Keith T. Weber

A review of remote sensing of invasive weeds and example of the early detection of spotted knapweed (Centaurea maculosa) and babysbreath (Gypsophila paniculata) with a hyperspectral sensor

Abstract Remote sensing technology is a tool for detecting invasive species affecting forest, rangeland, and pasture environments. This article provides a review of the technology, and algorithms used to process remotely sensed data when detecting weeds and a working example of the detection of spotted knapweed and babysbreath with a hyperspectral sensor. Spotted knapweed and babysbreath frequently invade semiarid rangeland and irrigated pastures of the western United States. Ground surveys to identify the extent of invasive species infestations should be more efficient with the use of classified images from remotely sensed data because dispersal of an invasive plant may have occurred before the discovery or treatment of an infestation. Remote sensing data were classified to determine if infestations of spotted knapweed and babysbreath were detectable in Swan Valley near Idaho Falls, ID. Hyperspectral images at 2-m spatial resolution and 400- to 953-nm spectral resolution with 12-nm increments were used to identify locations of spotted knapweed and babysbreath. Images were classified using the spectral angle mapper (SAM) algorithm at 1, 2, 3, 4, 5, and 10° angles. Ground validation of the classified images established that 57% of known spotted knapweed infestations and 97% of known babysbreath infestations were identified through the use of hyperspectral imagery and the SAM algorithm. Nomenclature: Babysbreath, Gypsophila paniculata L. GYPPA; spotted knapweed, Centaurea maculosa Lam. CENMA.

Desertification and livestock grazing: The roles of sedentarization, mobility and rest

Pastoralism is an ancient form of self-provisioning that is still in wide use today throughout the world. While many pastoral regions are the focus of current desertification studies, the long history of sustainability evidenced by these cultures is of great interest. Numerous studies suggesting a general trend of desertification intimate degradation is a recent phenomenon principally attributable to changes in land tenure, management, and treatment. This paper explores the suggested causes of land degradation and identifies the land management and grazing treatments shared by many pastoral cultures. The singular commonality found in nearly all studies of degradation is the prevalence of partial or total rest. While historical observations rightly suggest that desertification is the result of both climatic and anthropic factors, recent focus has been placed upon the effect of sedentarisation. This paper attempts to coalesce these two streams of thinking with particular focus upon inclusive planning processes which may improve arid and semiarid rangeland ecosystems using livestock as a solution to the problem of land degradation.

Intercalibration and Evaluation of ResourceSat-1 and Landsat-5 NDVI

ResourceSat-1 is a designated alternative to Landsat should the existing TM (Thematic Mapper) and ETM+ (Enhanced Thematic Mapper Plus) sensors fail prior to the successful launch of Landsat 8 in late 2012. However, to enable integration of ResourceSat-1 into the many existing long-term Landsat projects around the world, practicable similarity must be demonstrated. To quantify the potential for ResourceSat-1 to satisfy some of the needs of the remote sensing community, Normalized Difference Vegetation Index (NDVI) values derived from Landsat-5 were compared to NDVI values derived from ResourceSat-1. An intercalibration equation that converts ResourceSat-1 NDVI values to equivalent Landsat-5 NDVI values was derived thereby enabling direct comparison between the two sensors. Comparisons were made using imagery spanning a 3-year time period. Prior to intercalibration, NDVI values were highly correlated (mean R 2 > 0.73) but statistically different (P < 0.001). Following intercalibration, the resulting indices were statistically inseparable (P ≥ 0.56). The intercalibration technique described in this paper represents an easily repeatable process that demonstrates practicable similarity between ResourceSat-1 and Landsat-5 imagery.

Comparison of MODIS fPAR Products with Landsat-5 TM-Derived fPAR over Semiarid Rangelands of Idaho

While validation of the MODIS fPAR product is well behind that of the LAI product, it is recently receiving more attention. In this study, MODIS fPAR and Landsat-5 TM-derived fPAR (TM fPAR) were calculated and quantitatively compared using imagery from 2005 to 2008 for the semiarid rangelands of Idaho, USA. fPAR change maps were calculated between active growth and late-summer senescence periods. Accuracy of the MODIS fPAR and TM fPAR were determined indirectly by incorporating field-based measurements of above-ground forage biomass and percent ground cover from a variety of sites (n = 442).

Modeling Wildland Fire Susceptibility Using Fuzzy Systems

Due to fire suppression efforts, many areas have developed conditions whereby fire susceptibility is high. To help identify those areas and improve fire management, two fire susceptibility models were developed for a study area in southeastern Idaho. Both models used the same intrinsic parameters (topography, fuel characteristics, etc). The difference between the models is the first used expert knowledge to weight input parameters, whereas the second relied upon fuzzy systems to derive the weighting. Comparing the resulting output models indicates that the first more accurately capture fire susceptibility. This lends credibility to the use of expert knowledge in geo-spatial modeling.

Multi-sensor Analyses of Vegetation Indices in a Semi-arid Environment

Multi-sensor comparisons are sometimes used due to limited image availability and temporal coverage by a single sensor. However, multi-sensor comparability is not well documented. Factors affecting direct comparability such as atmospheric conditions, landscape heterogeneity, landscape changes, and sensor characteristics are difficult to quantify. This study compared several vegetation indices (VIs) from multi-sensor data to determine if VIs are comparable across scales and sensors. Within-sensor comparisons demonstrate that VIs are consistent across spatial resolutions indicating a direct multi-scale comparability. However, among-sensor comparisons indicate that VIs calculated from different sensors are not comparable with one another regardless of spatial resolution. Sensor-specific characteristics appear to offer the best explanation for the observed results.

Assessing the impact of seasonal precipitation and temperature on vegetation in a grass-dominated rangeland

Changes in vegetation are affected by many climatic factors and have been successfully monitored through satellite remote sensing over the past 20 years. In this study, the Normalised Difference Vegetation Index (NDVI), derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard the Terra satellite, was selected as an indicator of change in vegetation. Monthly MODIS composite NDVI at a 1-km resolution was acquired throughout the 2004–09 growing seasons (i.e. April–September). Data describing daily precipitation and temperature, primary factors affecting vegetation growth in the semiarid rangelands of Idaho, were derived from the Surface Observation Gridding System and local weather station datasets. Inter-annual and seasonal fluctuations of precipitation and temperature were analysed and temporal relationships between monthly NDVI, precipitation and temperature were examined. Results indicated NDVI values observed in June and July were strongly correlated with accumulated precipitation (R2 >0.75), while NDVI values observed early in the growing season (May) as well as late in the growing season (August and September) were only moderately related with accumulated precipitation (R2 ≥0.45). The role of ambient temperature was also apparent, especially early in the growing season. Specifically, early growing-season temperatures appeared to significantly affect plant phenology and, consequently, correlations between NDVI and accumulated precipitation. It is concluded that precipitation during the growing season is a better predictor of NDVI than temperature but is interrelated with influences of temperature in parts of the growing season.

Assessing the susceptibility of semiarid rangelands to wildfires using Terra MODIS and Landsat Thematic Mapper data

In order to monitor wildfires at broad spatial scales and with frequent periodicity, satellite remote sensing techniques have been used in many studies. Rangeland susceptibility to wildfires closely relates to accumulated fuel load. The normalised difference vegetation index (NDVI) and fraction of photosynthetically active radiation (fPAR) are key variables used by many ecological models to estimate biomass and vegetation productivity. Subsequently, both NDVI and fPAR data have become an indirect means of deriving fuel load information. For these reasons, NDVI and fPAR, derived from the Moderate Resolution Imaging Spectroradiometer on-board Terra and Landsat Thematic Mapper imagery, were used to represent prefire vegetation changes in fuel load preceding the Millennial and Crystal Fires of 2000 and 2006 in the rangelands of south-east Idaho respectively. NDVI and fPAR change maps were calculated between active growth and late-summer senescence periods and compared with precipitation, temperature, forage biomass and percentage ground cover data. The results indicate that NDVI and fPAR value changes 2 years before the fire were greater than those 1 year before fire as an abundance of grasses existed 2 years before each wildfire based on field forage biomass sampling. NDVI and fPAR have direct implication for the assessment of prefire vegetation change. Therefore, rangeland susceptibility to wildfire may be estimated using NDVI and fPAR change analysis. Furthermore, fPAR change data may be included as an input source for early fire warning models, and may increase the accuracy and efficiency of fire and fuel load management in semiarid rangelands.

Improving Classification Accuracy Assessments with Statistical Bootstrap Resampling Techniques

The use of remotely sensed imagery to generate land cover models is common today. Validation of these models typically involves the use of an independent set of ground-truth data that are used to calculate an error matrix resulting in estimates of omission, commission, and overall error. However, each estimate of error contains a degree of uncertainty itself due to: (1) conceptual bias; (2) location/registration and co-registration errors; and (3) variability in the sample sites used to produce and validate the model. In this study, focus was not placed upon describing land cover mapping techniques, but rather the application of bootstrap resampling to improve the characterization of classification error, demonstrate a method to determine uncertainty from sample site variability, and calculate confidence limits using statistical bootstrap resampling of 500 sample sites acquired within a single Landsat 5 TM image. The sample sites represented one of five land cover categories (water, roads, lava, irrigated agriculture, and rangelands), with each category containing 100 samples. The sample set was then iteratively resampled (n = 200) and 65 sites were randomly selected (without replacement) for use as classification training sites, while the balance (n = 35) were used for validation. Imagery was subsequently classified using a maximum likelihood technique and the model validated using a standard error matrix. This classification-validation process was repeated 200 times. Confidence intervals were then calculated using the resulting omission and commission errors. Results from this experiment indicate that bootstrap resampling is an effective method to characterize classification uncertainty and determine the effect of sample bias.

Detection thresholds for rare, spectrally unique targets within semiarid rangelands.

Many factors influence classification accuracy, and this study assessed detection thresholds for various sub-pixel targets using QuickBird multispectral imagery. Six iterations of maximum-likelihood classification were used to determine classification accuracy for 100 spectrally unique targets randomly placed over a semiarid rangeland site. Error matrices were calculated using independent validation sites and producer’s, user’s, and overall accuracy, Kappa Index of Agreement, and transformed divergence were analyzed to compare the performance of each classification and determine detection thresholds. Results indicate a strong relationship between target size and classification accuracy (R 2 0.94) as well as an increasingly prominent role played by training site selection as target size decreased. Strong spectral separability and good classification accuracies were achieved for targets 25 percent cover. Sub-pixel targets 25 percent in size were not detectable. This study highlights the effect of target size upon classification accuracy and has direct implications for invasive plant research and rare target detection.