Iryna Dronova

Object-Based Image Analysis in Wetland Research: A Review

The applications of object-based image analysis (OBIA) in remote sensing studies of wetlands have been growing over recent decades, addressing tasks from detection and delineation of wetland bodies to comprehensive analyses of within-wetland cover types and their change. Compared to pixel-based approaches, OBIA offers several important benefits to wetland analyses related to smoothing of the local noise, incorporating meaningful non-spectral features for class separation and accounting for landscape hierarchy of wetland ecosystem organization and structure. However, there has been little discussion on whether unique challenges of wetland environments can be uniformly addressed by OBIA across different types of data, spatial scales and research objectives, and to what extent technical and conceptual aspects of this framework may themselves present challenges in a complex wetland setting. This review presents a synthesis of 73 studies that applied OBIA to different types of remote sensing data, spatial scale and research objectives. It summarizes the progress and scope of OBIA uses in wetlands, key benefits of this approach, factors related to accuracy and uncertainty in its applications and the main research needs and directions to expand the OBIA capacity in the future wetland studies. Growing demands for higher-accuracy wetland characterization at both regional and local scales together with advances in very high resolution remote sensing and novel tasks in wetland restoration monitoring will likely continue active exploration of the OBIA potential in these diverse and complex environments.

Evaluation of a hierarchy of models reveals importance of substrate limitation for predicting carbon dioxide and methane exchange in restored wetlands

Wetlands and flooded peatlands can sequester large amounts of carbon (C) and have high greenhouse gas mitigation potential. There is growing interest in financing wetland restoration using C markets; however, this requires careful accounting of both CO2 and CH4 exchange at the ecosystem scale. Here we present a new model, the PEPRMT model (Peatland Ecosystem Photosynthesis Respiration and Methane Transport), which consists of a hierarchy of biogeochemical models designed to estimate CO2 and CH4 exchange in restored managed wetlands. Empirical models using temperature and/or photosynthesis to predict respiration and CH4 production were contrasted with a more process-based model that simulated substrate-limited respiration and CH4 production using multiple carbon pools. Models were parameterized by using a model-data fusion approach with multiple years of eddy covariance data collected in a recently restored wetland and a mature restored wetland. A third recently restored wetland site was used for model validation. During model validation, the process-based model explained 70% of the variance in net ecosystem exchange of CO2 (NEE) and 50% of the variance in CH4 exchange. Not accounting for high respiration following restoration led to empirical models overestimating annual NEE by 33–51%. By employing a model-data fusion approach we provide rigorous estimates of uncertainty in model predictions, accounting for uncertainty in data, model parameters, and model structure. The PEPRMT model is a valuable tool for understanding carbon cycling in restored wetlands and for application in carbon market-funded wetland restoration, thereby advancing opportunity to counteract the vast degradation of wetlands and flooded peatlands.

Modeling stormwater management at the city district level in response to changes in land use and low impact development

Abstract Mitigating the impact of increasing impervious surfaces on stormwater runoff by low impact development (LID) is currently being widely promoted at site and local scales. In turn, the series of distributed LID implementations may produce cumulative effects and benefit stormwater management at larger, regional scales. However, the potential of multiple LID implementations to mitigate the broad-scale impacts of urban stormwater is not yet fully understood, particularly among different design strategies to reduce directly connected impervious areas (DCIA). In this study, the hydrological responses of stormwater runoff characteristics to four different land use conversion scenarios at the city scale were explored using GIS-based Stormwater Management Model (SWMM). Model simulation results confirmed the effectiveness of LID controls; however, they also indicated that even with the most beneficial scenarios hydrological performance of developed areas was still not yet up to the pre-development level, especially where there were pronounced changes from pervious to impervious land.

Analysis of urbanization dynamics in mainland China using pixel-based night-time light trajectories from 1992 to 2013

ABSTRACT Understanding urbanization dynamics, or how intensity of urbanization changes over time, is an important basis for urban planning and management, which has been investigated using various data-driven approaches. Considering the advantages and constraints of different data sources, we use pixel-based, time-series night-time light (NTL) trajectories to characterize urbanization dynamics in mainland China where massive urban development has been occurring in recent decades. After pre-processing the data, we extracted time-series NTL trajectories for each 1 km × 1 km pixel between 1992 and 2013 and used the unsupervised k-means classification to identify the major typologies of these trajectories as urbanization dynamics based on their main statistical parameters. The classification identified five urbanization dynamics, namely, stable urban activity, high-level steady growth, acceleration, low-level steady growth, and fluctuation. Their distributions and spatial patterns were further summarized and compared among different Chinese administrative divisions. We specifically analysed the acceleration trajectories that showed rapid transitions from rural to urban, as we considered these trajectories as potential indicators for aggressive urbanization. We found several clusters at prefecture city and county levels with high proportion of the acceleration, and referred to the underlying socioeconomic characteristics and developmental history to understand how these clusters could had been formed. Through this study, we revealed the dominant tendencies of urbanization in China over space and time, and developed an analysis framework that could be extended to other regions.

Landscape-Level Associations of Wintering Waterbird Diversity and Abundance from Remotely Sensed Wetland Characteristics of Poyang Lake

Poyang Lake, the largest freshwater wetland in China, provides critical habitat for wintering waterbirds from the East Asian Flyway; however, landscape drivers of non-uniform bird diversity and abundance are not yet well understood. Using a winter 2006 waterbird survey, we examined the relationships among metrics of bird community diversity and abundance and landscape characteristics of 51 wetland sub-lakes derived by an object-based classification of Landsat satellite data. Relative importance of predictors and their sets was assessed using information-theoretic model selection and the Akaike Information Criterion. Ordinary least squares regression models were diagnosed and corrected for spatial autocorrelation using spatial autoregressive lag and error models. The strongest and most consistent landscape predictors included Normalized Difference Vegetation Index for mudflat (negative effect) and emergent grassland (positive effect), total sub-lake area (positive effect), and proportion of submerged vegetation (negative effect). Significant spatial autocorrelation in linear regression was associated with local clustering of response and predictor variables, and should be further explored for selection of wetland sampling units and management of protected areas. Overall, results corroborate the utility of remote sensing to elucidate potential indicators of waterbird diversity that complement logistically challenging ground observations and offer new hypotheses on factors underlying community distributions.

Implications of changing spatial dynamics of irrigated pasture, California's third largest agricultural water use

Irrigated agriculture is practiced on 680 million acres worldwide. Irrigated grazing land is likely a significant portion of that area but estimating an accurate figure has remained problematic. Due to its significant contribution to agricultural water use worldwide, we develop a methodology to remotely sense irrigated pasture using a California case study. Irrigated pasture is the third largest agricultural water use in California, yet its economic returns are low. As pressures mount for the agricultural sector to be more water efficient and for water to be directed towards its most economically valuable uses, there will likely be a reduction in irrigated pasture acreage. A first step in understanding the importance of irrigated pasture in California is establishing a methodology to quantify baseline information about its area, location, and current rate of loss. This study used a novel object-based image analysis and supervised classification on publicly-available, high resolution, remote sensing National Agriculture Imaging Program (NAIP) imagery to develop a highly accurate map of irrigated pasture in a rural county in Californias Sierra foothills. Irrigated pasture was found to have decreased by 19% during the ten-year period, 2005-2014, from 4,273 to 3,470 acres. The implications of this loss include potential impacts to wetland-dependent species, groundwater recharge, game species, traditional ranching culture, livestock production, and land conservation. Overall accuracy in classification across years was consistently over 89%. Comparing these results against available measurements of irrigated pasture provided by state and federal agencies reveals that this method significantly improves upon existing metrics and methods of data collection and points to critical needs for new targeted research and monitoring efforts. Broadly, the analysis presented here provides an improved methodology for mapping irrigated pasture that can be extended to provide accurate and spatially-explicit data for other counties in California and other arid and semi-arid regions worldwide.

Opportunities and Constraints in Characterizing Landscape Distribution of an Invasive Grass from Very High Resolution Multi-Spectral Imagery

Understanding spatial distributions of invasive plant species at early infestation stages is critical for assessing the dynamics and underlying factors of invasions. Recent progress in very high resolution remote sensing is facilitating this task by providing high spatial detail over whole-site extents that are prohibitive to comprehensive ground surveys. This study assessed the opportunities and constraints to characterize landscape distribution of the invasive grass medusahead (Elymus caput-medusae) in a ∼36.8 ha grassland in California, United States from 0.15m-resolution visible/near-infrared aerial imagery at the stage of late spring phenological contrast with dominant grasses. We compared several object-based unsupervised, single-run supervised and hierarchical approaches to classify medusahead using spectral, textural, and contextual variables. Fuzzy accuracy assessment indicated that 44–100% of test medusahead samples were matched by its classified extents from different methods, while 63–83% of test samples classified as medusahead had this class as an acceptable candidate. Main sources of error included spectral similarity between medusahead and other green species and mixing of medusahead with other vegetation at variable densities. Adding texture attributes to spectral variables increased the accuracy of most classification methods, corroborating the informative value of local patterns under limited spectral data. The highest accuracy across different metrics was shown by the supervised single-run support vector machine with seven vegetation classes and Bayesian algorithms with three vegetation classes; however, their medusahead allocations showed some “spillover” effects due to misclassifications with other green vegetation. This issue was addressed by more complex hierarchical approaches, though their final accuracy did not exceed the best single-run methods. However, the comparison of classified medusahead extents with field segments of its patches overlapping with survey transects indicated that most methods tended to miss and/or over-estimate the length of the smallest patches and under-estimate the largest ones due to classification errors. Overall, the study outcomes support the potential of cost-effective, very high-resolution sensing for the site-scale detection of infestation hotspots that can be customized to plant phenological schedules. However, more accurate medusahead patch delineation in mixed-cover grasslands would benefit from testing hyperspectral data and using our study’s framework to inform and constrain the candidate vegetation classes in heterogeneous locations.

Vegetation Development in a Tidal Marsh Restoration Project during a Historic Drought: A Remote Sensing Approach

Tidal wetland restoration efforts can be challenging to monitor in the field due to unstable local conditions and poor site access. However, understanding how restored systems evolve over time is essential for future management of their ecological benefits, many of which are related to vegetation dynamics. Physical attributes such as elevation and distance to channel play crucial roles in governing vegetation expansion in developing tidal wetlands. However, in Mediterranean ecosystems, drought years, wet years and their resulting influence on salinity levels may also play a crucial role in determining the trajectory of restoration projects, but the influence of weather variability on restoration outcomes is not well understood. Here, we use object-based image analysis (OBIA) and change analysis of the high-resolution IKONOS and WorldView-2 satellite imagery to explore whether mean annual rates of change from mudflat to vegetation are lower during drought years with higher salinity (2011-2015) compared to years with lower salinity (2009-2011) at a developing restoration site in California’s San Francisco Bay. We found that vegetation increased at a mean rate of 1979 m2/year during California’s historic drought, 10.4 times slower than the rate of 20580 m2/year between 2009 and 2011 when the state was not in drought. Vegetation was significantly concentrated in areas in closer to channel edges, where salinity stress is ameliorated, and the magnitude of the effect increased in the 2015 image. In our image analysis, we found that different distributions of water, mud and algae between years led to different segmentation settings for each set of images, highlighting the need for more robust and reproducible OBIA strategies in complex wetlands. Our results demonstrate that adaptive monitoring efforts in variable climates should take into account the influence of weather on tidal wetland ecosystems, and that high-resolution remote sensing can be an effective means of assessing these dynamics.

Aquatic plant functional type spectral characteristics analysis and comparison using multi-temporal and multi-sensor remote sensing over the Poyang Lake wetland, China

In systems with strong seasonal difference in vegetation structure and appearance, multi-temporal imagery can be particularly useful for community- and species-level discrimination. And, since the availability of past data for one source of time series images may be limited, so we need to develop multi-temporal and multi-source method for wetland ecosystem monitoring. To perform this type of analysis, the image spectral characteristics comparison between different aquatic macrophytes and different sensors should be studied firstly. We used TM images, Beijing-1 images and HJ-1 images for this analysis and based on the determination of aquatic plant functional types (PFTs). The objectives of this study were: (1) single-sensor single-date aquatic PFT analysis; (2) multi-source single-date diagnostic spectral characteristics analysis and comparison for different aquatic PFTs; (3) multi-source multitemporal diagnostic spectral characteristics analysis for different aquatic PFTs. From this analysis we found that: (1) For the single-date TM data, the diagnostic spectral band and indexes are Band 2, 4, 5, NDVI, and MNDWI; the best temporal for discriminating different Nonpersistent Emergent Wetland PFTs are in low water level periods, and water infilling and subsiding periods for seasonal submerged and floating aquatic macrophyte. Multi-spectral Decision Tree classification method lead the more good results for most of PFTs; (2) the same type of aquatic PFTs have similar and comparable reflectance characteristics between multi-sensor optical data which could satisfy the time series analysis by compensating more available past images; (3) phenological curves and relative canopy moisture curves extracted from time series remote sensing images provide important information for distinguish different PFTs.