MONITORING POST-FIRE REVEGETATION: MILFORD FLAT FIRE, UTAH Lisa Langs Stoner, GIS Analyst Remote Sensing & GIS Laboratory Utah State University Logan, UT 84322 [email protected] ABSTRACT In July 2007, Utah experienced the largest fire in the state’s history, burning approximately 1,469 square kilometers in Millard and Beaver counties, involving several different plant communities, primarily salt desert scrub, sagebrush, and juniper. The Milford Flat Fire was lightning-caused but fueled by a number of converging conditions including drought, high winds, and an abundance of annual weeds. The Remote Sensing & GIS Laboratory at Utah State University worked with the Utah Bureau of Land Management (BLM) to develop a remote sensing-based monitoring protocol to assess the effects of the fire on the landscape and to evaluate seeding program success. A 10year sequence of phenological profiles were developed from MODIS 16-day composite Normalized Differenced Vegetation Index (NDVI) data to compare vegetation responses before and after the fire at both a fire-wide and treatment scale. The phenological response of annual species such as cheatgrass (Bromus tectorum) was assessed relative to the re-establishment of perennial vegetation. With a favorable climate during the three years following the fire, the BLM’s seeding program is on a positive trajectory. KEYWORDS: wildfire, revegetation, monitoring, MODIS, Utah INTRODUCTION The Milford Flat Fire (MFF), Utah’s largest wildland fire, burned from 6 July 2007 to 15 July 2007. This lightning-caused, fast moving fire started three miles north of the town of Milford, Utah, consuming more than 1,469 square kilometers in the Black Rock Desert region of Beaver and Milford Counties (Figure 1). A number of environmental conditions contributed to the severity of the fire including periods of sustained drought prior to the fire, high winds, high temperatures, and an abundance of invasive annual weeds. Average annual precipitation in this area is between 20 and 52 centimeters that come mostly as snow and monsoonal thunderstorms in late summer. The vegetation fueling the fire included salt desert scrub, sagebrush, and juniper intermixed with an understory of annual weeds. Most of the area within the fire boundary burned contiguously, leaving only a few small unburned islands. Approximately 75% of the land is Bureau of Land Management (BLM)-owned, followed by private (17%), state (8%) and Bureau of Indian Affairs (0.2%). Substantial investments from federal, state, and private sources were directed toward fire rehabilitation, comprised of revegetation treatments and post-fire monitoring. Soil stabilization was the immediate goal following the fire. The success of seed treatments and high seedling survival rates were of primary importance. Ultimately, the desired outcome was to increase the perennial plant cover and minimize invasive annuals and noxious weeds. Postfire vegetation and wildlife surveys were conducted by the BLM and the Utah Division of Wildlife Resources (UDWR). The Remote Sensing & GIS Laboratory (RS/GIS Lab) at Utah State University (USU) was contracted to develop a remote sensing-based monitoring protocol to assess the effects of the fire on a landscape scale and to evaluate seeding program success. PROJECT OBJECTIVES The primary objective of this project was to develop a remote sensing-based monitoring system to: 1) determine the extent of the Milford Flat Fire, 2) identify land cover types affected by the fire, 3) map pre- and post-fire distribution and density of cheatgrass (Bromus tectorum), 4) identify pre- and post-fire phenological patterns, including greenness onset, peak, and end dates, and 5) assess the effectiveness of post-fire seeding treatments. ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012 Figure 1. Location of the Milford Flat Fire in Utah. REMOTE SENSING DATA Data from the Moderate Resolution Imaging Spectroradiometer (MODIS), Landsat Thematic Mapper (TM), and the National Agriculture Imagery Program’s (NAIP) 1-meter resolution digital ortho photos, all provided critical information for this project. The Normalized Differenced Vegetation Index (NDVI) derived from MODIS 16-Day (cloud-corrected) composite images at 250-meter spatial resolution, was a primary data set in this project for measuring changes in vegetation growth and productivity from years 2000 to 2010. NDVI is a ratio based on two bands, the visible red (.620–.670 micrometers) and the NIR (.841–.876 micrometers) regions of the electromagnetic spectrum. The visible red region of incoming sunlight is strongly absorbed by chlorophyll in plants, whereas the NIR region is reflected. NDVI is calculated as follows: NDVI = (NIR – red)/ (NIR + red) The NDVI produces a range of values between -1 and 1, where, increasing positive values are associated with increasing greenness. By exploiting the differences in plant reflectance, the spatial distribution of vegetation growth patterns can be determined. The finer spatial resolution of Landsat TM (30-meter2) was used to detect and delineate the area that burned. As with MODIS, we created a Landsat-derived fire intensity index by exploiting the near-infrared and middle-infrared bands of a Landsat image acquired on July 27, 2007. Lastly, the NAIP county-wide digital ortho-photos were invaluable sources of data used in base maps and selection of field sites. ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012 Figure 2. Progression from the original Landsat TM image captured July 27, 2007, to the Landsat-derived fire intensity index (panchromatic), to the Landsat-derived classification of the fire area (red) with BLM fire boundary (black). METHODS Identifying Fire Extent Fire-related decreases in vegetation are evident spectrally by decreases in reflectivity in the visible and NIR bands, and conversely, increases in the MIR band reflectance values. Accordingly, vegetation loss by fire can be detected remotely if the amount of vegetation removed changes significantly the amount of solar radiation reflected or absorbed (Patterns and Yool, 1998, and references therein). Figure 2 shows the raw Landsat TM image acquired on July 27, 2007 (visualized by bands: 7 (MIR), 4 (NIR), and 3 (red)). The intensity of the red color coincides with the amount of black ash resulting from the fire. Darker red areas occur where juniper was prevalent creating more ash. Lighter red areas seem to coincide with areas that had lower ground cover. To identify burned areas immediately after the MFF, we created a Normalized Difference Fire Index (NDFI) using Landsat’s band 4 (NIR, 0.76-0.90 micrometers) and band 7 (MIR, 2.08-2.35 micrometers) in the following calculation: NDFI = (NIR – MIR) / (NIR + MIR). In the Landsat-derived fire intensity map (Figure 2, panchromatic image) brighter areas indicate higher relative fire intensity due to higher amounts of combustible material. The third image in Figure 2 shows the classification of the areas that burned with the BLM-generated fire boundary for comparison. The method of mapping fire extent using the NDFI provided a more detailed depiction of what lands actually burned within the fire boundary. However, for this project, we used the extent of the BLM-generated fire boundary to retain the unburned areas for our analyses. Pre-fire Land Cover The vast majority (910 km2) of the pre-fire land cover present within the MFF boundary, were shrublands dominated by Wyoming big sagebrush (Artemisia tridentata spp. wyomingensis), black sagebrush (A. nova), shadscale saltbush (Atriplex confertifolia), fourwing saltbush (A. canescens), and greasewood (Sarcobatus vermiculatus). Juniper woodlands made up the second highest land cover type prior to the fire (285 km2), followed by a mixture of grasslands and forblands (115 km2). ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012 Distribution and Density of Cheatgrass Assuming that peak greenness for cheatgrass occurs around April 15 th (+/- 2 to 3 weeks) in this portion of Utah (R. Beckstrand (BLM-FFO), pers. comm.), an elevated early spring green-up is a possible indication of cheatgrass cover. Depending on local climate conditions, late June to early July is when “greenness” values would be at a minimum; after the cool season grasses have senesced and before the green-up of the warm season grasses. An index for cheatgrass density swas derived by selecting two dates of MODIS NDVI images for each year between 2000 and 2010, one to identify peak greenness in spring, and the second for typical summer greenness. Using the formula: (Spring NDVI – Summer NDVI) / (Spring NDVI + Summer NDVI) a normalized difference index was created, producing a range of greenness values between -1 and +1. Values above zero indicate areas that were greener in spring than they were in summer. Higher positive values should indicate higher cheatgrass cover and density. We compared the output of cheatgrass density indices Table 1. Dates for estimating annual cheatgrass using the higher spatial resolution of Landsat-derived NDVI density using MODIS NDVI data images versus the coarser MODIS-derived NDVI images. We used MODIS NDVI data to calculate the probability of Year Spring Summer cheatgrass across the extent of the MFF for the following 2000 Apr-22 Jul-11 reasons: 1) more dates to select from, enabling more 2001 Apr-23 Jul-12 phenologically accurate representations of the difference 2002 Apr-23 Jun-26 between peaks and troughs in annual greenness curves, 2) 2003 Apr-23 Jul-12 MODIS 16-day composites are cloud free, and 3) spatially, the 2004 Apr-22 Jul-11 results were similar between Landsat and MODIS when the 2005 May-09 Jul-12 images were acquired on the same dates by the two sensors. A list of dates used to create annual cheatgrass density 2006 May-09 Jul-12 estimates from MODIS NDVI data is provided in Table 1. The 2007 Apr-07 Jun-26 peak of spring greenness during the years 2000 to 2010 2008 Apr-06 Jul-11 occurred between April 6 and May 9. In 2007 and 2008, the 2009 Apr-23 Jul-12 greenness peaked a little early during the first week of April. 2010 May-09 Jul-12 Also, in 2007, an early summer date (June 26th) was selected for minimum greenness, since the next available date, July 12 th, was while the fire was burning and would have biased the cheatgrass estimate. Phenological Profiles The MODIS NDVI images were also used to create annual “greenness curves” or phenological profiles for each year between 2000 and 2010. By plotting mean NDVI values over time, changes in vegetation can be seen at both fire-wide and treatment scales. A total of 253 images of MODIS NDVI 16-day composite data were downloaded (23 dates/year * 11 years) from the USGS Land Processes Distributed Active Archive Center (https://lpdaac.usgs.gov/). Each date required three scenes to encompass the state of Utah, which were mosaicked, reprojected, and subset for subsequent analysis at the scale of the Milford Flat Fire. We used phenological profiles to develop within- and among-year comparisons, including identification of greenness onset, peak, and end dates, and as an overall index of the response of vegetation to the treatments. Seed Treatments and Remote Sensing Monitoring Field Sites The BLM’s seeding program began in the fall of 2007 and continued through the spring of 2009. Four types of mechanical treatments were employed depending on the terrain and remnant vegetation. Areas with steeper slopes received aerial seeding only, while areas with more even terrain were seeded by chaining or drilling. A small proportion of the treatments involved imprinting in areas with remnant greasewood. Approximately 647 km2 were treated initially. However, some areas required subsequent re-seeding, which took place in 2008 and 2009. Approximately 1100 km2 were seeded in total. Emergency stabilization seed mixes for the MFF were composed primarily of cool season perennial bunchgrasses and forbs and some warm season perennial bunchgrasses (both native and non-native species). A shrub component was included in some areas during re-seeding. Shrubs included fourwing saltbush, shadscale saltbush, and Gardner’s saltbush (A. gardneri) in lower precipitation zones, and Wyoming big sagebrush, Mountain big sagebrush (A. tridentata spp. vaseyana), and antelope bitterbrush (Purshia tridentata) in higher precipitation ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012 zones. Forage kochia (Bassia prostrata) was commonly used in re-seeded areas. Seeding application occurred during the fall, winter, and spring. Consequently, many large areas that were once dominated by shrubland and juniper were converted to grasses and forbs following the initial soil stabilization efforts. Twenty-five field sites representing a range of pre-fire vegetation types and post-fire treatments were selected within the perimeter of the MFF to evaluate the ability of remotely-sensed imagery to monitor treatment success. Field site selection was based on the following principles: 1) site familiarity developed during field reconnaissance lead by the BLM prior to remote sensing analyses, 2) stratification across the three most prevalent treatment types, and identification of untreated and unburned areas, 3) easily accessed by roads in preparation for the second field reconnaissance, and 4) areas that could be correlated to BLM vegetation monitoring sites. The orientation of the field sites started in the southern portion of the fire within the Cedar City Field Office (CCFO) district which is generally within higher precipitation zones, and run northward into the Fillmore Field Office (FFO) district which is generally within a lower precipitation zone (Figure 3). Of the 25 field sites, 3 were aerially seeded, 12 were chained, 6 were drilled, 3 burned but with no treatment, and one was unburned. We intersected each MODIS NDVI 16-day composite image with the field site boundaries to calculate mean greenness values for all dates at each site. Similarly to the fire-wide analysis (see Phenological Profiles section), we plotted mean NDVI values over time to show differences in phenology (i.e. greenness) within and among years by treatment type. Figure 3. Distribution of BLM treatments and field sites selected for remote sensing monitoring. ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012 RESULTS Distribution and Density of Cheatgrass To identify potential hotspots for cheatgrass before and after the MFF, we calculated a normalized difference index by optimizing the differences between spring and summer NDVI greenness values. We created a series of maps to show the estimated distribution and abundance of cheatgrass from year 2000 to 2010 (Figure 4). The same histogram breaks were used each year to show relative cheatgrass density. Density levels were classified as the following: none (clear/transparent), low (green), medium (orange), and high (red). Figure 4. Milford Flat Fire relative cheatgrass density indices for 2000 through 2010. (Density classes: low = green, medium = orange, high = red, no cheatgrass predicted = transparent). There is a noticeable increase in both cheatgrass distribution and density in the years leading up to the fire, with a spike in cheatgrass density in 2001, 2005, and 2007. This part of Utah received an unusual amount of precipitation ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012 in the fall of 2000, 2004, and 2006. The map for 2007 is an estimate of cheatgrass immediately preceding the MFF, which clearly shows higher cheatgrass density across the MFF landscape. In the three years post-fire, 2008 to 2010, the distribution and relative amount of cheatgrass were significantly reduced compared to pre-fire densities. The first year following the fire, the landscape was mostly void of vegetation. In 2008, the density of cheatgrass was generally low with only a few areas of high density (red). There was an increase in density in 2009, but the actual value for greenness (i.e. NDVI means) in spring was lower than all other years (except 2008) and was slightly delayed (see Phenological Patterns section, Figure 5). The cheatgrass response in 2009 was higher than expected, being that the fall of 2008 was not particularly wet. That said, the density calculated for 2009 may be a response to other species in addition to cheatgrass growing at the same time. Plants such as forage kochia and alfalfa (Medicago sativa), both have early spring greenness similar to cheatgrass, but have active growing seasons throughout the summer and fall. Therefore, areas with a high component of forage kochia and/or alfalfa would lower the spring-to-summer difference. The climate in 2010 appeared abnormal, with lower than average temperatures in April, and an increase in spring precipitation (D. Whitaker (BLM-FFO), pers. comm.). Cheatgrass density decreased in 2010 compared to 2009 estimates, but had more high density areas than were present in 2008. Figure 4 shows the variability in the area estimated to have cheatgrass and changes in density levels from 2000 to 2010. The area with cheatgrass including all density classes (i.e. low, medium, and high levels) before the fire (2000 – 2007) averaged approximately 1175 km2 (SD ± 231), or 87% of the MFF burn area. The area with cheatgrass after the fire (2008 – 2010) dropped to an average of 811 km2 (SD ± 150), or 60% of the MFF burn area. The average area with just “high” levels of cheatgrass dropped from 35% to 11% of the MFF burn area, with 473 km2 (SD ± 310) before the fire and 145 km2 (SD ± 188) after the fire. Percent cover of cheatgrass within the different cheatgrass density levels was not available, as this information requires additional ground-truth data to calibrate. These maps represent relative values of cheatgrass identifying “hot spots” for each year throughout the MFF burned area. Phenological Patterns Greenness curves constructed from MODIS NDVI mean values for water years 2000 to 2010 were used to assess changes in phenology before and after the MFF (Figure 5). We used “water years,” defined as the 12-month period from October 1 to September 30. Values for each date from the first six water years, 2000 to 2006, were averaged together to provide a “benchmark” for comparing post-fire revegetation efforts. All subsequent years, 2007 through 2011, are shown as single lines for viewing purposes, but represent the same type of data as the 2000 to 2006 mean. Each of these response curves represent average “greenness” across the extent of the fire. The year of the fire, 2007, is shown as a black line. The timing and height of peak greenness during the spring of 2007, is a typical response for cheatgrass. The black vertical line represents the start date of the fire, after which, greenness drops immediately. The pattern in the vegetation phenologically, was variable during the first two years following the fire. However, at three years post-fire, the response in greenness is promising. In 2010 (Figure 5, red line), the onset of greenness is nearly one month after the 2000-2006 average onset. The gradual increase in greenness, and trend of increased greenness continuing later into the summer beyond the typical green-up and senescence of cheatgrass, indicates other species have emerged. The continued elevated greenness pattern is likely a response by warm season grasses, perennial forbs, or shrubs that were seeded during post-fire treatments, re-sprouted, or immigrated into the area following the MFF. Regardless of the species present, the prolonged greenness into summer months is a positive result. MODIS NDVI greenness curves were used to identify the approximate dates for greenness onset, peak, and end between 2000 and 2010 (Figure 6). In pre-fire years, greenness onset began in early to middle March (March 5 th – 22nd). The greenness peak occurred in late April and early May (April 22 nd - May 9th), with a decline in greenness values in middle to early July (June 11th – July 12th). The year of the MFF was an unusual year, in several respects. In the fall of 2006, there was a slight increase in greenness values relative to prior years, indicating a possible fall emergence of cheatgrass. In 2007, greenness peaked earlier than normal, approximately April 7 th, following an initial steep green-up, surpassing all years’ peak greenness values (Figure 5). The green-up period lasted through late June (June 26th), approximately 2 weeks prior to the fire’s starting date of July 6 th. In 2009 and 2010, the start date for greenness onset was approximately April 23 rd, which was related to later than normal snow cover. ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012 Figure 5. Greenness curves at Milford Flat for water years 2000 to 2011. Figure 6. Dates for greenness onset, peak, and end at Milford Flat, 2000 - 2010. If cheatgrass receives sufficient fall precipitation, and temperatures are agreeable, germination will begin that same season, leading to potentially greater cheatgrass density the following spring (R. Beckstrand (BLM-FFO) and E. Schupp, pers. comm.). The cheatgrass density maps for 2001, 2005 and 2007 (Figures 4) may be an indication of this tendency. By adding climate data to the NDVI greenness curves, this phenological pattern was observed: spring greenness values were higher following a wet fall, particularly if the precipitation came during the month of October (Figure 7). A similar pattern was apparent in 2001 and 2005 of relatively high spring greenness following an unusually wet October. October precipitation in 2001, 2005, and 2007, was at least 6.7 cm (≥1 SD) above the 2000 – 2011 mean. The relationship between mean October precipitation was positively correlated with peak spring greenness (R2 = 0.58, p = 0.006) (Figure 8, inset). We applied this relationship using the recorded mean October ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012 2010 precipitation of 6.4 cm to predict peak spring greenness for water year 2011. The predicted spring greenness for water year 2011 was 0.34, also above the 2000 to 2010 average (Figure 8). Figure 7. Monthly greenness (NDVI means) plotted against mean precipitation for the 2007 water year. Figure 8. October precipitation and peak spring greenness by water year at Milford Flat. ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012 Seed Treatment Assessment The final objective for this project was to evaluate the response of post-fire seeding treatments with remotelysensed imagery. Twenty-five sample field sites (Figure 3) were examined including: 3 aerial seedings, 12 chained and seeded sites, 6 drilled and seeded sites, 3 burned sites that did not receive a treatment, and 1 unburned site. Changes in mean NDVI values were used to assess the response of the treatments in the following analyses: 1. Phenological curves for water years 2000 to 2011 were created for each field site (similar to the Figure 5, not included in this document due to page length restrictions), using mean greenness values from all dates using the mean greenness value from water years 2000 to 2006 as a “benchmark” to compare the vegetation response in pre- versus post-fire years. 2. Mean greenness values in 2010 were used to compare post-fire differences in spring (May 25th) versus summer (July 28th) (Figure 9a) to assess cheatgrass potential three years post-fire. 3. Differences in mean greenness between pre-fire benchmarks (i.e. mean between 2000 and 2006) and postfire (2010) in spring (May 25th) and summer (July 28th) were calculated to see before and after greenness by season by field site (Figure 9b). Patterns in phenology at the field site scale are similar to the fire-wide results, but there are some notable differences based on pre-fire vegetation, treatment type, and location. When comparing year 2010 (Figure 5, red line) relative to the benchmark (Figure 5, gold bars), the average greenness at juniper sites dropped considerably after the fire compared to pre-fire values, particularly in the summer. As would be expected, live juniper had higher greenness values pre-fire compared to the vegetation signal three years post-fire. The opposite response was seen at many of the big sagebrush sites where greenness values were higher in the summer post-fire than the pre-fire summer benchmark, evidence of the establishment of perennial seedlings. At some sites, the cheatgrass population may be increasing based on a precipitous drop in spring versus summer greenness values. Figures 9a and b summarize seasonal changes in phenology at each field site, color-coded by treatment type. In general, chained sites had slightly higher greenness than aerial and drilled sites in both spring and summer, but the response among treatment types was similar (i.e. no significant differences among treatment types) (Figure 9a). In areas where juniper dominated pre-fire, the 2010 summer greenness was greater at chained sites versus those that were aerially seeded. There is greater growth in chained areas compared to aerially seeded sites (D. Fletcher BLMCCFO, pers. comm.). The 2010 summer greenness was higher in drilled sites relative to pre-fire benchmarks, except for two sites, #24 (salt desert scrub pre-MFF) and #25 (salt desert scrub/ greasewood flat pre-MFF). These two sites are probably the driest sites, the latter with highly saline soils that experienced extreme wind erosion after the fire. Of the three untreated sites, there were minimal differences in summer greenness in pre-fire benchmarks and 2010 values, with the exception of a small peak in greenness on August 13th, possibly a response of warm season grasses to the precipitation received in late July/early August. Both big sagebrush and James’ galleta (Pleuraphis jamesii) seedlings were re-establishing post-MFF at field site #7. Prior seed treatments put in place following the 2006 wildfires may be responsible for the increased greenness at field site #16. The amount of greenness in 2010 at the unburned site, site #3, a juniper site with sagebrush understory, was analogous to pre-fire benchmarks. Figure 9b shows the seasonal difference in greenness between pre-fire benchmarks (first bar) and post-fire values (second bar). Positive NDVI values (i.e. NDVI > 0) indicate increased greenness after the fire, while negative NDVI values indicate lower greenness after the fire. Field sites #1 to #9, #12, and #13 were all dominated by juniper prior to the MFF. Field sites #1 to #6 show negative greenness values in summer when comparing 2010 to the prefire average. For these sites, in summer, live juniper (i.e. the pre-fire condition) had higher greenness than the vegetation in 2010. Field sites #24 and #25 prior to the fire were classified as Inter-Mountain Basins Mixed Salt Desert Scrub and Inter-Mountain Basins Greasewood Flat, respectively. These sites have experienced substantial erosion post-fire where revegetation has been difficult, hence the negative greenness values in summer when comparing pre-fire to 2010 means. As described above, the MFF covered a broad area fueled by intense winds. The fire raged through canyons, crossing over ridges, and into multiple precipitation zones. The northern most portion of the fire passed through areas and vegetation types where fire is relatively infrequent and less resilient to disturbance. A landscape pattern caused by northeasterly winds is clearly evident in both NAIP and Landsat imagery. In some areas, seeds eroded along with the soil and/or were buried due to high winds (R. Beckstrand (BLM-FFO), pers. comm.). The environmental conditions of the lower, drier sites and those with heavy erosion problems have likely contributed to a lower success rate of seedling establishment and productivity and will ultimately be slower to recover. ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012 Figure 9. Post-fire NDVI greenness in spring (May 25th) and summer (July 28th) by treatment type (a). Differences in NDVI greenness between pre-fire benchmarks (means from 2000-2006) and post-fire (2010) in spring (May 25th) and summer (July 28th) by treatment type (b). (Treatment type: aerially seeded = blue, chained = lavender, drilled = yellow, no treatment = tan, unburned = green.) CONCLUSIONS Relative to the state of Utah, MFF covered ~0.7% of the total land area. The intensity of this fire removed most of the pre-existing vegetation and being a highly wind-prone area, the unprotected soils were therefore susceptible to ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012 severe wind erosion. The immediate concern of the BLM was emergency stabilization of the soil. The BLM’s seeding treatments started during the fall of 2007 and were completed by spring 2008. In 2009, the RS/GIS Laboratory began to work with the BLM to develop a remote sensing monitoring program to evaluate the success of post-fire landscape treatments. From a remote sensing perspective, the outcome of the BLM’s seeding program was positive. The MODISderived phenological profiles showed that although the vegetation responded slowly at first, the relative greenness detected in summer months increased compared to pre-fire benchmarks. This response was likely a result of perennial species (i.e. grasses, forbs, and some shrub seedlings) that were seeded, and/or certain immigrant species such as Russian thistle (Salsola iberica) that are occupying some areas. From annual cheatgrass density maps, the amount and distribution of cheatgrass was lower when compared to pre-fire estimates. At the field site scale, we found that areas that were chained (i.e. mostly juniper sites) had higher greenness in the summer compared to sites that were aerially seeded. Many of the sites that were drill seeded also had a strong greenness response in summer, often greater than the greenness exhibited by pre-existing vegetation (i.e. salt desert scrub and low elevation sagebrush sites). No matter how much greenness increased in each treatment, the greenness peak in all plots occurred nearly one month later in 2010. The delayed greenness response in 2010 was largely due to climate variation due to an extended winter and snow cover through April, but the gradual increase in spring values and continued elevated response through summer was likely a result of the treatments. While the trajectory for recovery is positive, as our data shows, the response was variable in certain areas, and our results are based mostly on the vegetation responses in a single year, 2010. Additional monitoring will be required to properly assess the success of seeding treatments across the MFF landscape. The 2011 water year will be interesting. The points of concern are mostly related to the favorable conditions for cheatgrass growth during the 2011 water year. The higher than average precipitation received in the fall of 2010, may promote early germination of cheatgrass. Greenness recorded in the fall of 2010 was also elevated compared to pre-fire benchmarks. Temperatures rose statewide in early spring of 2011, followed by additional precipitation. Fortunately, the response of the treatments appears to be good, evidenced by elevated summer greenness patterns in 2010 at the majority of the MFF field sites. REFERENCES Aronoff, S., 2005. Remote sensing basics, In: Remote sensing for GIS managers. S. Aronoff, editor. ESRI Press, Redlands, CA-USA., pp. 53–67. Patterns, M.W., and S.R. Yool, 1998. Mapping Fire-Induced Vegetation Mortality Using Landsat Thematic Mapper Data: A Comparison of Linear Transformation Techniques, Remote Sensing of the Environment, 65(2):132142. Piwowar, J.M., 2005. Digital image analysis, In: Remote sensing for GIS managers. S. Aronoff, editor. ESRI Press, Redlands, CA-USA., pp. 288-335. ASPRS 2012 Annual Conference Sacramento, California March 19-23, 2012
© Copyright 2026 Paperzz