Scale Of Spatial Autocorrelation Research Articles

Analysis of spatial pattern of aerosol optical depth and affecting factors using spatial autocorrelation and spatial autoregressive model

Previous studies neglected the implicit information in the spatial pattern of the aerosol optical depth (AOD), such as spatial configuration characteristics, spatial heterogeneity and spatial dependence. The understanding of AOD is helpful in accurately estimating the environmental and climatic effects from aerosols. In this study, a method of spatial autocorrelation statistics from geostatistics was applied to investigate the spatial pattern of AOD in Hubei province in central China during the period 2003–2008. Spatial autoregressive models were used to quantize the correlations between AOD and affecting factors, such as elevation, forest coverage and population density. In addition, the difference between the standard linear regression model and the spatial models was discussed. The results were as follows: the spatial pattern of AOD in Hubei province shows significant spatial autocorrelation, indicating that AODs are clustered such that higher AODs tend to be surrounded by higher AOD neighbors, while lower AODs are surrounded by lower AOD neighbors, and the spatial autocorrelation scale of the AOD over Hubei is approximately 400 km. The high–high zone is mainly distributed in the Wuhan city circle and the Jianghan plain areas, while the low–low zones are mainly located in the middle and high mountain areas of northwest Hubei province. The overall degree and pattern of spatial autocorrelation do not change largely from 2003 to 2008, which indicates a stable spatial configuration of AOD. A significant negative spatial autocorrelation exists between AOD and elevation, which may suggest that AOD and elevation have an inverse spatial distribution, and the same applies for forest coverage. Population density and the AOD show a significant positive spatial autocorrelation, which may imply that they have similar spatial distribution, while industrial production and AOD do not show an obvious positive spatial autocorrelation. Spatial autoregressive models show better predictive ability and stability than the standard linear regression model because the spatial autocorrelation of the AOD is taken into consideration.

Spatial variability of surface fuels in treated and untreated ponderosa pine forests of the southern Rocky Mountains

There is growing consensus that spatial variability in fuel loading at scales down to 0.5 m may govern fire behaviour and effects. However, there remains a lack of understanding of how fuels vary through space in wildland settings. This study quantifies surface fuel loading and its spatial variability in ponderosa pine sites before and after fuels treatment in the southern Rocky Mountains, USA. We found that spatial semivariance for 1- and 100-h fuels, litter and duff following thin-and-burn treatments differed from untreated sites, and was lower than thin-only sites for all fuel components except 1000-h fuels. Fuel component semivariance increased with mean fuel component loading. The scale of spatial autocorrelation for all fuel components and sites ranged from <1 to 48 m, with the shortest distances occurring for the finest fuel components (i.e. duff, litter). Component mean fuel particle diameter strongly predicted (R2 = 0.88) the distance needed to achieve sample independence. Additional work should test if these scaling relationships hold true across forested ecosystems, and could reveal fundamental processes controlling surface fuel variability. Incorporating knowledge of spatial variability into fuel sampling protocols will enhance assessment of wildlife habitat, and fire behaviour and effects modelling, over singular stand-level means.

Joint effects of habitat heterogeneity and species' life-history traits on population dynamics in spatially structured landscapes.

Both habitat heterogeneity and species’ life-history traits play important roles in driving population dynamics, yet there is little scientific consensus around the combined effect of these two factors on populations in complex landscapes. Using a spatially explicit agent-based model, we explored how interactions between habitat spatial structure (defined here as the scale of spatial autocorrelation in habitat quality) and species life-history strategies (defined here by species environmental tolerance and movement capacity) affect population dynamics in spatially heterogeneous landscapes. We compared the responses of four hypothetical species with different life-history traits to four landscape scenarios differing in the scale of spatial autocorrelation in habitat quality. The results showed that the population size of all hypothetical species exhibited a substantial increase as the scale of spatial autocorrelation in habitat quality increased, yet the pattern of population increase was shaped by species’ movement capacity. The increasing scale of spatial autocorrelation in habitat quality promoted the resource share of individuals, but had little effect on the mean mortality rate of individuals. Species’ movement capacity also determined the proportion of individuals in high-quality cells as well as the proportion of individuals experiencing competition in response to increased spatial autocorrelation in habitat quality. Positive correlations between the resource share of individuals and the proportion of individuals experiencing competition indicate that large-scale spatial autocorrelation in habitat quality may mask the density-dependent effect on populations through increasing the resource share of individuals, especially for species with low mobility. These findings suggest that low-mobility species may be more sensitive to habitat spatial heterogeneity in spatially structured landscapes. In addition, localized movement in combination with spatial autocorrelation may increase the population size, despite increased density effects.

Spatial Autocorrelation of Denitrification in a Restored and a Natural Floodplain

Floodplain wetlands have the potential to act as significant nitrogen (N) sinks through both assimilatory uptake and denitrification. However, the methodological constraint of making denitrification measurements at the point scale makes estimating field-scale rates difficult without a good understanding of how point measurements might be extrapolated. To address this issue, we compared spatial heterogeneity and scale of spatial autocorrelation of soil denitrification and its potential drivers (moisture, total N, nitrate and organic matter) in neighboring natural and restored floodplain systems of the Wisconsin and Baraboo Rivers in Wisconsin, USA at two spatial scales (large 81 m2 plot, and small 5 m2 plot). At the coarser scale, the range of autocorrelation was shorter at the restored site for 4 of the 5 soil parameters including denitrification (range = 9 m, vs. 21 m at the natural site). At the finer spatial scale, significant autocorrelation was detected for 4 of the 5 variables at the natural site, but only 1 variable (NO3-N) at the restored site. We suggest that high variability in spatial distribution of denitrification could render some interpolation techniques inappropriate.

Analyzing the spatial autocorrelation of regional urban datum land price

This study focuses on spatial autocorrelation and the spatial distribution of urban land prices from a regional perspective. Taking Hubei province, China, as a case study area, spatial autocorrelation degree, spatial autocorrelation pattern, and the mechanism of its formation were discussed. The study employs Moran’s I, local Moran’s I, and Moran’s I correlogram to analyze spatial autocorrelation degree and its change along with contiguity order. Some local clustering hot spots are found. This paper uses semi-variance statistic for land price based on route distance to find the spatial autocorrelation scale. We also adopt spatial clustering based on a kind of composite distance to probe into the clustering characteristic of land prices. By Moran’s I and Moran’s I correlogram, we find that datum price of the cities in Hubei province has faint spatial autocorrelation degree at the first and the second-order contiguity. Spatial variance hints that the scale of the autocorrelation is about 200 km in route distance. Spatial clustering result indicates that the spatial distribution of city land price is a kind of hierarchy structure similar to administrative regions. From principal factors analysis and stepwise linear regression, we find that the value added of city secondary and tertiary industry and the urban population are two of the most influential factors to urban datum land price. The value added of city secondary and tertiary industry has higher spatial autocorrelation than urban datum land price and has a bigger autocorrelation scale. But urban population has little spatial autocorrelation. It can be inferred that the spatial autocorrelation of urban land price is mainly caused by economic spatial autocorrelation. But its spatial autocorrelation degree is lower than economic factors because urban datum land price is also influenced by other special local factors, such as population, city infrastructure, land supply, etc.

Scales of temporal and spatial variability of midlatitude land surface temperature

[1] Scales of temporal and spatial variability of clear-sky land surface temperature (LST) in middle latitudes are empirically evaluated using data from satellite and land surface observations. We consider separately the time-dependent expected value, its spatial variations, weather-related temporal and spatial anomalies, and errors of LST observation. Seasonal and diurnal cycles in the time-dependent expected value of LST are found to be the main components of temporal variations of clear-sky LST. The scale of spatial variability in the expected value of LST is found to be much smaller than the scale of spatial variability of the weather-related signal. The scale of temporal autocorrelation of weather-related LST variations is found to be in a good agreement with our earlier preliminary estimate and equal to 3 d, which corresponds to the time scale of weather system variations. This weather-related signal in clear-sky LST is statistically the same as in surface air temperature (SAT) observations at regular meteorological stations. The scale of spatial autocorrelation of weather-related LST variations exceeds 1000 km, which is the spatial scale of synoptic weather systems. These estimates provide us with a basis for better understanding and interpretation of LST observations from past, current, and future geostationary satellites and polar orbiters.

Spatial and temporal ecology of Scots pine ectomycorrhizas

Spatial analysis was used to explore the distribution of individual species in an ectomycorrhizal (ECM) fungal community to address: whether mycorrhizas of individual ECM fungal species were patchily distributed, and at what scale; and what the causes of this patchiness might be. Ectomycorrhizas were extracted from spatially explicit samples of the surface organic horizons of a pine plantation. The number of mycorrhizas of each ECM fungal species was recorded using morphotyping combined with internal transcribed spacer (ITS) sequencing. Semivariograms, kriging and cluster analyses were used to determine both the extent and scale of spatial autocorrelation in species abundances, potential interactions between species, and change over time. The mycorrhizas of some, but not all, ECM fungal species were patchily distributed and the size of patches differed between species. The relative abundance of individual ECM fungal species and the position of patches of ectomycorrhizas changed between years. Spatial and temporal analysis revealed a dynamic ECM fungal community with many interspecific interactions taking place, despite the homogeneity of the host community. The spatial pattern of mycorrhizas was influenced by the underlying distribution of fine roots, but local root density was in turn influenced by the presence of specific fungal species.

The application of geostatistics in grain size trend analysis: A case study of eastern Beibu Gulf

There are 71 surface sediment samples collected from the eastern Beibu Gulf. The moment parameters (i.e. mean size, sorting and skewness) were obtained after applying grain size analysis. The geostatistical analysis was then applied to study the spatial autocorrelation for these parameters; while range, a parameter in the semivariogram that meters the scale of spatial autocorrelation, was estimated. The results indicated that the range for sorting coefficient was physically meaningful. The trend vectors calculated from grain size trend analysis model were consistent with the annual ocean circulation patterns and sediment transport rates according to previous studies. Therefore the range derived from the semivariogram of mean size can be used as the characteristic distance in the grain size trend analysis, which may remove the bias caused by the traditional way of basing on experiences or testing methods to get the characteristic distance. Hence the results from geostatistical analysis can also offer useful information for the determination of sediment sampling density in the future field work.

A diver survey method to quantify the clustering of sedentary invertebrates by the scale of spatial autocorrelation

Sedentary benthic invertebrates exhibit clustering at a range of spatial scales. Animal clustering reduces the precision of diver surveys and can accelerate overexploitation in dive fisheries. Dive harvesters target the densest aggregations of males and females that produce the highest rates of egg fertilisation during mass spawning events. By quantifying these effects of harvesting on fertilisation success, measuring animal clustering can inform stock management for reproductive sustainability. We present a method to measure the spatial extent of density aggregations down to 1 m, extending a previously described leaded-line survey design. Applying this method to abalone, research divers counted individuals in successive 1 × 2 m2 quadrats lying along adjoining pairs of 1 × 100 m2 transects. Clusters were observed as neighbouring quadrats of high animal density. Spatial autocorrelations at inter-quadrat distances of 1 to 100 m were calculated for four surveys, with eight pairs of transects swum in each survey. For all four surveys, inside two survey regions, spatial autocorrelation declined to non-significant levels at a distance of ~20 m. Quantified by the distance within which density counts are correlated, this quadrat-within-transect method provides a diver survey measure of the scale of spatial aggregation for sedentary invertebrates such as abalone, sea cucumbers and urchins.

Using spatial analysis to monitor tree diversity at a large scale: a case study in Northeast China Transect

Aims Monitoring and assessing diversity change at a large scale is important for any meaningful biodiversity conservation and management. Spatial analysis techniques can provide information about different aspects of diversity distribution including change. We applied some common spatial analysis methods and additive partitioning of species diversity in the Northeast China Transect as a case study to show how to characterize the distribution and change of tree diversity in this area from different perspectives. Methods The field data were collected from the permanent plots conducted every 4 km. The additive partitioning of species diversity was used to characterize the diversity of tree species at different scales. Moran’s I was used for identifying the spatial scale of autocorrelation, lacunarity was studied for diversity patch contagion and dispersion and spectral entropy was used for assessing the overall spatial distribution. Important findings Datacollectedfrom1986 to1994indicatethat the changeofa diversity was not significant in the study area, but the change of b diversity was significant. The percentage ofa diversity in total diversity (c) increased from 14.2 to 17.2%, and the percentage of b diversity decreased from 85.8 to 82.8%. For botha andb diversities, the scale of spatial autocorrelation decreased at the scale of 25–40 km and increased around 15–20 and 200 km. The lacunarity ofa diversity decreased significantly andtherewasasuddenchangeatthescaleof56–68km,butthelacunarity of b diversity increased across scales. The spectral entropy decreased slightly ina diversity and remained similar forb diversity. By using spatial analysis, we can monitor the diversity change over a large area and also assess the effectiveness of the current conservation strategies.

Effects of wind erosion on the spatial heterogeneity of soil nutrients in two desert grassland communities

Wind is known to affect the spatial heterogeneity of soil resources in arid and semiarid systems, but multi-year, quantified observations are largely absent. We studied the effects of wind erosion on the spatial distribution of soil organic carbon (SOC) and other soil nutrients at the Jornada Experimental Range, in southern New Mexico. Enhanced wind erosion was encouraged by grass cover reduction in a Sporobolus-mesquite dominated site (SM) and a Bouteloua-mesquite dominated site (BM). The scale and magnitude of spatial dependence for the soil analytes were quantified using geostatistical analyses. Results of this study show that soil organic matter related analytes such as SOC, TN, Navail, and SO4 2­ are among the first to be eroded and redistributed; cations such as Ca2+ and Mg2+ may not be removed and redistributed significantly; and other ions such as K+, Na+ and Cl− showed no discernible pattern of change. Geostatistics show that wind appeared to increase the scale of spatial autocorrelation, but decrease the scale of spatial dependence of most soil analytes over 2–3 windy seasons. In the wind enhanced plot of the SM site, up to 99% of the spatial dependence of SOC was autocorrelated at the distance of 1.45 m before the initiation of wind erosion, but the spatial dependence dropped significantly to only 60% at a larger autocorrelation distance of 2.76 m after three windy seasons. Similar but less significant changes were observed for SOC in the BM site. Despite the differential effects of wind on the soil analytes, we conclude that the overall results of wind on the grass cover reduction plots are the disappearance of small, strong fertile islands, which may be related to grasses; and the reinforcement of large fertile islands, which are likely related to mesquite shrubs. In addition, the change of the spatial patterns of SOC and other soil nutrients induced by enhanced wind erosion may persist and reinforce soil islands associated with shrubs, thus allowing a positive feedback for further desertification in this arid grassland.

Spatial dependency of soil nutrient availability and microbial properties in a mixed forest of Tsuga heterophylla and Pseudotsuga menziesii, in coastal British Columbia, Canada

Spatial variations in nutrient concentrations and turnover may contribute to variations in productivity within forest ecosystems and be responsible for creating and maintaining diversity of plant species. The aim of this study was to relate spatial patterns in soil nutrient availability and microbial properties in the forest floor and mineral soil in order to explore the controls on variations in nutrient availability in the two horizons. Microbial C, N and P, extractable N and P, and plant-available nutrients as estimated by plant root simulator probes were measured. We then used geostatistical techniques to determine the scale of spatial autocorrelation in the forest floor and mineral soil. Several of the measured variables were spatially autocorrelated at a scale of tens to hundreds of meters. Although more variables were autocorrelated in the mineral soil than in the forest floor, spatial patterns in gravimetric moisture content and nutrient concentrations in the two horizons generally overlapped. The spatial patterns were probably caused by differences in topography, except for ammonium whose shorter range of spatial autocorrelation may reflect variations in nitrogen content of the canopy.

Local Time- and Space Scales of Organized Tropical Deep Convection

The time- and space scales of tropical deep convection are estimated via analysis of 3-hourly Global Cloud Imagery (GCI) data for 3 yr at 35‐70-km resolution. The emphasis is on estimating local time- and space scales rather than traditional zonal wavenumber‐frequency spectra. This is accomplished through estimation of local spatial lag autocorrelations, the conditional probability of convection at neighboring points, and the expected duration of convective events. The spatial autocorrelation scale is found to be approximately 130 km, and the mean duration of convective events approximately 5.5 h, in the convectively active areas of the Tropics. There is a tendency for the spatial autocorrelation scales to be shorter over the continents than oceans (95‐155 versus 110‐170 km). The expected duration of convective events likewise tends to be shorter (4‐6 versus 5‐7 h). In the far western Pacific, these differences are sharp enough to legitimize the notion of the Indonesian archipelago as an extended maritime continent with a distinctive shape. Consistent with many other studies, the diurnal variation of the convection is also found to be strikingly different over the continents and oceans. The diurnal amplitude over land is comparable to the long-term mean, raising the possibility of significant aliasing across timescales. The simple analysis of this paper should be useful in evaluating and perhaps even improving the representation of convective processes in general circulation models.