreated sites used for analysis are denoted with black circles. The locations of the unburned and untreated control sites (Fountain Creek and Hunter’s Run) are shown with red and blue squares, respectively. 2

V.D. Vo and A.M. Kinoshita

Journal of Environmental Management 260 (2020) 109993

despite artificial intervention, natural regeneration produced the most abundant species and higher vertical canopy density. These previous studies demonstrated the use of remote sensing products to increase the area and sample size used for analyses. This research investigates post-fire rehabilitation and vegetative recovery using widely available remote sensing data to develop a greater understanding of post-fire vegetative response at a higher temporal (i.e. more frequent observa­ tions) and larger spatial scales and demonstrates the ability to augment the number of sites studied. The objective of this work are to: 1) estimate monthly and seasonal vegetation changes after a fire, 2) investigate the spatial and temporal patterns of vegetation with respect to geomorphic and climatic factors, and 3) determine the impacts of post-fire rehabil­ itation on vegetative recovery.

using remote sensing. 2.3. Enhanced Vegetation Index (EVI) Enhanced Vegetation Index (EVI) is sensitive to canopy structure and is also calibrated to quantify vegetation biomass and reduce the back­ ground noise, atmospheric noise, and potential for saturation (Huete et al., 2002; Chen et al., 2005). EVI is selected, for the Waldo Canyon Fire area, and defined by the equation: � � NIR RED (1) EVI ¼ 2:5 NIR þ C1 �RED C2 �BLUE þ Ic where NIR is the near-infrared reflectance, RED is the red band reflec­ tance, and BLUE is the blue band reflectance, C1 and C2 are the red and blue corrections for atmospheric resistance, respectively, and L is the canopy background brightness correction factor (Huete et al., 2002). The coefficients are defined as L ¼ 1, C1 ¼ 6, and C2 ¼ 7.5 (Huete et al., 2002). EVI ranges from 1 to 1 to represent vegetation conditions, with lower values indicating lower vegetation (negative values typically barren rock or water) and higher values representing abundant vegetation. To analyze the vegetative conditions and the effects of applied mulch treatments before and after the Waldo Canyon Fire, 8-day and 30-m EVI composites were acquired from the United States Geological Survey (USGS) Landsat 7. The image collection contains Tier 1 data and is the highest quality data that meets geometric and radiometric quality re­ quirements (Chander et al., 2009). The remote sensing data was pro­ cessed with Google Earth Engine. Cloud cover was corrected using a cloud score parameter. Images with a cloud score over 20 (greater than 20% cloud cover) were not used in this analysis.

2. Materials & methods 2.1. Waldo Canyon Fire The Waldo Canyon Fire started on June 23, 2012 and was contained by July 10, 2012. The fire burned approximately 74 km2 in the Pike National Forest, about 8 km west of Colorado Springs, CO (Fig. 1). Young and Rust (2012) used a combination of aerial survey and field reconnaissance to classify 19% of the burned area as high, 40% as moderate, and 41% as low soil burn severity. The Waldo Canyon Fire was the second most destructive wildfire in Colorado history, causing a significant amount of damage and destroying 346 homes (Chin et al., 2019). Elevations in the burned area ranged from 2040 to 2860 m (USGS, 2012). The dominant land cover classification is evergreen forest, which contains trees that are generally taller than 5 m, have greater than 20% total vegetation cover, and 75% of the trees remain green all year (Homer et al., 2015). The vegetative species in the forest was 50% Ponderosa pine, and 28% mixed conifer (Moore and Park, 2012). The soils have a shallow depth with low infiltration and high runoff potential (Young and Rust, 2012). Peaks in streamflow typically occur during the summer due to thunderstorms and the spring due to snowmelt (Moore and Park, 2012). Precipitation in the region averages about 600 mm (23.6 in) a year (NOAA, 2019).

2.4. Data analysis EVI for each of the study sites were evaluated for four main cate­ gories for 10-years (2007–2016). A total of 142 locations were selected and categorized by treatment type and burn severity: Burned-Untreated, Burned-Treated (Burned-Straw and Burned-Wood), and two UnburnedUntreated reference sites (Hunter’s Run and Fountain Creek). Sixtythree burned-Untreated points were selected from within the burn area with an equal representation of low, moderate, and high burn severity types. These points were selected to represent different geomorphic and climatic conditions. Twenty-eight Burned-Straw sites were selected in burned areas that were treated with straw. Forty-nine Burned-Wood sites were selected from burned areas that were treated with wood. A 200-m radius around each location was developed to form a study site. At each site, EVI was acquired and averaged to estimate the mean EVI for each satellite image. All data was analyzed monthly and seasonally. Seasons were divided into four periods of three months based on this region (NOAA, 2019): Winter (December to February), Spring (March to May), Summer (June to August), and Autumn (September to November). Slope, elevation, slope aspect, land cover, and precipitation were acquired from Google Earth Engine datasets. Elevation data was ac­ quired from a USGS digital elevation model (DEM) of the continental United States with a resolution of 1/3 arc second (USGS, 2019). The DEM was processed to acquire slope and slope aspect. Slopes were calculated in degrees and divided into three categories: low (0–15� ), moderate (15–30� ), and high (greater than 30� ). A total of 27 low, 102 moderate, and 13 high slopes were determined from the 142 study sites. Slope aspects were categorized by cardinal directions north (316–45� ), east (46–135� ), south (136–225� ), and west (226–315� ). A total of 9 north, 50 east, 44 south, and 39 west slope aspects were categorized. Precipitation data were acquired from the Parameter-elevation Re­ gressions on Independent Slopes Model (PRISM) Monthly Spatial Data­ set from Oregon State University at 2.5 arc minute resolution (Daly

2.2. Treatment types Wood and straw mulching were utilized in the Waldo Canyon Fire for post-fire rehabilitation (Fig. 1). Wood mulching was ideal due to the availability of on-site hazardous trees (trees affected by the fire that can potentially fall and pose a risk to people, properties, and utilities) and availability of scrap wood off-site from nearby local timber contractors (Robichaud et al., 2013). Straw mulching was also implemented, as it was less dense than the wood mulch and therefore more efficient and economical to distribute. Aerial mulching methods were utilized with field crews to quickly cover the affected burn areas. The objectives of the post-fire treatments were to mitigate soil loss, slow the spread of weeds, preserve the Preble’s mouse riparian habitat, and protect values at risk from flooding and sedimentation (Moore and Park, 2012). In total, 4.23 km2 were treated with straw mulch, and 8.09 km2 were treated with wood mulch (BAER 2012; Robichaud et al., 2013). This work selected combinations of treatments and burn severity for analysis. For comparison, two control sites (unburned and untreated) were selected. Hunter’s Run is approximately 8.7 km outside the burn area and was not affected by the fire. Similar to Chin et al. (2019), Hunter’s Run was considered an unburned or control site in this study. Fountain Creek was also selected as a control site. While the upper watershed of Fountain Creek was within the burn area (63% of the upper watershed was burned), the outlet, which was analyzed in the current study, was not burned (Rosgen and Rosgen, 2013a). No post-fire treat­ ment was applied to Fountain Creek outside of the burned area, thus the downstream area was selected as a control site to monitor vegetation 3

Journal of Environmental Management 260 (2020) 109993

V.D. Vo and A.M. Kinoshita

Fig. 2. Monthly precipitation is shown with blue horizontal bars. The vegetation time series for Burned-Untreated, Burned-Treated (Burned-Straw and BurnedWood), and Unburned (Hunter’s Run and Fountain Creek) are shown for 2007–2016. The solid black line denotes seasonal averages, the dashed green line de­ notes monthly averages, and the solid horizontal red line represents the date of the Waldo Canyon Fire on June 23, 2012.

et al., 2008, 2015). Data for soil burn severity was acquired from the US Forest Service (Young and Rust. 2012). The locations of straw and wood mulch treatment areas were acquired from the US Forest Service (BAER 2012). All geospatial data and image collections were composited within the Google Earth Engine platform and used to analyze the change in EVI across the entire dataset for the 142 locations. We compared the annual, monthly, and seasonal vegetative condi­ tions using a two-factor Analysis of Variance (ANOVA) with replication to determine the significance between observed parameters. This method was used to account for the uneven sample sizes between the Unburned-Untreated, Burned-Untreated, Burned-Wood, and BurnedStraw sites. If the samples were comparable based on the ANOVA, we tested the null hypothesis that the pre-fire (2006–2011) EVI conditions were similar to post-fire (2012–2016) using an F-test. Treatment type (wood or straw) was considered the fixed variable, and slope or aspect was used as the secondary factor. If F-values were greater than F-critical then the null hypothesis was rejected. Confidence values were calcu­ lated at α ¼ 0.05 and represent the statistical significance of hypothesis testing for each treatment type over the study period.

3. Results 3.1. Monthly and seasonal vegetation patterns Using EVI as a proxy for vegetation conditions, an immediate and statistically significant (α ¼ 0.05) decrease in vegetation was observed after the fire (Fig. 2, Table 1). Vegetative levels in Burned-Untreated, Burned-Wood, and Burned-Straw sites immediately reduced by 43.6% (0.140), 63.5% (0.102), and 63.9% (0.082), respectively, following the first three months after the fire (Fig. 2, Appendix Table A-6). The two control study sites Hunter’s Run and Fountain Creek experienced a 24% and 26% change of vegetation in the first post-fire season, respectively. Greater variability in vegetative levels was observed after the fire (see appendix for all values). The monthly EVI averages for straw treated study sites before the fire had a mean value of 0.288 � 0.063 (Fig. 2, Table 1). After the fire, monthly EVI for Burned-Straw means were 0.150 � 0.081. Monthly EVI for Burned-Wood study sites before the fire ranged from 0.003 to 0.665, with a mean of 0.293. After the fire, monthly Burned-Wood EVI had a mean of 0.169 � 0.083. In the Burned-Untreated study sites, the pre-fire monthly EVI had a mean of 0.278 � 0.068. Post-fire monthly EVI had a

Table 1 Average EVI and standard deviations for pre-fire (January 2007 to June 2011) and post-fire (July 2012 to December 2016) time periods and the absolute percent change. Each average is calculated from monthly values within the pre-fire or post-fire period. Burned-Untreated Burned-Wood Burned-Straw Unburned-Untreated (Hunter’s Run) Unburned-Untreated (Fountain Creek)

Pre-Fire Average

Post Fire Average

Absolute % Change

0.278 � 0.293 � 0.288 � 0.282 � 0.213 �

0.196 þ 0.169 � 0.150 � 0.265 � 0.211 �

29.3% 42.5% 47.8% 5.9% 1.1%

0.068 0.077 0.063 0.167 0.069

4

0.075 0.083 0.081 0.148 0.091

V.D. Vo and A.M. Kinoshita

Journal of Environmental Management 260 (2020) 109993

mean of 0.196 � 0.075. While the Unburned-Untreated (Hunter’s Run) pre-fire monthly EVI had a mean of 0.282 � 0.167. The post-fire monthly EVI had a mean of 0.265 � 0.148. Lastly, the UnburnedUntreated (Fountain Creek) sites had a pre-fire monthly EVI of 0.213 � 0.069 and a post-fire EVI of 0.211 � 0.091. Yearly values of EVI were estimated for Burned-Untreated, BurnedStraw, Burned-Wood, and Unburned-Untreated (Fig. 3, Appendix Table A-1). After the fire, Burned-Straw, Burned-Wood, and BurnedUntreated areas had a statistically significant (α ¼ 0.05) decrease in vegetation compared to pre-fire averages. Straw sites had an average decrease in vegetation of 73%, where pre-fire EVI was 0.271 � 0.015 and post-fire EVI was 0.073 � 0.013. Wood sites had an average decrease in vegetation of 68.9%, where EVI was 0.265 � 0.021 before the fire and 0.0823 � 0.011 after the fire. Burned-Untreated sites also observed a decline of 28%, where pre-fire EVI was 0.257 � 0.017 and post-fire EVI was 0.122 � 0.0.008. Burned-Untreated sites had higher overall vegetation levels than treated sites. Wood and straw sites expe­ rienced larger vegetative regrowth after the fire, but did not return to pre-fire conditions (Appendix A1). Wood had vegetative regrowth of 31.2% and straw recovered 25.9% between the first post-fire year (2012) and the fourth post-year (2016), while Burned-Untreated sites had vegetative regrowth of about 15.9% between the first and fourth post-fire year. Comparing the means of post-fire and pre-fire years, Burned-Untreated sites had 47.3%, Burned-Straw had 27.0%, and Burned Wood had 31.1% of pre-fire conditions after 5 years. Variations between annual EVI decreased after the fire for Burned-Straw and Burned-Wood sites, but none of the Burned study sites exceeded their respective pre-fire average for the period of this study. For the Unburned-Untreated sites there was no statistically significant change in vegetation between the pre-fire and post-fire time periods.

3.2. Geomorphic and climatic impacts on vegetative patterns All 142 study sites were categorized by slope aspect (north, east, south, west) and slope (low, moderate, high; Fig. 4; Tables 2 and 3). Hunter’s Run and Fountain Creek had south facing slope aspect and east facing slope aspect, respectively and were classified with moderate slopes. Wood mulch was only applied to moderate slopes. North aspects did not have a significant application of straw treatment, and thus was excluded from analysis. North facing slopes declined 64% and 73% in EVI for Burned-Untreated and Burned-Wood sites, respectively. North facing Burned-Wood had the lowest vegetation levels of all the aspects with negative EVI values, which are typically associated with bare rock or soil. East aspects had the highest average EVI for all study sites, where EVI for Fountain Creek was 0.205, Burned-Untreated was 0.133, Burned-Straw was 0.096, and Burned-Wood was 0.096. Burned sites with south and west slopes had similar declines of about 60%–80% of pre-fire vegetation. Moderate slopes experienced the greatest recovery for burned study sites. Burned-Untreated slopes had a 55% decline in vegetation after the fire, but recovered 18% from 2012 (EVI ¼ 0.114) to the end of 2016 (EVI ¼ 0.139). Straw treatments on moderate slopes declined by 71% post-fire, but increased in vegetative levels by 26% between 2012 (EVI ¼ 0.070) and 2016 (EVI ¼ 0.095). Based on post-fire means, moderate slopes with wood treatment experienced the most recovery of 31% be­ tween 2012 (EVI ¼ 0.070) and 2016 (EVI ¼ 0.102). Straw treatments on high slopes tended to have greater vegetation recovery than straw on moderate or low slopes. Wood mulch treatment on moderate slopes had similar recovery patterns to the moderate slopes in straw and untreated areas. The average annual post-fire precipitation was statistically higher

Fig. 3. Annual vegetation levels with confidence intervals for Burned-Untreated, Burned-Treated (Burned-Straw and Burned-Wood), and Unburned (Hunter’s Run and Fountain Creek) for 2007 through 2016. The solid horizontal red line indicates the Waldo Canyon Fire on June 23, 2012. The dashed vertical green lines represent the pre-fire vegetation level averages. Circles represent the annual average EVI and the whiskers (lines extending from the circles) are the annual ranges between the upper and lower quartiles. Filled orange circles indicate statistically different means from the pre-fire averages. Years that are within the shaded green regions denote averages that are statistically similar to pre-fire vegetation conditions at a confidence level of α ¼ 0.05. 5

V.D. Vo and A.M. Kinoshita

Journal of Environmental Management 260 (2020) 109993

Fig. 4. (Left) Average annual vegetation levels for Burned-Untreated, Burned-Straw, Burned-Wood, and Unburned-Untreated (Hunter’s Run and Fountain Creek). Where available, the slope (low, moderate, and high) is noted for each category. (Right) Average annual vegetation levels for Burned-Untreated, Burned-Straw, Burned-Wood, and Unburned-Untreated. Where available, the slope aspect (north, east, south, and west) is noted for each category.

vegetation levels for burned sites (Fig. 3). Robichaud et al. (2013) noted that first year ground cover was not indicative of the rates of ground cover recovery in the following years for the Hayman Fire (Colorado), Hot Creek Fire (California), Myrtle Creek Fire (Oregon), and School Fire (Washington). In two of the fires, the vegetative levels declined in the first year, and then was dominated by litter and natural regrowth. In the other two fires, 56–68% of the first post-fire year ground cover was composed of the applied treatment and minimal actual plant regrowth occurred. After two to three years, natural vegetative regrowth exceeded ground cover levels of the treatment. Vegetation patterns in this study showed similar responses. Beyond 2014 (two years after the fire), we observed small annual increases in vegetation in the area burned within the Waldo Canyon Fire (Fig. 2; Appendix Table A-1). Seasonal changes had significantly greater variation than annual vegetation levels, thus the ANOVA analysis (Table A-2) concluded that seasonal values could not be compared directly (F ¼ 36.34 > F-critical ¼ 1.55). However, annual trends could be compared (F-value ¼ 2.71 < Fcritical ¼ 3.55). The average annual post-fire EVI conditions were different than the average pre-fire conditions. The Burned-Untreated, Burned-Straw, and Burned-Wood sites had similar declines in the first post-fire season, which persisted until the wetter seasons in Winter 2013 (Fig. 2). This seasonal trend is similar to Wagenbrenner et al. (2006), who noted that in the first season, only straw mulch sites had signifi­ cantly more ground vegetation than the control (untreated) sites, and average vegetation cover remained low. The following seasons found similar recoveries across all the plots, including the control (untreated), seeded, mulch, and contour felled sites. It was not until Spring 2003, five seasons later, when a large upward trend was observed. In the current study, the largest post-fire seasonal impacts for the Waldo Canyon Fire occurred from Summer 2014 to Autumn 2014, where Burned-Untreated, Burned-Straw, and Burned-Wood had a decrease of 48%, 46%, and 44% respectively. Morgan et al. (2014) observed that grass and forb vegeta­ tion increased the first two years, then stabilized over the following four years The authors suggested that a longer study (~6 years post-fire) could provide valuable information on how post-fire rehabilitation in­ fluences the overall trajectory of recovery, as shorter studies (i.e. year-to-year or seasonal) may not provide a comprehensive under­ standing of vegetation response.

than the average annual pre-fire precipitation, mostly due to large pulses that occurred in September 2013 with 193 mm and May 2015 with 246 mm of precipitation (Fig. 2). Corresponding peaks in EVI were observed during these storm events, most notably for the Burned-Straw sites, in which the vegetative levels in September 2013 (EVI ¼ 0.503) exceeded the pre-fire vegetation conditions (EVI ¼ 0.288). Burned-Wood had an increase of 43% (EVI ¼ 0.420) during the September 2013 storm event and Burned-Untreated sites had an increase of 61% (EVI ¼ 0.447). In the Unburned-Untreated sites, Hunter’s Run and Fountain Creek experi­ enced vegetation regrowth of 138% (EVI ¼ 0.595) and 73% (EVI ¼ 0.449), respectively. The Unburned-Untreated seasonal vegetation conditions were more consistent and not as sensitive to large storm events or wet years (Figs. 2 and 3). For example, the large storm in June 2013 increased the vegetation response across all four conditions. However, this vegetation regrowth was not sustained and the BurnedUntreated and Burned-Treated sites were followed by a statistically lower than average EVI in August and September. 4. Discussion 4.1. Monthly and seasonal vegetation patterns after fire Post-fire vegetation regrowth reflected the seasonal and monthly changes over time (Figs. 2 and 3). Generally, all burned study sites did not return to pre-fire average vegetation levels. Large increases in vegetation, such as July through September 2013, were followed by lower vegetation levels later in the year or in the following seasons (Fig. 2). Badia and Marti (2000) found that sites treated with seeding and straw mulch had a 30% increase in plant ground cover compared to untreated burned sites for the first year after the fire. Most of the plants used for seeding died by the second year, which reduced the difference between untreated and treated sites. In the Waldo Canyon Fire, no sig­ nificant first year peak was observed. The vegetative levels decreased from 2012 to 2013 for Burned-Straw and Burned-Wood sites (Fig. 3, Tables 2 and 3). A small increase in vegetation was observed for Burned-Straw and Burned-Wood sites starting in 2014. The treated sites experienced the most variance in the first post-fire year (Fig. 3). Variation in annual post-fire vegetation levels was similar to pre-fire 6

V.D. Vo and A.M. Kinoshita

Journal of Environmental Management 260 (2020) 109993

equivalent slopes (angles of 11.3� and 31.0� ) and found that the burned moderate slopes generally had the most vegetation regrowth regardless of treatment type or application within the post-fire period of study. A study by Rachels et al. (2016) observed mixed results for slopes in the Cedar Fire in San Diego County, California. Low (less than 16.7� ) and moderate (16.7� –45.0� ) slope angles had similar recoveries between three sites, with moderate slopes having slightly more recovery (3–5%). High slope angles (greater than 45.0� ) had larger shrub regrowth than the other slopes in two of the three sites. However, it was noted that the third site received less precipitation than the other two sites. Aspect was also a major factor in the amount of vegetation regrowth observed. The aspect categories were statistically comparable between treatment types as the F-value (1.22) was less than F-critical (2.18) (Appendix Table A-4). North aspects had lower vegetation levels compared to other aspects. The greatest recovery in the Waldo Canyon Fire occurred in the west aspects, where vegetation returned to 81% of pre-fire vegetation for wood and 80% for straw treatments after 5 years. Burned-Untreated sites had the smallest vegetation recovery in the west aspect and largest vegetation recovery in north aspect (46% and 64% of pre-fire vegetation, respectively). This is contrary to Schroeder et al. (2007), who used a Landsat analysis and observed that north aspects in Oregon encountered three times the vegetation regrowth compared to other slope aspects. However, in that study, the planting was performed on shallow slopes and low elevations, which was not the case in the Waldo Canyon Fire. Most of the north-facing wood treatments were implemented on slopes greater than 24� . Thus, there were less data points available for this study; only three of the 49 wood treated areas were north aspects. Rachels et al. (2016) found that in studies of the 2003 Cedar Fire (San Diego, California), shrub regrowth was lowest on south aspects, while northern slope aspects had the lowest relative change from pre-fire conditions. In Kinoshita and Hogue (2011), north aspects also recovered at a greater pace, at 90% of pre-fire vegetation, and 75–85% for the other aspects for 7 years after fire. It is expected that a longer-term study of the Waldo Canyon Fire vegetative growth would observe larger recovery trends as evergreen trees reestablish. Vegetative levels responded to large storm events with a peak in vegetation regrowth, but this response was not sustained and over the year the vegetation decreased (Fig. 2). Abundant precipitation in the Waldo Canyon Fire likely promoted temporary vegetation regrowth, such as July 2014 (99% of pre-fire average for Burned-Wood and 69% for Burned-Straw) and July 2016 (79% for Burned-Wood and 81% for Burned-Straw). However, the vegetative levels then decreased to postfire averages, often within the same year. This suggests that the vege­ tation could not become established, even after periods of accelerated growth encouraged by a large storm. Kunze and Stednick (2006) had a similar observation in the Bobcat Fire in northern Colorado. Dry mulch, seeding, and contour LEBs were implemented as post-fire rehabilitation. However, large storm events occurred two months after the fire and again the next summer, overwhelming the treatments and obscuring the effects of the treatment. Kinoshita and Hogue (2011) observed a stron­ ger recovery for the Old Fire in California and found that in an extremely wet post-fire year (2005), vegetation levels increased to 80–100% of pre-fire levels. This wet year season surge of vegetation was not sus­ tained and vegetation levels returned to lower levels for the next four years.

Table 2 Average annual EVI values and standard deviation with respect to north, east, south, and west-facing slope aspects. Grey shading highlights statistically different values from the pre-fire average (α ¼ 0.05). N/A denotes slope aspect and treatment with unavailable information. Aspect

Year

BurnedUntreated

BurnedStraw

BurnedWood

UnburnedHunter’s Run

UnburnedFountain Creek

North

PreFire Avg 2012

0.135 � 0.050

N/A

0.098 � 0.036

N/A

N/A

N/A

0.278 � 0.070

2013 2014 2015 2016 East

PreFire Avg 2012 2013 2014 2015 2016

South

PreFire Avg 2012 2013 2014 2015 2016

West

PreFire Avg 2012 2013 2014 2015 2016

0.021 � 0.073 0.079 � 0.056 0.038 � 0.052 0.052 � 0.087 0.051 � 0.054 0.283 � 0.018

0.252 � 0.028

0.071 � 0.041 0.067 � 0.048 0.016 � 0.041 0.036 � 0.051 0.016 � 0.0.44 0.292 � 0.028

0.130 � 0.056 0.127 � 0.047 0.126 � 0.041 0.128 � 0.039 0.153 � 0.043 0.293 � 0.022

0.076 � 0.020 0.083 � 0.037 0.090 � 0.028 0.114 � 0.074 0.119 � 0.037 0.305 � 0.016

0.080 � 0.032 0.086 � 0.033 0.092 � 0.028 0.097 � 0.047 0.112 � 0.028 0.282 � 0.021

0.120 � 0.036 0.125 � 0.039 0.133 � 0.026 0.135 � 0.039 0.145 � 0.041 0.248 � .017

0.080 � 0.030 0.072 � 0.026 0.086 � 0.018 0.056 � 0.036 0.102 � 0.024 0.241 � 0.013

0.095 � 0.018 0.109 � 0.031 0.118 � 0.025 0.101 � 0.034 0.137 � 0.027 0.207 � 0.029

0.138 � 0.083 0.131 � 0.067 0.124 � 0.064 0.122 � 0.068 0.149 � 0.064

0.055 � 0.024 0.041 � 0.022 0.053 � 0.024 0.019 � 0.018 0.070 � 0.019

0.037 � 0.035 0.029 � 0.035 0.045 � 0.033 0.029 � 0.036 0.057 � 0.031

0.210 0.190 0.201 0.211 0.213 0.256 � 0.124

N/A

0.256 0.262 0.265 0.261 0.278 N/A

N/A

4.2. Geomorphic and climatic vegetation patterns after fire The three slope categories could be compared (F-value ¼ 0.282 < Fcritical ¼ 2.484). Burned-Straw sites on steeper slopes exceeding 30� had EVI values that increased to about 33% of the pre-fire levels. This is in part due to the lower density and weight of straw mulch, which could provide more effective soil cover on the steeper slopes (Robichaud et al., 2013). Badia and Marti (2000) observed that despite decomposition and removal of straw by wind and water, straw mulch could still maintain about 43%–69% vegetative biomass. For low to moderate slopes, vegetation levels were similar. Field-based studies have shown both wood and straw mulches performed best for ground cover applications for slopes within 20–60% (Napper, 2006). This study evaluated

4.3. Impacts of post-fire rehabilitation on vegetation Post-fire treated areas had mixed impacts on vegetation, which is also noted in other field-based studies. Vegetation increased in burned areas with straw application from 23% to 35% from 2012 to 2016, but did not return to pre-fire conditions. The results from this study had a much lower observed vegetation recovery after straw treatment than �ndez and Vega (2016). Ferna �ndez and Vega (2016) observed that Ferna straw treatment reduced soil loss and increased vegetation coverage in Sierra do Barbanza, Spain. The plant recovery in Spain varied from 32% 7

V.D. Vo and A.M. Kinoshita

Journal of Environmental Management 260 (2020) 109993

Table 3 Average annual EVI values and standard deviations with respect to low, moderate, and high slopes. Shading highlights statistically different values from the pre-fire period (α ¼ 0.05). N/A denotes slope aspect and treatment with unavailable information. Slope

Year

Burned-Untreated

Burned-Straw

Burned-Wood

Unburned-Hunter’s Run

Unburned-Fountain Creek

Low (0–15� )

Pre-Fire Avg 2012 2013 2014 2015 2016 Pre-Fire Avg 2012 2013 2014 2015 2016 Pre-Fire Avg 2012 2013 2014 2015 2016

0.250 0.137 0.143 0.129 0.116 0.154 0.276 0.114 0.124 0.121 0.124 0.139 0.213 0.088 0.085 0.084 0.107 0.105

0.286 � 0.069 � 0.061 � 0.079 � 0.036 � 0.092 � 0.252 � 0.070 � 0.064 � 0.072 � 0.068 � 0.095 � 0.289 � 0.080 � 0.072 � 0.086 � 0.056 � 0.102 �

N/A

N/A

N/A

0.256 � 0.124 0.256 0.262 0.265 0.261 0.278 N/A

0.278 � 0.070 0.210 0.190 0.201 0.211 0.213 N/A

Moderate (15–30� )

High (30� þ)

� 0.013 � 0.59 � 0.045 � 0.037 � 0.061 � 0.047 � 0.019 � 0.073 � 0.057 � 0.057 � 0.063 � 0.062 � 0.034 � 0.078 � 0.044 � 0.063 � 0.062 � 0.061

0.023 0.022 0.024 0.015 0.032 0.023 0.015 0.025 0.35 0.028 0.065 0.035 0.026 0.037 0.037 0.040 0.063 0.038

to 52% in the first year and about 77%–99% in the third year. Fernandez and Vega noted that the precipitation during the first two years of the study were similar to the average annual precipitation, while the third �ndez and Vega (2016) also found that year was below average. Ferna straw mulch cover declined rapidly in the first year, up to 53%. The decline in vegetative ground cover in the first post-fire year for the Waldo Canyon Fire was comparatively less, at 8.8%. However, �ndez and Vega (2016) used much smaller control plots in their Ferna study, and only observed the first three years post-fire. Kruse et al. (2004) also observed mixed results following the Megram Fire in the northern California, where straw mulching and seeding with barley did not promote regrowth beyond natural recovery. Similarly, Dodson and Peterson (2010) noted that vegetation recovery in the Tripod Fire in Washington was largely unaffected by mulching. Straw mulching had the small benefit of retaining soil moisture and limiting establishment of non-native vegetation. Differences between these studies and the cur­ rent study may be attributed to the experimental set up, site conditions, or discrepancies between satellite-based information and in situ observations. The primary goal of wood and straw mulching implementation is not to improve vegetative recovery. Rather, mulching is used to control soil moisture and temperature, prevent invasive species propagation, reduce overland flow, prevent erosion, and lessen the impact of rain drops (Robichaud et al., 2000). Mulching provides immediate ground cover and protection to soils where erosion potential is high (Napper, 2006). While vegetation recovery is often a secondary objective, using EVI as a proxy for vegetative growth can be an indirect measure of the treatment effectiveness to stabilize the soil and provide ground cover. Regulating soil moisture and temperature promotes plant growth (Goodwin et al., 2002). Effective ground cover from the straw and wood mulch may discourage plant mortality. There is a general observed trend across the entire study period (2007–2016), in which the burned areas treated with wood and straw have less seasonal variability compared to the Unburned-Untreated sites, indicating potentially different vegetation types or processes. This may be attributed to the soil and ambient air temperature. In the Mediterranean forests in the Zuera Mountains, Spain, areas that were not treated with wood management strategies after a fire observed an average of 1 � C cooler temperatures, and were 10% greener (Vlassova and P�erez-Cabello, 2016). The Aleppo pine in Spain was more amicable to post-fire regrowth due to seed release and germination by smoke or heat, which assisted in the reestablishment of plant communities. In comparison, the vegetation in the Waldo Canyon Fire area did not have the same plant adaptation characteristics or climate conditions. Rosgen and Rosgen (2013b) noted that the Pike National Forest, the site of Waldo Canyon Fire, had shallow soils, cold

0.265 0.070 0.074 0.085 0.080 0.102 N/A

� 0.023 � 0.042 � 0.048 � 0.041 � 0.051 � 0.044

climates, and low precipitation, which caused unfavorable conditions for regrowth. These factors inhibited the rapid and significant vegetative regeneration found at other post-fire rehabilitation studies. Future in­ vestigations that provide field information could provide insight into favorable vegetative regrowth conditions created by mulching. 4.4. Uncertainties and limitations While the satellite-based products are convenient and widely avail­ able in time and space, we note that there are inherent uncertainty and limitations associated with the data and analyses. Study sites were selected randomly to minimize bias and provide a large sample of lo­ cations and conditions throughout the Waldo Canyon Fire. This method resulted in a smaller dataset of north-facing aspects and high slopes. Areas treated with mulch tended to be aspects other than north and at low or moderate slopes. On average, Burned-Treated sites experienced lower post-fire vegetation regrowth than Burned-Untreated areas. Considering the dominant evergreen forest cover, the difference be­ tween treated and untreated areas may have been more pronounced. However, this effect would have lessened throughout the study as nat­ ural vegetation regrowth overtook the post-fire treatments. Typically, this period could be the first three years after the fire, where the EVI was about 80% of pre-fire levels (Wittenberg et al., 2007). The ANOVA analysis confirmed that the means between the various slope and aspect categories were comparable. However, if possible, future studies should incorporate more sites with a greater balance between geomorphic properties (slope and aspect). The evaluation of vegetation regrowth is highly dependent on the number of years analyzed. In the Waldo Canyon Fire, vegetation con­ ditions remained at about 60–75% of pre-fire levels, even after five years. This is likely attributed to the evergreen vegetation type in the Pike National Forest. Wittenberg et al. (2007), who focused on an arid region in Mt. Carmel, Israel, vegetation that generally showed seasonal responses returned to 80%–100% of pre-disturbance levels within less than five years. Meanwhile, Fornwalt et al. (2018) observed that tree snag density returned to pre-fire levels by the tenth year after fire in high burn severity areas in the Colorado Hayman Fire. The high-severity areas in the Hayman Fire had experienced an immediate and long-lived reduction in live overstory density. Thus, an extension of the study period for the Waldo Canyon Fire may be needed to further assess when vegetation levels returns to pre-fire conditions. This research used all available data from Landsat 7 for the duration of the study. More sensitive bands for reflectance and corrections for interference (USGS, 2019) could also improve future representation of vegetation recovery. Further, this work could be enhanced as additional Landsat 8 scenes or 8

V.D. Vo and A.M. Kinoshita

Journal of Environmental Management 260 (2020) 109993

newer remote sensing products become available.

high slopes. Similar to other vegetation recovery studies, the treated areas also responded to large storm events with a sudden increase in biomass, which was not sustained for the duration of the study. This is one of the first studies to contribute to the understanding of post-fire rehabilitation impacts on vegetation recovery using remote sensing data, augmenting the number of sites and length of time analyzed. Additional field studies to corroborate the results of this work could help to provide better information for management and treatment implementation. As the frequency and severity of wildfires increase and the need to minimize secondary impacts to society, studies that utilize remote sensing can be used to improve our insights into the effects of post-fire rehabilitation treatment on vegetation recovery.

5. Conclusion The use of large-scale remote sensing data has not been specifically used to investigate the impacts of post-fire rehabilitation treatments on vegetation. Remote sensing as a proxy for vegetation regrowth can be utilized to efficiently evaluate vegetation in regions with large spatial and temporal resolution. This work used satellite-based datasets, pri­ marily consisting of EVI, elevation, slope, slope aspect, precipitation, and land cover, which was integrated into a single collection and analyzed collectively. In general, for the study period, the post-fire conditions did not re­ turn to pre-fire levels. After the Waldo Canyon Fire, there was an observed reduction in vegetation of 30–40% compared to pre-fire con­ ditions in all cases. The total amount of vegetation levels lost in the Burned-Treated sites was larger than Burned-Untreated areas following the fire. Consequently, during the post fire years, Burned-Untreated sites tended to stabilize to pre-fire conditions sooner than Burned-Treated sites. Thus there is typically a preference to treat areas with more se­ vere loss of vegetation. Of the two treatments investigated, the increase from the first post-fire year (2012) to the last post-year (2016) in vegetative levels was 31.2% in Burned-Wood, 25.9% in Burned-Straw, and 15.7% in Burned-Untreated. While there was substantial regrowth in mulch treated sites during the post-fire study period, vegetation did not return to pre-fire vegetation conditions. Burned-Treated sites were more resistant to seasonal changes, with an average seasonal vegetation loss of 45% compared to 72% loss in Burned-Untreated sites. East slope aspects experienced the most vegetation regrowth of 32% compared to 26% in the other slopes. It was observed that moderate slopes had the vegetation regrowth of 26%–31% compared to 25% for low and 22% for

Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. CRediT authorship contribution statement Viet D. Vo: Conceptualization, Methodology, Software, Formal analysis, Writing - original draft, Writing - review & editing, Visualiza­ tion. Alicia M. Kinoshita: Conceptualization, Methodology, Writing original draft, Writing - review & editing, Supervision. Acknowledgements The authors would like to acknowledge Mandy Moore of Waldo Canyon BAER Implementation Team for providing the treatment maps for the Waldo Canyon Fire. The authors would also like to thank five anonymous reviewers and the Editor for their suggestions.

Appendix. A: Data and ANOVA tests

Table A-1 Annual EVI for treatment type and slope. Slope

Straw

Wood

B-Ut

Year

Low (1–15� )

Medium (15–30� )

High (>30� )

2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2007 2008 2009 2010 2011 2012 2013

0.256 0.279 0.308 0.309 0.278 0.069 0.061 0.079 0.036 0.092 0.233 0.259 0.289 0.286 0.256 0.070 0.074 0.085 0.080 0.102 0.250 0.235 0.247 0.272 0.248 0.137 0.143

0.251 0.253 0.273 0.280 0.247 0.070 0.064 0.072 0.068 0.095 0.233 0.259 0.289 0.286 0.256 0.070 0.074 0.085 0.080 0.102 0.257 0.270 0.287 0.303 0.262 0.114 0.124

0.261 0.268 0.287 0.275 0.259 0.081 0.076 0.076 0.072 0.103 0.233 0.259 0.289 0.286 0.256 0.070 0.074 0.085 0.080 0.102 0.194 0.203 0.194 0.274 0.202 0.088 0.085 (continued on next page)

9

V.D. Vo and A.M. Kinoshita

Journal of Environmental Management 260 (2020) 109993

Table A-1 (continued ) Slope Year

Low (1–15� )

Medium (15–30� )

High (>30� )

2014 2015 2016

0.129 0.116 0.154

0.121 0.124 0.139

0.084 0.107 0.105

Table A-2 ANOVA analysis for treatment type and slope Anova: Two-Factor With Replication SUMMARY straw Count Sum Average Variance wood Count Sum Average Variance b-ut Count Sum Average Variance Total Count Sum Average Variance

Low (1–15)

Medium (15–30)

High (30þ)

Total

10 1.7675 0.1768 0.0137

10 1.6749 0.1675 0.0098

10 1.7583 0.1758 0.0100

30 5.2007 0.1734 0.0104

10 1.736 0.174 0.010

10 1.736 0.174 0.010

10 1.736 0.174 0.010

30 5.207 0.174 0.009

10 1.930 0.193 0.004

10 2.002 0.200 0.007

10 1.536 0.154 0.005

30 5.468 0.182 0.005

30 5.433 0.181 0.008

30 5.412 0.180 0.008

30 5.030 0.168 0.008

ANOVA Source of Variation

SS

df

MS

F

P-value

F crit

Sample Columns Interaction Within Total

0.002 0.003 0.010 0.695 0.709

2 2 4 81 89

0.001 0.002 0.002 0.009

0.091 0.201 0.282

0.914 0.819 0.889

3.109 3.109 2.484

Table A-3 Annual EVI for treatment type and aspect. Aspect

Straw

Wood

B-ut

Year

North

East

South

West

2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2007 2008 2009 2010

0.232 0.234 0.253 0.257 0.231 0.055 0.041 0.053 0.019 0.070 0.084 0.087 0.080 0.162 0.077 0.071 0.067 0.016 0.036 0.016 0.097 0.150 0.108 0.218

0.225 0.241 0.262 0.295 0.235 0.076 0.083 0.090 0.114 0.119 0.253 0.294 0.327 0.307 0.280 0.080 0.086 0.092 0.097 0.112 0.264 0.274 0.300 0.304

0.287 0.300 0.328 0.311 0.297 0.080 0.072 0.086 0.056 0.102 0.250 0.281 0.309 0.288 0.283 0.095 0.109 0.118 0.101 0.137 0.306 0.266 0.289 0.323

0.232 0.234 0.253 0.257 0.231 0.055 0.041 0.053 0.019 0.070 0.189 0.183 0.212 0.255 0.197 0.037 0.029 0.045 0.029 0.057 0.223 0.246 0.250 0.272 (continued on next page)

10

V.D. Vo and A.M. Kinoshita

Journal of Environmental Management 260 (2020) 109993

Table A-3 (continued ) Aspect Year

North

East

South

West

2011 2012 2013 2014 2015 2016

0.105 0.021 0.079 0.038 0.052 0.051

0.274 0.130 0.127 0.126 0.128 0.153

0.281 0.120 0.125 0.133 0.135 0.145

0.247 0.138 0.131 0.124 0.122 0.149

Table A-4 ANOVA analysis for treatment type and aspect. Anova: Two-Factor With Replication SUMMARY Straw Count Sum Average Variance wood Count Sum Average Variance b-ut Count Sum Average Variance Total Count Sum Average Variance

North

East

South

West

Total

10 1.444 0.144 0.011

10 1.739 0.174 0.007

10 1.918 0.192 0.014

10 1.444 0.144 0.011

40 6.545 0.164 0.010

10 0.356 0.036 0.006

10 1.927 0.193 0.011

10 1.970 0.197 0.008

10 1.233 0.123 0.008

40 5.486 0.137 0.012

10 0.918 0.092 0.003

10 2.081 0.208 0.006

10 2.123 0.212 0.008

10 1.901 0.190 0.004

40 7.023 0.176 0.007

30 2.718 0.091 0.008

30 5.747 0.192 0.008

30 6.011 0.200 0.009

30 4.578 0.153 0.008

ANOVA Source of Variation

SS

df

MS

F

P-value

F crit

Sample Columns Interaction Within Total

0.031 0.225 0.060 0.880 1.1958872

2 3 6 108 119

0.015 0.075 0.010 0.008

1.896 9.189 1.221

0.155 0.000 0.301

3.080 2.689 2.184

Table A-5 Annual EVI by Treatment Year

Y2007

Y2008

Y2009

Y2010

Y2011

Y2012

Y2013

Y2014

Y2015

Y2016

Straw Wood B-UT

0.254 0.233 0.243

0.264 0.259 0.248

0.287 0.289 0.258

0.289 0.286 0.289

0.260 0.256 0.247

0.071 0.070 0.115

0.065 0.074 0.121

0.076 0.085 0.116

0.059 0.080 0.119

0.096 0.102 0.137

Table A-6 ANOVA analysis for annual EVI by treatment SUMMARY

Count

Sum

Average

Variance

Straw Wood B-UT Y2007 Y2008 Y2009 Y2010 Y2011 Y2012 Y2013 Y2014 Y2015

10 10 10 3 3 3 3 3 3 3 3 3

1.720821 1.735571 1.893762 0.729861 0.771355 0.834787 0.865137 0.762834 0.256575 0.260409 0.277177 0.257409

0.172082 0.173557 0.189376 0.243287 0.257118 0.278262 0.288379 0.254278 0.085525 0.086803 0.092392 0.085803

0.011063 0.009564 0.005282 0.000105 6.72E-05 0.000303 2.68E-06 4.84E-05 0.000669 0.000916 0.000444 0.00093 (continued on next page)

11

Journal of Environmental Management 260 (2020) 109993

V.D. Vo and A.M. Kinoshita

Table A-6 (continued ) SUMMARY

Count

Sum

Average

Variance

Y2016

3

0.334611

0.111537

0.000489

ANOVA Source of Variation

SS

df

MS

F

P-value

F crit

Rows Columns Error Total

0.001838 0.227071 0.00611 0.235019

2 9 18 29

0.000919 0.02523 0.000339

2.707834 74.32668

0.093733 2.17E-12

3.554557 2.456281

Table A-7 EVI by season and treatment. Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall 2007 2007 2007 2007 2008 2008 2008 2008 2009 2009 2009 2009 2010 2010 2010 2010 2011 2011 2011 2011 Wood 0.215 Straw 0.234 B-UT 0.214

0.288 0.285 0.279

0.311 0.264 0.300

0.203 0.346 0.240 0.306 0.229 0.305

0.315 0.321 0.311

0.312 0.278 0.301

0.301 0.258 0.309 0.250 0.272 0.212

0.282 0.285 0.286

0.322 0.305 0.311

0.281 0.289 0.271 0.293 0.246 0.252

0.413 0.394 0.395

0.360 0.344 0.336

0.295 0.261 0.312 0.284 0.284 0.249

0.291 0.302 0.289

0.288 0.282 0.287

0.275 0.277 0.246

Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall 2012 2012 2012 2012 2013 2013 2013 2013 2014 2014 2014 2014 2015 2015 2015 2015 2016 2016 2016 2016 Wood 0.267 Straw 0.277 B-UT 0.242

0.280 0.227 0.264

0.102 0.082 0.149

0.117 0.089 0.104 0.090 0.140 0.129

0.212 0.232 0.244

0.227 0.144 0.228

0.128 0.096 0.117 0.091 0.161 0.126

0.163 0.181 0.212

0.261 0.187 0.272

0.116 0.200 0.102 0.141 0.143 0.195

0.193 0.177 0.230

0.232 0.204 0.263

0.184 0.129 0.174 0.104 0.210 0.146

0.210 0.219 0.253

Table A-8 ANOVA analysis for treatment time and season. Anova: Two-Factor Without Replication SUMMARY

Count

Sum

Average

Variance

Wood Straw B-UT Winter 2007 Spring 2007 Summer 2007 Fall 2007 Winter 2008 Spring 2008 Summer 2008 Fall 2008 Winter 2009 Spring 2009 Summer 2009 Fall 2009 Winter 2010 Spring 2010 Summer 2010 Fall 2010 Winter 2011 Spring 2011 Summer 2011 Fall 2011 Winter 2012 Spring 2012 Summer 2012 Fall 2012 Winter 2013 Spring 2013 Summer 2013 Fall 2013 Winter 2014 Spring 2014 Summer 2014 Fall 2014 Winter 2015 Spring 2015 Summer 2015

40 40 40 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3

9.488667 9.048 9.642 0.663 0.852 0.875333 0.671667 0.957 0.947333 0.891333 0.881333 0.719667 0.852667 0.938333 0.798333 0.833667 1.202 1.041 0.891 0.794333 0.882 0.856667 0.798667 0.786 0.770667 0.332667 0.360667 0.307333 0.688333 0.599333 0.406 0.312667 0.556667 0.72 0.361333 0.535667 0.6 0.699

0.237217 0.2262 0.24105 0.221 0.284 0.291778 0.223889 0.319 0.315778 0.297111 0.293778 0.239889 0.284222 0.312778 0.266111 0.277889 0.400667 0.347 0.297 0.264778 0.294 0.285556 0.266222 0.262 0.256889 0.110889 0.120222 0.102444 0.229444 0.199778 0.135333 0.104222 0.185556 0.24 0.120444 0.178556 0.2 0.233

0.006349 0.006709 0.003848 0.000134 2.41E-05 0.000607 0.000359 0.00056 2.71E-05 0.000302 0.000384 0.000585 4.93E-06 7.38E-05 0.000338 0.000507 0.000108 0.000149 0.000193 0.000323 5.28E-05 9.93E-06 0.000297 0.000317 0.000754 0.001193 0.000346 0.000516 0.000266 0.002334 0.000513 0.000363 0.000622 0.002165 0.000442 0.001045 0.000739 0.000891 (continued on next page)

12

0.239 0.230 0.272

0.138 0.129 0.159

V.D. Vo and A.M. Kinoshita

Journal of Environmental Management 260 (2020) 109993

Table A-8 (continued ) Anova: Two-Factor Without Replication SUMMARY

Count

Sum

Average

Variance

Fall 2015 Winter 2016 Spring 2016 Summer 2016 Fall 2016

3 3 3 3 3

0.568 0.379 0.681333 0.741 0.425667

0.189333 0.126333 0.227111 0.247 0.141889

0.000342 0.000441 0.000512 0.000492 0.000234

ANOVA Source of Variation

SS

df

MS

F

P-value

F crit

Rows Columns Error Total

0.004754 0.624941 0.034385 0.664081

2 39 78 119

0.002377 0.016024 0.000441

5.39258 36.34969

0.006403 1.95E-36

3.113792 1.553239

References

Moody, J.A., Shakesby, R.A., Robichaud, P.R., Cannon, S.H., Martin, D.A., 2013. Current research issues related to post-wildfire runoff and erosion processes. Earth Sci. Rev. 122, 10–37. Elsevier B.V. Moore, M., Park, D., 2012. Hydrology Resource Report Waldo Canyon Fire BAER Assessment – Pikes Peak Ranger District Pike National Forest, pp. 1–10. Morgan, P., Moy, M., Lentile, L.B., Lentile, L.B., Lewis, S.A., Robichaud, P.R., Hudak, A. T., 2014. Vegetation response after post-flre mulching and native grass seeding. Fire Ecol. 10 (3), 49–62. Napper, C. (2006). “Burned area emergency response treatments catalog." https://www. fs.fed.us/t-d/pubs/pdf/BAERCAT/lo_res/06251801L.pdf.” (December), 1–266. Neary, D.G., Ryan, K.C., DeBano, L.F., 2005. Wildland Fire in Ecosystems: Effects of Fire on Soils and Water. Gen. Tech. Rep. RMRS-GTR-42, vol. 4. U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Ogden, UT, p. 250. NOAA, 2019. The Climate of Colorado Springs. CO.”. Parson, A., Robichaud, P.R., Lewis, S.A., Napper, C., Clark, J.T., 2010. Field Guide for Mapping Post-fire Soil Burn severity.” Gen. Tech. Rep. RMRS-GTR-243. U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, CO, p. 49. Pyke, D.A., Wirth, T.A., Beyers, J.L., 2013. Does seeding after wildfires in rangelands reduce erosion or invasive species? Restor. Ecol. 21 (4), 415–421. Rachels, D.H., Stow, D.A., O’Leary, J.F., Johnson, H.D., Riggan, P.J., 2016. Chaparral recovery following a major fire with variable burn conditions. Int. J. Remote Sens. https://doi.org/10.1080/01431161.2016.1204029. Robichaud, P.R., Beyers, J.L., Neary, D.G., 2000. Evaluating the Effectiveness of Post-fire Rehabilitation Treatments. USDA Forest Service, Rocky Mountain Research Station, General Technical Report (September). Robichaud, P.R., Ashmun, L.E., Sims, B.D., 2010. Post-fire Treatment Effectiveness for Hillslope Stabilization. USDA Forest Service General Technical Report RMRS-GTR240. Robichaud, P.R., Ashmun, L.E., Foltz, R.B., Showers, C.G., Groenier, J.S., Kesler, J., DeLeo, C., Moore, M., 2013. Production and Aerial Application of Wood Shreds as a Post-fire Hillslope Erosion Mitigation Treatment. USDA Forest Service - General Technical Report RMRS-GTR, pp. 1–16, 307 GTR. Robichaud, P.R., Rhee, H., Lewis, S.A., 2014. A synthesis of post-fire Burned Area Reports from 1972 to 2009 for western US Forest Service lands: Trends in wildfire characteristics and post-fire stabilisation treatments and expenditures. Int. J. Wildland Fire 23 (7), 929–944. Rosgen, D., Rosgen, B., 2013a. Waldo Canyon Fire Watershed Assessment : the WARSSS Results. Rosgen, D., Rosgen, B., 2013b. The Waldo Canyon Fire Master Plan for Watershed Restoration & Sediment Reduction. Scasta, J.D., Weir, J.R., Stambaugh, M.C., 2016. Droughts and Wildfires in Western U.S. Rangelands. Rangelands 38 (4), 197–203. The Authors. Schroeder, T.A., Cohen, W.B., Yang, Z., 2007. Patterns of forest regrowth following clearcutting in western Oregon as determined from a Landsat time-series. For. Ecol. Manag. 243 (2–3), 259–273. Shvetsov, E.G., Kukavskaya, E.A., Buryak, L.V., Barrett, K., 2019. Assessment of post-fire vegetation recovery in Southern Siberia using remote sensing observations. Environ. Res. Lett. 14 (5), 055001. IOP Publishing. Smith, H.G., Sheridan, G.J., Lane, P.N.J., Nyman, P., Haydon, S., 2011. Wildfire effects on water quality in forest catchments: A review with implications for water supply. J. Hydrol. 396 (1–2), 170–192. Elsevier B.V. Storey, E.A., Stow, D.A., O’Leary, J.F., 2016. Assessing Postfire Recovery of Chamise Chaparral Using Multi-Temporal Spectral Vegetation Index Trajectories Derived from Landsat Imagery. Remote Sensing of Environment. USGS, 2012. USGS National Elevation Dataset Courtesy of the U.S. Geological Survey. USGS, 2019. USGS Landsat 7 Collection Courtesy of the U.S. Geological Survey. Viana-Soto, A., Aguado, I., Martínez, S., 2017. Assessment of Post-Fire Vegetation Recovery Using Fire Severity and Geographical Data in the Mediterranean Region (Spain). Environments 4 (4), 90. Vlassova, L., P� erez-Cabello, F., 2016. Effects of post-fire wood management strategies on vegetation recovery and land surface temperature (LST) estimated from Landsat images. Int. J. Appl. Earth Obs. Geoinf. 44 https://doi.org/10.1016/j. jag.2015.08.011.

BAER, 2012. Waldo Canyon BAER Implementation Aerial Mulching Units (Final). US Department of Agriculture, Forest Service. Badia, D., Marti, C., 2000. Seeding and mulching treatments as conservation measures of two burned soils in the central ebro valley, ne Spain. ” Arid Soil Research, Rehabilitation. Badía, D., Martí, C., Aguirre, J., Echeverría, M.T., Ibarra, P., 2008. Erodibility and hydrology of arid burned soils: soil type and revegetation effects. Arid Land Res. Manag. 22 (4), 286–295. Cerd� a, A., Robichaud, P.R., 2009. Fire Effects on Soils and Restoration Strategies. Volume 5 of Series: Land Reconstruction and Management. Science Publishers, Enfield, NH. Chander, G., Markham, B.L., Helder, D.L., 2009. Summary of Current Radiometric Calibration Coefficients for Landsat MSS, TM, ETMþ, and EO-1 ALI Sensors. Remote Sensing of Environment. Chen, X., Vierling, L., Deering, D., Conley, A., 2005. Monitoring boreal forest leaf area index across a Siberian burn chronosequence: a MODIS validation study. Int. J. Remote Sens. 26 (24), 5433–5451. Chen, W., Moriya, K., Sakai, T., Koyama, L., Cao, C., 2014. Monitoring of post-fire forest recovery under different restoration modes based on time series Landsat data. Europ. J. Rem. Sens. 47 (1), 153–168. https://doi.org/10.5721/EuJRS20144710. Chin, A., Solverson, A.P., O’Dowd, A.P., Florsheim, J.L., Kinoshita, A.M., Nourbakhshbeidokhti, S., Sellers, S.M., Tyner, L., Gidley, R., 2019. Interacting geomorphic and ecological response of step-pool streams after wildfire. GSA Bull. Daly, C., Halbleib, M., Smith, J.I., Gibson, W.P., Doggett, M.K., Taylor, G.H., Curtis, J., Pasteris, P.P., 2008. Physiographically sensitive mapping of climatological temperature and precipitation across the conterminous United States. Int. J. Climatol. Daly, C., Smith, J.I., Olson, K.V., 2015. Mapping atmospheric moisture climatologies across the conterminous United States. PLoS One. Dodson, E.K., Peterson, D.W., 2010. Mulching effects on vegetation recovery following high severity wildfire in north-central Washington State, USA. For. Ecol. Manag. 260, 1816–1823. https://doi.org/10.1016/j.foreco.2010.08.026. Fern� andez, C., Vega, J.A., 2016. Are erosion barriers and straw mulching effective for controlling soil erosion after a high severity wildfire in NW Spain. Ecol. Eng. 87, 132–138. Elsevier B.V. Fornwalt, P.J., Stevens-Rumann, C.S., Collins, B.J., 2018. Overstory structure and surface cover dynamics in the decade following the hayman fire, Colorado. Forests 9 (3), 1–17. Goodwin, K., Sheley, R., Clark, J., 2002. Integrated Noxious Weed Management after Wildfires. Hallema, D.W., Sun, G., Caldwell, P.V., Norman, S.P., Cohen, E.C., Liu, Y., Bladon, K.D., McNulty, S.G., 2018. Burned forests impact water supplies. Nat. Commun. 9 (1), 1–8. Springer US. Higuera, P.E., Abatzoglou, J.T., Littell, J.S., Morgan, P., 2015. The changing strength and nature of fire-climate relationships in the northern Rocky Mountains, U.S.A., 19022008. PLoS One 10 (6), 1–21. Homer, C., Dewitz, J., Yang, L., Jin, S., Danielson, P., Xian, G., Coulston, J., Herold, N., Wickham, J., Megown, K., 2015. Completion of the 2011 national land cover database for the conterminous United States – representing a decade of land cover change information. Photogramm. Eng. Remote Sens. Huete, A., Didan, K., Miura, T., Rodriguez, E., Gao, X., Ferreira, L., 2002. Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sens. Environ. 83 (1–2), 195–213. Kinoshita, A.M., Hogue, T.S., 2011. Spatial and temporal controls on post-fire hydrologic recovery in Southern California watersheds. Catena 87 (2), 240–252. Elsevier B.V. Kinoshita, A.M., Chin, A., Simon, G.L., Briles, C., Hogue, T.S., O’Dowd, A.P., Gerlak, A.K., Albornoz, A.U., 2016. Wildfire, water, and society: Toward integrative research in the ‘Anthropocene. Elsevier B.V Anthropocene 16, 16–27. February 2018. Kruse, R., Bend, E., Bierzychudek, P., 2004. Native plant regeneration and introduction of non-natives following post-fire rehabilitation with straw mulch and barley seeding. For. Ecol. Manag. 196 (2–3), 299–310. Kunze, M.D., Stednick, J.D., 2006. Streamflow and Suspended Sediment Yield Following the 2000 Bobcat Fire, Colorado. Hydrological Processes.

13

V.D. Vo and A.M. Kinoshita

Journal of Environmental Management 260 (2020) 109993

Wagenbrenner, J.W., MacDonald, L.H., Rough, D., 2006. Effectiveness of tree post-fire rehabilitation treatments in the Colorado Front Range. Hydrol. Process. 20 (14), 2989–3006. Wittenberg, L., Malkinson, D., Beeri, O., Halutzy, A., Tesler, N., 2007. Spatial and temporal patterns of vegetation recovery following sequences of forest fires in a Mediterranean landscape, Mt. Carmel Israel. Catena 71 (1), 76–83.

Young, D., Rust, B., 2012. Waldo Canyon Fire – Burned Area Emergency Response Soil Resource Assessment Executive Summary – Soil Resource Condition Assessment, pp. 1–15.

14