With the growing size and use of night light time series from the Visible Infrared Imaging Radiometer Suite Day/Night Band (DNB), it is important to understand the stability of the dataset. All satellites observe differences in pixel values during repeat observations. In the case of night light data, these changes can be due to both environmental effects and changes in light emission. Here we examine the stability of individual locations of particular large scale light sources (e.g., airports and prisons) in the monthly composites of DNB data from April 2012 to September 2017. The radiances for individual pixels of most large light emitters are approximately normally distributed, with a standard deviation of typically 15–20% of the mean. Greenhouses and flares, however, are not stable sources. We observe geospatial autocorrelation in the monthly variations for nearby sites, while the correlation for sites separated by large distances is small. This suggests that local factors contribute most to the variation in the pixel radiances and furthermore that averaging radiances over large areas will reduce the total variation. A better understanding of the causes of temporal variation would improve the sensitivity of DNB to lighting changes.
Geographic information system (GIS)-based predictive modelling is widely used in archaeology to identify suitable zones for ancient settlement locations and determine underlying factors of their distribution. In this study, we developed predictive models on Roman viticulture in the Laetanian Region (Hispania Citerior-Tarraconensis), using the location of 82 ancient wine-pressing facilities or torcularia as response variables and 15 topographical and 6 socio-economic cost distance datasets as predictor variables. Several predictor variable subsets were selected either by expert knowledge of similar studies or by using a semi-automatization algorithm based on statistical distribution metrics of the input data. The latter aims at simplifying modelling and minimizing the necessity of a priori knowledge. Both approaches predicted the distribution of archeological sites sufficiently well. However, the best prediction performance was obtained by an expert knowledge model utilizing a predictor variable combination based on recommendations on viticulture by Lucius Junius Moderatus Columella, the prominent ancient Roman agronomist. The results indicate that the accessibility of a location and its connectivity to trade routes and distribution centres, determined by terrain steepness, was decisive for the settlement of viticultural facilities. With the knowledge gained, the ancient cultivated area and number of wine-pressing facilities needed for processing the vineyard yields were extrapolated for the entire study region.
26 April 2018 in International Journal of Remote Sensing
Remote sensing, which is a common method to examine land-use/land-cover (LULC) changes, could be useful in the analysis of livestock ecosystem transformations. In the last two decades, before Landsat images were free, developing countries could not afford monitoring through remote sensing because of the high cost of acquiring satellite imagery and commercial software. However, Landsat time series nowadays allows the characterization of changes in vegetation across large areas over time. The aim of this study is to analyse the LULC changes affecting forest frontiers and traditional silvopastoral systems (TSPS) in a representative livestock area of Nicaragua. Nearly cloud-free Landsat scenes – a Landsat 5 Thematic Mapper (TM) scene from 1986 and a Landsat 8 Operational Land Imager (OLI) scene from 2015 – have been the data sets used in the study. A process chain following a four-step definition of the remote-sensing process was conceptually developed and implemented based onfree open source software components and by applying the random forest (RF) algorithm. A conceptual LULC classification scheme representing TSPS was developed. Although the imagery shows a heterogeneous surface cover and mixed pixels, it is possible to achieve promising classification results with the RF algorithm with out-of-the-bag (OOB) errors below 13% for both images along with an overall accuracy level of 85.9% for the 2015 subset and 85.2% for the 1986 subset. The classification shows that from 1986 to 2015 (29 years) the intervened secondary forest (ISF) increased 2.6 times, whereas the degraded pastures decreased by 34.5%. The livestock landscape in Matiguás is in a state of constant transformation, but the main changes head towards the positive direction of tree-cover recovery and an increased number of areas of natural regeneration.
1 April 2017 in International Journal of Agricultural and Environmental Information Systems
From a geoinformation science perspective real estate portals apply non-spatial methods to analyse and visualise rental price data. Their approach shows considerable shortcomings. Portal operators neglect real estate agents' mantra that exactly three things are important in real estates: location, location and location (Stroisch, 2010). Although real estate portals retacord the spatial reference of their listed apartments, geocoded address data is used insufficiently for analyses and visualisation, and in many cases the data is just used to “pin” map the listings. To date geoinformation science, spatial statistics and geovisualization play a minor role for real estate portals in analysing and visualising their housing data. This contribution discusses the analytical and geovisual status quo of real estate portals and addresses the most serious deficits of the employed non-spatial methods. Alternative analysing approaches from geostatistics, machine learning and geovisualization demonstrate potentials to optimise real estate portals´ analysing and visualisation capacities.
From a geoinformation science perspective real estate portals apply non-spatial methods to analyse and visualise rental price data. Their approach shows considerable shortcomings. Portal operators neglect real estate agents' mantra that exactly three things are important in real estates: location, location and location (Stroisch, 2010). Although real estate portals retacord the spatial reference of their listed apartments, geocoded address data is used insufficiently for analyses and visualisation, and in many cases the data is just used to “pin” map the listings. To date geoinformation science, spatial statistics and geovisualization play a minor role for real estate portals in analysing and visualising their housing data. This contribution discusses the analytical and geovisual status quo of real estate portals and addresses the most serious deficits of the employed non-spatial methods. Alternative analysing approaches from geostatistics, machine learning and geovisualization demonstrate potentials to optimise real estate portals´ analysing and visualisation capacities.