Milan Kilibarda

SoilGrids250m: Global gridded soil information based on machine learning

This paper describes the technical development and accuracy assessment of the most recent and improved version of the SoilGrids system at 250m resolution (June 2016 update). SoilGrids provides global predictions for standard numeric soil properties (organic carbon, bulk density, Cation Exchange Capacity (CEC), pH, soil texture fractions and coarse fragments) at seven standard depths (0, 5, 15, 30, 60, 100 and 200 cm), in addition to predictions of depth to bedrock and distribution of soil classes based on the World Reference Base (WRB) and USDA classification systems (ca. 280 raster layers in total). Predictions were based on ca. 150,000 soil profiles used for training and a stack of 158 remote sensing-based soil covariates (primarily derived from MODIS land products, SRTM DEM derivatives, climatic images and global landform and lithology maps), which were used to fit an ensemble of machine learning methods—random forest and gradient boosting and/or multinomial logistic regression—as implemented in the R packages ranger, xgboost, nnet and caret. The results of 10–fold cross-validation show that the ensemble models explain between 56% (coarse fragments) and 83% (pH) of variation with an overall average of 61%. Improvements in the relative accuracy considering the amount of variation explained, in comparison to the previous version of SoilGrids at 1 km spatial resolution, range from 60 to 230%. Improvements can be attributed to: (1) the use of machine learning instead of linear regression, (2) to considerable investments in preparing finer resolution covariate layers and (3) to insertion of additional soil profiles. Further development of SoilGrids could include refinement of methods to incorporate input uncertainties and derivation of posterior probability distributions (per pixel), and further automation of spatial modeling so that soil maps can be generated for potentially hundreds of soil variables. Another area of future research is the development of methods for multiscale merging of SoilGrids predictions with local and/or national gridded soil products (e.g. up to 50 m spatial resolution) so that increasingly more accurate, complete and consistent global soil information can be produced. SoilGrids are available under the Open Data Base License.

Spatio-temporal interpolation of daily temperatures for global land areas at 1 km resolution

Combined Global Surface Summary of Day and European Climate Assessment and Dataset daily meteorological data sets (around 9000 stations) were used to build spatio-temporal geostatistical models and predict daily air temperature at ground resolution of 1km for the global land mass. Predictions in space and time were made for the mean, maximum, and minimum temperatures using spatio-temporal regression-kriging with a time series of Moderate Resolution Imaging Spectroradiometer (MODIS) 8 day images, topographic layers (digital elevation model and topographic wetness index), and a geometric temperature trend as covariates. The accuracy of predicting daily temperatures was assessed using leave-one-out cross validation. To account for geographical point clustering of station data and get a more representative cross-validation accuracy, predicted values were aggregated to blocks of land of size 500x500km. Results show that the average accuracy for predicting mean, maximum, and minimum daily temperatures is root-mean-square error (RMSE) =2 degrees C for areas densely covered with stations and between 2 degrees C and 4 degrees C for areas with lower station density. The lowest prediction accuracy was observed at high altitudes (>1000m) and in Antarctica with an RMSE around 6 degrees C. The model and predictions were built for the year 2011 only, but the same methodology could be extended for the whole range of the MODIS land surface temperature images (2001 to today), i.e., to produce global archives of daily temperatures (a next-generation repository) and to feed various global environmental models. Key Points Global spatio-temporal regression-kriging daily temperature interpolation Fitting of global spatio-temporal models for the mean, maximum, and minimum temperatures Time series of MODIS 8 day images as explanatory variables in regression part

Spatial analysis of the temperature trends in Serbia during the period 1961–2010

The spatial analysis of annual and seasonal temperature trends in Serbia during the period 1961–2010 was carried out using mean monthly data from 64 meteorological stations. Change year detection was achieved using cumulative sum charts. The magnitude of trends was derived from the slopes of linear trends using the least square method. The same formalism of least square method was used to assess the statistical significance of the determined trends. Maps of temperature trends were generated by applying a spatial regression method to visualize the detected tendencies. The obtained results indicate a negative temperature trend for the period before the change year except for winter and a more pronounced positive trend after the change year. Besides being more pronounced, the vast majority of trends after the change year were also clearly statistically significant. Our estimate of the average temperature trend over Serbia is in agreement with those obtained at the global and European scale. Calculated global autocorrelation statistics (Moran’s I) indicate an apparent random spatial pattern of temperature trends across the Serbia for both periods before and after the change year.

Mapping population change index in Southern Serbia (1961-2027) as a function of environmental factors

Niche analysis methods developed within the biogeography community are routinely used for species distribution modeling of wildlife and endangered species. So far, such techniques have not been used to explain distribution of people in an area, nor to assess spatio-temporal dynamics of human populations. In this paper, the MaxEnt approach to species distribution modeling and publicly available gridded predictors were used to analyze the population dynamics in Southern Serbia (South Pomoravlje Region) for the period 1961-2027. Population values from the census administrative units were first downscaled to 200 m grid using a detailed map of populated places and dasymetric interpolation. In the second step, a point pattern representing the whole population (468,500 inhabitants in 2002) was simulated using the R package spatstat. MaxEnt was then used to derive habitat suitability index (HSI) as a function of gridded predictors: distance to roads, elevation, slope, topographic wetness index, enhanced vegetation index and land cover classes. HSI and environmental predictors were further used to explain spatial patterns in the population change index (PCI) through regression modeling. The results show that inhabiting preference for year 1961 is mainly a function of topography (TWI, elevation). The HSI for year 2027 shows that large portions of remote areas are becoming less preferred for inhabiting. The results of cross-validation in MaxEnt show that distribution of population is distinctly controlled by environmental factors (AUC > 0.84). Population decrease is particularly significant in areas >25 km distant from the main road network. The results of regression analysis show that 40% of variability in the PCI values can be explained with these environmental maps, distance to roads and urban areas being the main drivers of migration process. This approach allows precise mapping of demographic patterns that otherwise would not be visible from the census data alone.

Estimating the Performance of Random Forest versus Multiple Regression for Predicting Prices of the Apartments

The goal of this study is to analyse the predictive performance of the random forest machine learning technique in comparison to commonly used hedonic models based on multiple regression for the prediction of apartment prices. A data set that includes 7407 records of apartment transactions referring to real estate sales from 2008–2013 in the city of Ljubljana, the capital of Slovenia, was used in order to test and compare the predictive performances of both models. Apparent challenges faced during modelling included (1) the non-linear nature of the prediction assignment task; (2) input data being based on transactions occurring over a period of great price changes in Ljubljana whereby a 28% decline was noted in six consecutive testing years; and (3) the complex urban form of the case study area. Available explanatory variables, organised as a Geographic Information Systems (GIS) ready dataset, including the structural and age characteristics of the apartments as well as environmental and neighbourhood information were considered in the modelling procedure. All performance measures (R2 values, sales ratios, mean average percentage error (MAPE), coefficient of dispersion (COD)) revealed significantly better results for predictions obtained by the random forest method, which confirms the prospective of this machine learning technique on apartment price prediction.

Dasymetric Mapping of Population Distribution in Serbia Based on Soil Sealing Degrees Layer

This paper outlines a methodology used to disaggregate a census population in order to more accurately determine the population distribution over a regional area or a state scale. Data regarding population distributions are usually accessible at the level of individual census designation places and are usually mapped as aggregated polygons by the choropleth method with the assumption of a homogeneous distribution of population within a cartographic unit. In contrast, dasymetric mapping provides a more reliable view into the allocation of inhabitants, which can be of significant importance when estimating population distributions. Coupling this methodology with the GIS environment and a free open access database of soil sealing facilitates the acquisition of population surface models for human and urban geography applications.

Spatial Hedonic Modeling of Housing Prices Using Auxiliary Maps

The latest applications of hedonic dwelling price models have included recent advances in spatial analysis that control for spatial dependence and heterogeneity. The study of spatial aspects of hedonic modelling pertains to spatial econometrics, which is relevant to this study because it clearly accounts for the influence and peculiarities related by space in real estate price modeling analysis.

Sparse regression interaction models for spatial prediction of soil properties in 3D

Abstract An approach for using lasso (Least Absolute Shrinkage and Selection Operator) regression in creating sparse 3D models of soil properties for spatial prediction at multiple depths is presented. Modeling soil properties in 3D benefits from interactions of spatial predictors with soil depth and its polynomial expansion, which yields a large number of model variables (and corresponding model parameters). Lasso is able to perform variable selection, hence reducing the number of model parameters and making the model more easily interpretable. This also prevents overfitting, which makes the model more accurate. The presented approach was tested using four variable selection approaches – none, stepwise, lasso and hierarchical lasso, on four kinds of models – standard linear model, linear model with polynomial expansion of depth, linear model with interactions of covariates with depth and linear model with interactions of covariates with depth and its polynomial expansion. This framework was used to predict Soil Organic Carbon (SOC) in three contrasting study areas: Bor (Serbia), Edgeroi (Australia) and the Netherlands. Results show that lasso yields substantial improvements in accuracy over standard and stepwise regression — up to 50 % of total variance. It yields models which contain up to five times less nonzero parameters than the full models and that are usually more sparse than models obtained by stepwise regression, up to three times. Extension of the standard linear model by including interactions typically improves the accuracy of models produced by lasso, but is detrimental to standard and stepwise regression. Regarding computation time, it was demonstrated that lasso is several orders of magnitude more efficient than stepwise regression for models with tens or hundreds of variables (including interactions). Proper model evaluation is emphasized. Considering the fact that lasso requires meta-parameter tuning, standard cross-validation does not suffice for adequate model evaluation, hence a nested cross-validation was employed. The presented approach is implemented as publicly available sparsereg3D R package.

3D urban solar potential maps - case study of the i-SCOPE project

Solar maps as web cartographic products that provide information on solar potential of surfaces on the Earth have been exploited in decision making, awareness raising, and promoting the use of solar energy. Web based solar maps of cities have become popular services as the use of solar energy is especially attractive in urban environments. The article discusses the concept and aspects of urban solar potential maps on the example of the i-Scope project as a case study. The i-Scope roof solar potential service built on 3-D urban information models was piloted in eight European cities. To obtain precise data on solar irradiation, a good quality digital surface model is required. A cost efficient innovative method for generation of digital surface model from stereophotogrammetry for urban areas where no advanced source data (e. g. LiDAR) exist is developed. The method works for flat, shed and gable roofs and provides sufficient accuracy of digital surface model .