Shirong Liu

The influence of vegetation type on the hydrological process at the landscape scale

The relationship between vegetation and hydrological processes is still a critical issue in ecology and environment science, especially at the landscape scale. Mingjiang valley plays an important role in water and soil resources conservation and erosion control in the upper Yangtze River. In this paper, the influence of vegetation type on hydrological processes at the landscape scale was studied using remote sensing and spatial analysis in Mingjiang valley and its five catchments. First, the vegetation distribution was mapped with high accuracy using three scenes of Landsat thematic mapper (TM) imagery and the optimal iterative unsupervised classification method. Then the spatial precipitation and actual evapotranspiration (AET) database was developed by converting the point-based data of meteorological stations to spatial surface with spatial interpolation. Cross-tabulation spatial analysis was employed to study the relationship between vegetation and rainfall, evaporation, and runoff. The results show that dominant vegetation types are grasslands, forests, and shrublands in the Mingjiang valley, with the proportions of 37.44%, 29.97%, and 22.62%, respectively. The annual precipitation ranges from 560 to 720 mm in areas of conifer and mixed forests, shrublands, and grasslands. For broadleaf forests, croplands, and other vegetation types, the precipitation distribution ranges from 480 to 800 mm, indicating a broader variation than that for the dominant vegetation type. In high-precipitation regions of the valley, forest vegetation covers the largest area. The precipitation is positively correlated with vegetation cover. We found that AET has a nonlinear relationship with vegetation cover, but this relationship is complicated. Our results demonstrated that the relative evapotranspiration rate (ER) is negatively correlated with precipitation, and water remaining (WR) is positively correlated with precipitation in the landscape. From the hydrological records in the Mingjiang valley, the annual mean runoff is 502 m3·s–1, the mean annual runoff amount is 140 × 109 m3, and the annual runoff rate is 0.0213 m3·s–1·km–2. We found that percent forest cover is positively correlated with percent runoff. This supports the results of previous nonspatial investigation in the valley. From scale analysis, we found that most spatial patterns of climate and hydrological variations are scale dependent, e.g., precipitation, AET, ER, WR, and runoff vary at different levels of landscape scales.

Monitoring the change of urban wetland using high spatial resolution remote sensing data

Accurate and timely information describing wetland resources and their changes over time, especially in coastal urban areas, is becoming more important. In this study, we mapped and monitored land-cover change in an urban wetland using high spatial resolution IKONOS images acquired in June 2003 and January 2006. An optimal iterative unsupervised classification (OIUC) method was used to overcome the limitations of unsupervised classification. The images were categorized into six classes, and an accuracy assessment was conducted using error matrices and the Kappa coefficient. The overall accuracies were 83.2% and 86.3% for the 2003 and 2006 images, respectively. A post-classification comparison method was used to detect the wetland change by calculating a detailed land-cover type transformation matrix. The results indicated a decrease in the area of water bodies and an increase in the area of vegetation in the wetland. This paper shows that high spatial resolution remote sensing data is advanced in studying an urban wetland at a local scale. An OIUC method, combined with visual interpretation, could yield high classification accuracy. A post-classification comparison method is also efficient in wetland change detection.

Estimation of forest-ecosystem site index using remote-sensed data

Abstract The estimation of site index and site quality forms the fundamental theory and basic tools in forest-ecosystem management and silviculture practice. The study on the spatial pattern and temporal dynamics of site index and site quality of forest ecosystem still lacks technological advancement. It is a novel approach for estimating forest productivity in large areas using satellite remote-sensed data. The site-index spatial distribution pattern of spruce (Picea asperata) forest in Songpan-Zhengjiangguan watershed, northwestern Sichuan Province, China, was described using the remote-sensing vegetation index application and the established inverse models. The application potential of the methodology in broad regions and forests using the accuracy assessment was evaluated. The results show that the site index of the spruce forest is in linear correlation with the remote-sensed vegetation indices (normalized difference vegetation index (NDVI) and soil adjust NDVI (TNDVI)), as well as with these inverse models with high accuracy. This study demonstrated that this approach can be used in similar estimation of different forest ecosystems.

Seasonal Timing for Estimating Carbon Mitigation in Revegetation of Abandoned Agricultural Land with High Spatial Resolution Remote Sensing

Dryland salinity is a major land management issue globally, and results in the abandonment of farmland. Revegetation with halophytic shrub species such as Atriplex nummularia for carbon mitigation may be a viable option but to generate carbon credits ongoing monitoring and verification is required. This study investigated the utility of high-resolution airborne images (Digital Multi Spectral Imagery (DMSI)) obtained in two seasons to estimate carbon stocks at the plant- and stand-scale. Pixel-scale vegetation indices, sub-pixel fractional green vegetation cover for individual plants, and estimates of the fractional coverage of the grazing plants within entire plots, were extracted from the high-resolution images. Carbon stocks were correlated with both canopy coverage (R2: 0.76–0.89) and spectral-based vegetation indices (R2: 0.77–0.89) with or without the use of the near-infrared spectral band. Indices derived from the dry season image showed a stronger correlation with field measurements of carbon than those derived from the green season image. These results show that in semi-arid environments it is better to estimate saltbush biomass with remote sensing data in the dry season to exclude the effect of pasture, even without the refinement provided by a vegetation classification. The approach of using canopy cover to refine estimates of carbon yield has broader application in shrublands and woodlands.

Vegetation dynamics and rainfall sensitivity for different vegetation types of the Australian continent in the dry period 2002–2010

Over the past 100 years, Australia has experienced pronounced changes in rainfall patterns, which in south-western Australia (SWAU) has resulted in both a >50% decrease of runoff since 1975 and effects on vegetation health. Resolving the dynamics of vegetation and climate is a prerequisite for predicting the response of vegetation to future climates and carbon mitigation objectives. With multi-resource vegetation indices (VIs; normalized difference vegetation index and leaf area index) and gridded climate data, this paper examines vegetation dynamics and sensitivity to rainfall change on the Australian continent for the past long drought period (2002–2010). We found that rainfall and VIs declined across 90% and 80% of the whole continent, respectively, compared to the baseline period of 2000–2001. The most dramatic declines in VIs occurred in open shrublands near the center of Australia and in SWAU, coinciding with significant reductions in rainfall and soil moisture. Overall, a strong relationship between water (rainfall and soil moisture) and VIs was detected in places where rainfall declined dramatically (up to 5 mm/year since 1970). For five major vegetation types, croplands showed the highest sensitivity to water change, followed by grasslands and woody savanna. Moderate sensitivity of open shrublands to water change was found, while evergreen broadleaf forests only showed a slight sensitivity to soil moisture change. Although there was no consistent significant relationship between rainfall and VIs of evergreen broadleaf forests, forests in south-eastern Australia, where rainfall had declined since 1997, have become more sensitive to rainfall change than in SWAU. Our results provide evidence that, at this scale of assessment and with interpolated data sets, a lasting reduced rainfall pattern has been a key factor constraining vegetation growth over the Australian continent.

Parallelization of a distributed ecohydrological model

Abstract WaSSI-C is an ecohydrological model which couples water and carbon cycles with water use efficiency (WUE) derived from global eddy flux observations. However, a significant limitation of the WaSSI-C model is that it only runs serially. High resolution simulations at a large scale are therefore computationally expensive and cause a run-time memory burden. Using distributed (MPI) and shared (OpenMP) memory parallelism techniques, we revised the original model as dWaSSI-C. We showed that using MPI was effective in reducing the computational run-time and memory use. Two experiments were carried out to simulate water and carbon fluxes over the Australian continent to test the sensitivity of the parallelized model to input data-sets of different spatial resolutions, as well as to WUE parameters for different vegetation types. These simulations were completed within minutes using dWaSSI-C, whereas they would not have been possible with the serial version. The dWaSSI-C model was able to simulate the seasonal dynamics of gross ecosystem productivity (GEP) reasonably well when compared to observations at four eddy flux sites. Sensitivity analysis showed that simulated GEP was more sensitive to WUE during the summer compared to winter in Australia, and woody savannas and grasslands showed higher sensitivity than evergreen broadleaf forests and shrublands. Although our results are model-specific, the parallelization approach can be adopted in other similar ecosystem models for large scale applications.