Abstract: The waters of the Mirna River basin have been extensively used for supply of a large po... more Abstract: The waters of the Mirna River basin have been extensively used for supply of a large portion of the population, tourists, industry and agricultural irrigation in Istria. An enlargement of water reserve for current uses, as well as for construction of several ...
Abstract: The purpose of this study was to analyze preliminary results and to determine the frequ... more Abstract: The purpose of this study was to analyze preliminary results and to determine the frequency and types of habitats used by the wolf marked with GPS collar. The wolf was captured and marked in Jelovac above village Krasno in Lika region. In the first 18 days ...
iSDAsoil dataset soil extractable Aluminium log-transformed predicted at 30 m resolution for 0–20... more iSDAsoil dataset soil extractable Aluminium log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, LandPKS, and other national and regional soil datasets). Layer description: sol_db_od_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Aluminium mean value, sol_db_od_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Aluminium model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.al_mehlich3 R-square: 0.881 Fitted values sd: 0.872 RMSE: 0.321 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -5.7042 -0.1036 0.0059 0.1189 3.3777 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) -0.675492 2.771906 -0.244 0.807 regr.ranger 0.879567 0.005464 160.969 &lt;2e-16 *** regr.xgboost 0.071537 0.005813 12.306 &lt;2e-16 *** regr.cubist 0.150157 0.004553 32.979 &lt;2e-16 *** regr.nnet 0.087603 0.431261 0.203 0.839 regr.cvglmnet -0.084440 0.003182 -26.534 &lt;2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.3208 on 63551 degrees of freedom Multiple R-squared: 0.8808, Adjusted R-squared: 0.8808 F-statistic: 9.391e+04 on 5 and 63551 DF, p-value: &lt; 2.2e-16</code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request support please vis [...]
Digital Terrain Model for Continental Europe based on the three publicly available Digital Surfac... more Digital Terrain Model for Continental Europe based on the three publicly available Digital Surface Models and predicted using an Ensemble Machine Learning (EML). EML was trainined using GEDI level 2B points (column "elev.lowestmode"): about 7 million GEDI points were overlaid vs NASADEM, AW3D, EU DEM, canopy height, tree cover and surface water cover maps, then a an ensemble prediction model was fitted using random forest, GLM with Lasso, Cubist and GLMnet, and used to predict most probable terrain height (bare earth). Input layers used to train the EML include: "lcv_bare.earth_glcf.landsat": UMD GLAD bare earth estimate for year 2010 based on Landsat time series, "dtm_elev.dsm_alos.aw3d": Digital Surface Model based on ALOS AW3D v8104, "dtm_canopy.height_glad.umd": UMD GLAD canopy height for 2019 based on GEDI data, "dtm_elev.dsm_eudem.eea": Copernicus EU DEM based on the SRTM and ASTER DEMs, "hyd_surface.water_jrc.gswe": JRC Global Surface Water Explorer surface water probability based on the Landsat time-series, "dtm_elev.dsm_nasadem.hgt": Digital Surface Model based on the USGS NASADEM, "lcv_landcover.12_pflugmacher2019": land cover map of Europe at 30 based on Pflugmacher et al. (2019), "lcv_tree.cover_umd.landsat_2000": forest tree cover for year 2000 based on the Global Forest Change data, "lcv_tree.cover_umd.landsat_2010": forest tree cover for year 2010 based on the Global Forest Change data, Detailed processing steps can be found <strong>here</strong>. Summary results of the model training (mlr::makeStackedLearner) report: <pre><code>Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -65.580 -2.630 0.648 3.120 181.769 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) -4.1448129 0.4663283 -8.888 &lt; 2e-16 *** regr.ranger 0.2667469 0.0009676 275.677 &lt; 2e-16 *** regr.glmnet -4.7183974 0.6038334 -7.814 5.54e-15 *** regr.cvglmnet 4.6966219 0.6042481 7.773 7.69e-15 *** regr.cubist 0.7643997 0.0012860 594.378 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' [...]
The paper describes production steps and accuracy assessment of an analysis-ready open environmen... more The paper describes production steps and accuracy assessment of an analysis-ready open environmental data cube (2000--2021+) for continental Europe; at working resolutions from 10~m to 30~m and with quarterly to annual estimates. The data cube is based on processing and harmonizing earth observation (EO) images: Landsat GLAD ARD (2000- -2020+), Sentinel-2 images (2017--2021+) and Digital Elevation data. These datasets were created with accessibility, user-friendliness, interoperability and synthesis in mind. This has required systematic spatiotemporal harmonization, efficient compression, and imputation of missing values. To ensure a missing value percentage below 1%, the EO data was ûrst aggregated into 4 quarterly periods approximating the 4 seasons common in Europe (winter, spring, summer and autumn), and then split into three percentiles (25th, 50th and 75th). Remaining missing data in the Landsat time-series was imputed with a temporal moving window median (TMWM) approach. The accuracy assessment shows TMWM gap-filling achieves higher performance in Southern Europe, and lower performance in mountainous regions such as the Scandinavian Mountains, the Alps, and the Pyrenees. The intended uses of the EcoDataCube platform include vegetation, soil, land cover and land use mapping projects, environmental monitoring and automated generation of data for statistical oûces including Eurostat. Combining all four datasets produced in this work (DTM, Landsat 30m, Sentinel-2 30m and Sentinel-2 10m) yields the highest land cover classification accuracy, with different datasets improving the results for diûerent land cover classes. The Environmental data cube for Europe is available under CC-BY license as Cloud-Optimized GeoTIFFs (ca. 12TB in size) through STAC and the EcoDataCube data portal.
iSDAsoil dataset soil texture classes derived from sand, silt and clay fractions at 30 m resoluti... more iSDAsoil dataset soil texture classes derived from sand, silt and clay fractions at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_texture.class_c_30m_*..*cm_2001..2017_v0.13_wgs84.tif = soil texture class, Classes: <pre><code>Code,Name,Value,Color Cl,clay,1,#d5c36b SiCl,silty clay,2,#b96947 SaCl,sandy clay,3,#9d3706 ClLo,clay loam,4,#ae868f SiClLo,silty clay loam,5,#f86714 SaClLo,sandy clay loam,6,#46d143 Lo,loam,7,#368f20 SiLo,silt loam,8,#3e5a14 SaLo,sandy loam,9,#ffd557 Si,silt,10,#fff72e LoSa,loamy sand,11,#ff5a9d Sa,sand,12,#ff005b NODATA,,255,#ffffff </code></pre> To submit an issue or request support please visit <strong>https://isda-africa.com/isdasoil</strong>
iSDAsoil dataset soil total organic Nitrogen (N) log-transformed predicted at 30 m resolution for... more iSDAsoil dataset soil total organic Nitrogen (N) log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.n_tot_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil total N mean value, sol_log.n_tot_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil total N model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.n_tot_ncs R-square: 0.732 Fitted values sd: 0.326 RMSE: 0.197 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -1.87298 -0.09584 -0.00985 0.07613 3.14728 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 0.267429 0.493235 0.542 0.588 regr.ranger 1.128208 0.005766 195.669 &lt; 2e-16 *** regr.xgboost -0.048780 0.006108 -7.987 1.4e-15 *** regr.cubist 0.143954 0.004424 32.539 &lt; 2e-16 *** regr.nnet -0.482261 0.797938 -0.604 0.546 regr.cvglmnet -0.170889 0.004955 -34.489 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.1972 on 99249 degrees of freedom Multiple R-squared: 0.7319, Adjusted R-squared: 0.7319 F-statistic: 5.419e+04 on 5 and 99249 DF, p-value: &lt; 2.2e-16</code></pre> To back-transform values (y) to g/kg use the following formula: <pre><code>g/kg = expm1( y / 100 )</code></pre> To submit an issue or request support please [...]
iSDAsoil dataset soil total carbon in permilles log-transformed predicted at 30 m resolution for ... more iSDAsoil dataset soil total carbon in permilles log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.c_tot_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil total Carbon mean value, sol_log.c_tot_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil total Carbon model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.c_tot R-square: 0.794 Fitted values sd: 0.571 RMSE: 0.291 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -2.70312 -0.16714 -0.00549 0.15691 3.01116 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 0.025841 0.032713 0.790 0.429570 regr.ranger 0.902240 0.008462 106.619 &lt; 2e-16 *** regr.xgboost 0.066535 0.008145 8.169 3.18e-16 *** regr.cubist 0.145730 0.006927 21.039 &lt; 2e-16 *** regr.nnet -0.048957 0.013466 -3.636 0.000278 *** regr.cvglmnet -0.075212 0.005556 -13.537 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.291 on 50140 degrees of freedom Multiple R-squared: 0.7938, Adjusted R-squared: 0.7938 F-statistic: 3.861e+04 on 5 and 50140 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request supp [...]
iSDAsoil dataset soil extractable Calcium log-transformed predicted at 30 m resolution for 0–20 a... more iSDAsoil dataset soil extractable Calcium log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.ca_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Calcium mean value, sol_log.ca_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Calcium model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.ca_mehlich3 R-square: 0.84 Fitted values sd: 1.24 RMSE: 0.543 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -6.0376 -0.2577 0.0076 0.2756 5.3825 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 5.737959 3.850998 1.490 0.136 regr.ranger 1.054018 0.003175 331.978 &lt; 2e-16 *** regr.xgboost -0.030930 0.003939 -7.853 4.1e-15 *** regr.cubist 0.061829 0.003561 17.364 &lt; 2e-16 *** regr.nnet -0.855297 0.561006 -1.525 0.127 regr.cvglmnet -0.065040 0.003225 -20.166 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.5428 on 144593 degrees of freedom Multiple R-squared: 0.8403, Adjusted R-squared: 0.8402 F-statistic: 1.521e+05 on 5 and 144593 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request support pl [...]
iSDAsoil dataset soil extractable Iron (Fe) log-transformed predicted at 30 m resolution for 0–20... more iSDAsoil dataset soil extractable Iron (Fe) log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.fe_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Iron mean value, sol_log.fe_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Iron model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.ca_mehlich3 R-square: 0.84 Fitted values sd: 1.24 RMSE: 0.543 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -6.0376 -0.2577 0.0076 0.2756 5.3825 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 5.737959 3.850998 1.490 0.136 regr.ranger 1.054018 0.003175 331.978 &lt; 2e-16 *** regr.xgboost -0.030930 0.003939 -7.853 4.1e-15 *** regr.cubist 0.061829 0.003561 17.364 &lt; 2e-16 *** regr.nnet -0.855297 0.561006 -1.525 0.127 regr.cvglmnet -0.065040 0.003225 -20.166 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.5428 on 144593 degrees of freedom Multiple R-squared: 0.8403, Adjusted R-squared: 0.8402 F-statistic: 1.521e+05 on 5 and 144593 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request support please [...]
iSDAsoil dataset soil extractable Sulphur (S) log-transformed predicted at 30 m resolution for 0–... more iSDAsoil dataset soil extractable Sulphur (S) log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.s_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Sulphur mean value, sol_log.s_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Sulphur model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.s_mehlich3 R-square: 0.548 Fitted values sd: 0.423 RMSE: 0.384 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -2.5729 -0.2102 -0.0264 0.1694 5.0049 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 1.459208 4.154229 0.351 0.725 regr.ranger 0.937179 0.016167 57.967 &lt; 2e-16 *** regr.xgboost 0.002587 0.016252 0.159 0.874 regr.cubist 0.145396 0.010890 13.351 &lt; 2e-16 *** regr.nnet -0.672062 1.796642 -0.374 0.708 regr.cvglmnet -0.045157 0.011256 -4.012 6.04e-05 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.3841 on 37530 degrees of freedom Multiple R-squared: 0.5481, Adjusted R-squared: 0.548 F-statistic: 9103 on 5 and 37530 DF, p-value: &lt; 2.2e-16</code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request support please visit <stron [...]
iSDAsoil dataset soil clay content (USDA system) in % predicted at 30 m resolution for 0–20 and 2... more iSDAsoil dataset soil clay content (USDA system) in % predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, LandPKS, and other national and regional soil datasets). Layer description: sol_clay_tot_psa_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil clay content mean value, sol_clay_tot_psa_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil clay content (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: clay_tot_psa R-square: 0.746 Fitted values sd: 16.5 RMSE: 9.63 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -75.803 -4.512 -0.178 3.748 82.146 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 4.494652 8.914671 0.504 0.61413 regr.ranger 1.076957 0.003611 298.210 &lt; 2e-16 *** regr.xgboost -0.012617 0.004678 -2.697 0.00699 ** regr.cubist 0.030730 0.003930 7.820 5.32e-15 *** regr.nnet -0.238376 0.365390 -0.652 0.51415 regr.cvglmnet -0.044547 0.004379 -10.174 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 9.629 on 122269 degrees of freedom Multiple R-squared: 0.7458, Adjusted R-squared: 0.7458 F-statistic: 7.175e+04 on 5 and 122269 DF, p-value: &lt; 2.2e-16</code></pre> To submit an issue or request support please visit <strong>https://isda-africa.com/isdasoil</strong>
iSDAsoil dataset soil extractable Potassium (K) log-transformed predicted at 30 m resolution for ... more iSDAsoil dataset soil extractable Potassium (K) log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.k_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Potassium mean value, sol_log.k_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Potassium model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.k_mehlich3 R-square: 0.773 Fitted values sd: 0.938 RMSE: 0.509 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -4.3088 -0.2648 -0.0037 0.2639 6.8136 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 10.907726 6.134422 1.778 0.0754 . regr.ranger 1.004487 0.003878 259.026 &lt;2e-16 *** regr.xgboost -0.004081 0.004739 -0.861 0.3892 regr.cubist 0.084556 0.004346 19.454 &lt;2e-16 *** regr.nnet -2.205286 1.228586 -1.795 0.0727 . regr.cvglmnet -0.064510 0.003933 -16.401 &lt;2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.5092 on 139122 degrees of freedom Multiple R-squared: 0.7725, Adjusted R-squared: 0.7725 F-statistic: 9.451e+04 on 5 and 139122 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request s [...]
iSDAsoil Africa dataset soil pH (in H2O) predicted at 30 m resolution for 0-20 and 20-50 cm depth... more iSDAsoil Africa dataset soil pH (in H2O) predicted at 30 m resolution for 0-20 and 20-50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Layer description: sol_ph_h2o_m_30m_20..50cm_2001..2017_v0.13_wgs84.tif = predicted soil pH mean value, sol_ph_h2o_md_30m_20..50cm_2001..2017_v0.13_wgs84.tif = predicted soil pH model (prediction) errors, Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (ALOS AW3D, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, LandPKS, and other national and regional soil datasets). Model errors are derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) indicates: <pre><code>Variable: ph_h2o R-square: 0.818 Fitted values sd: 0.972 RMSE: 0.459 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -5.5939 -0.2328 -0.0066 0.2222 4.7477 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 1.113440 1.164473 0.956 0.338986 regr.ranger 1.032918 0.003138 329.116 &lt; 2e-16 *** regr.xgboost -0.014201 0.004185 -3.393 0.000691 *** regr.cubist 0.049667 0.003709 13.392 &lt; 2e-16 *** regr.nnet -0.188570 0.188214 -1.002 0.316398 regr.cvglmnet -0.059763 0.003636 -16.438 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.4591 on 133378 degrees of freedom Multiple R-squared: 0.8176, Adjusted R-squared: 0.8176 F-statistic: 1.195e+05 on 5 and 133378 DF, p-value: &lt; 2.2e-16</code></pre> To submit an issue or request support please visit <strong>https://soil.isda-africa.com</strong>
iSDAsoil dataset soil extractable Phosphorus (P) log-transformed predicted at 30 m resolution for... more iSDAsoil dataset soil extractable Phosphorus (P) log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.p_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Phosphorus mean value, sol_log.p_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Phosphorus model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.p_mehlich3 R-square: 0.486 Fitted values sd: 0.687 RMSE: 0.707 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -3.2892 -0.3942 -0.0637 0.2614 4.9466 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 3.378801 3.143200 1.075 0.282 regr.ranger 0.861655 0.011099 77.631 &lt; 2e-16 *** regr.xgboost 0.066139 0.013091 5.052 4.38e-07 *** regr.cubist 0.157674 0.008886 17.744 &lt; 2e-16 *** regr.nnet -1.649621 1.442240 -1.144 0.253 regr.cvglmnet 0.013628 0.010407 1.310 0.190 --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.7066 on 53493 degrees of freedom Multiple R-squared: 0.486, Adjusted R-squared: 0.486 F-statistic: 1.012e+04 on 5 and 53493 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request support please [...]
iSDAsoil dataset soil stone content / coarse fragments log-transformed predicted at 30 m resoluti... more iSDAsoil dataset soil stone content / coarse fragments log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.wpg2_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil stone content mean value, sol_log.wpg2_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil stone content model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.wpg2 R-square: 0.709 Fitted values sd: 1.25 RMSE: 0.803 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -4.0555 -0.3113 -0.0222 0.2378 4.5794 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) -0.008606 1.361982 -0.006 0.995 regr.ranger 0.972265 0.004443 218.854 &lt; 2e-16 *** regr.xgboost 0.034649 0.006404 5.411 6.3e-08 *** regr.cubist 0.069589 0.005229 13.308 &lt; 2e-16 *** regr.nnet -0.012756 0.796535 -0.016 0.987 regr.cvglmnet -0.056645 0.005509 -10.283 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.8032 on 92785 degrees of freedom Multiple R-squared: 0.7092, Adjusted R-squared: 0.7092 F-statistic: 4.525e+04 on 5 and 92785 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to % use the following formula: <pre><code>% = expm1( y / 10 )</code></pre> To submit an issue or request support please v [...]
Digital Terrain Model for Continental Europe based on the three publicly available Digital Surfac... more Digital Terrain Model for Continental Europe based on the three publicly available Digital Surface Models and predicted using an Ensemble Machine Learning (EML). EML was trainined using GEDI level 2B points (Level 2A; "elev_lowestmode") and ICESat-2 (ATL08; "h_te_mean"): about 9 million points were overlaid vs MERITDEM, AW3D30, GLO-30, EU DEM, GLAD canopy height, tree cover and surface water cover maps, then an ensemble prediction model (mlr package in R) was fitted using random forest, Cubist and GLM, and used to predict most probable terrain height (bare earth). Input layers used to train the EML include: "lcv_bare.earth_glcf.landsat": UMD GLAD bare earth estimate for year 2010 based on Landsat time series, "dtm_elev.dsm_alos.aw3d": Digital Surface Model based on ALOS AW3D, "dtm_canopy.height_glad.umd": UMD GLAD canopy height for 2019 based on GEDI data, "dtm_elev.dsm_eudem.eea": Copernicus EU DEM based on the SRTM and ASTER DEMs, "hyd_surface.water_jrc.gswe": JRC Global Surface Water Explorer surface water probability based on the Landsat time-series, "lcv_landcover.12_pflugmacher2019": land cover map of Europe at 30 based on Pflugmacher et al. (2019), "lcv_tree.cover_umd.landsat_2000": forest tree cover for year 2000 based on the Global Forest Change data, "lcv_tree.cover_umd.landsat_2010": forest tree cover for year 2010 based on the Global Forest Change data, Detailed processing steps can be found <strong>here</strong>. Read more about the processing steps <strong>here</strong>. Training data set can be obtained in the file "gedi_elev.lowestmode_2019_eumap.RDS". The initial linear model fitted using the four independent Digital Surface / Digital Terrain models shows: <pre><code>Residuals: Min 1Q Median 3Q Max -124.627 -1.097 0.973 2.544 59.324 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) -1.6220640 0.0032415 -500.4 &lt;2e-16 *** eu_dem25m_ -0.1092988 0.0005531 -197.6 &lt;2e-16 *** eu_AW3Dv2012_30m_ 0.0933774 0.0005957 156.7 &lt;2e-16 *** eu_GLO30_30m_ 0.2637153 0.0006062 435.1 &lt;2e-16 [...]
Abstract: The waters of the Mirna River basin have been extensively used for supply of a large po... more Abstract: The waters of the Mirna River basin have been extensively used for supply of a large portion of the population, tourists, industry and agricultural irrigation in Istria. An enlargement of water reserve for current uses, as well as for construction of several ...
Abstract: The purpose of this study was to analyze preliminary results and to determine the frequ... more Abstract: The purpose of this study was to analyze preliminary results and to determine the frequency and types of habitats used by the wolf marked with GPS collar. The wolf was captured and marked in Jelovac above village Krasno in Lika region. In the first 18 days ...
iSDAsoil dataset soil extractable Aluminium log-transformed predicted at 30 m resolution for 0–20... more iSDAsoil dataset soil extractable Aluminium log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, LandPKS, and other national and regional soil datasets). Layer description: sol_db_od_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Aluminium mean value, sol_db_od_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Aluminium model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.al_mehlich3 R-square: 0.881 Fitted values sd: 0.872 RMSE: 0.321 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -5.7042 -0.1036 0.0059 0.1189 3.3777 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) -0.675492 2.771906 -0.244 0.807 regr.ranger 0.879567 0.005464 160.969 &lt;2e-16 *** regr.xgboost 0.071537 0.005813 12.306 &lt;2e-16 *** regr.cubist 0.150157 0.004553 32.979 &lt;2e-16 *** regr.nnet 0.087603 0.431261 0.203 0.839 regr.cvglmnet -0.084440 0.003182 -26.534 &lt;2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.3208 on 63551 degrees of freedom Multiple R-squared: 0.8808, Adjusted R-squared: 0.8808 F-statistic: 9.391e+04 on 5 and 63551 DF, p-value: &lt; 2.2e-16</code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request support please vis [...]
Digital Terrain Model for Continental Europe based on the three publicly available Digital Surfac... more Digital Terrain Model for Continental Europe based on the three publicly available Digital Surface Models and predicted using an Ensemble Machine Learning (EML). EML was trainined using GEDI level 2B points (column "elev.lowestmode"): about 7 million GEDI points were overlaid vs NASADEM, AW3D, EU DEM, canopy height, tree cover and surface water cover maps, then a an ensemble prediction model was fitted using random forest, GLM with Lasso, Cubist and GLMnet, and used to predict most probable terrain height (bare earth). Input layers used to train the EML include: "lcv_bare.earth_glcf.landsat": UMD GLAD bare earth estimate for year 2010 based on Landsat time series, "dtm_elev.dsm_alos.aw3d": Digital Surface Model based on ALOS AW3D v8104, "dtm_canopy.height_glad.umd": UMD GLAD canopy height for 2019 based on GEDI data, "dtm_elev.dsm_eudem.eea": Copernicus EU DEM based on the SRTM and ASTER DEMs, "hyd_surface.water_jrc.gswe": JRC Global Surface Water Explorer surface water probability based on the Landsat time-series, "dtm_elev.dsm_nasadem.hgt": Digital Surface Model based on the USGS NASADEM, "lcv_landcover.12_pflugmacher2019": land cover map of Europe at 30 based on Pflugmacher et al. (2019), "lcv_tree.cover_umd.landsat_2000": forest tree cover for year 2000 based on the Global Forest Change data, "lcv_tree.cover_umd.landsat_2010": forest tree cover for year 2010 based on the Global Forest Change data, Detailed processing steps can be found <strong>here</strong>. Summary results of the model training (mlr::makeStackedLearner) report: <pre><code>Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -65.580 -2.630 0.648 3.120 181.769 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) -4.1448129 0.4663283 -8.888 &lt; 2e-16 *** regr.ranger 0.2667469 0.0009676 275.677 &lt; 2e-16 *** regr.glmnet -4.7183974 0.6038334 -7.814 5.54e-15 *** regr.cvglmnet 4.6966219 0.6042481 7.773 7.69e-15 *** regr.cubist 0.7643997 0.0012860 594.378 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' [...]
The paper describes production steps and accuracy assessment of an analysis-ready open environmen... more The paper describes production steps and accuracy assessment of an analysis-ready open environmental data cube (2000--2021+) for continental Europe; at working resolutions from 10~m to 30~m and with quarterly to annual estimates. The data cube is based on processing and harmonizing earth observation (EO) images: Landsat GLAD ARD (2000- -2020+), Sentinel-2 images (2017--2021+) and Digital Elevation data. These datasets were created with accessibility, user-friendliness, interoperability and synthesis in mind. This has required systematic spatiotemporal harmonization, efficient compression, and imputation of missing values. To ensure a missing value percentage below 1%, the EO data was ûrst aggregated into 4 quarterly periods approximating the 4 seasons common in Europe (winter, spring, summer and autumn), and then split into three percentiles (25th, 50th and 75th). Remaining missing data in the Landsat time-series was imputed with a temporal moving window median (TMWM) approach. The accuracy assessment shows TMWM gap-filling achieves higher performance in Southern Europe, and lower performance in mountainous regions such as the Scandinavian Mountains, the Alps, and the Pyrenees. The intended uses of the EcoDataCube platform include vegetation, soil, land cover and land use mapping projects, environmental monitoring and automated generation of data for statistical oûces including Eurostat. Combining all four datasets produced in this work (DTM, Landsat 30m, Sentinel-2 30m and Sentinel-2 10m) yields the highest land cover classification accuracy, with different datasets improving the results for diûerent land cover classes. The Environmental data cube for Europe is available under CC-BY license as Cloud-Optimized GeoTIFFs (ca. 12TB in size) through STAC and the EcoDataCube data portal.
iSDAsoil dataset soil texture classes derived from sand, silt and clay fractions at 30 m resoluti... more iSDAsoil dataset soil texture classes derived from sand, silt and clay fractions at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_texture.class_c_30m_*..*cm_2001..2017_v0.13_wgs84.tif = soil texture class, Classes: <pre><code>Code,Name,Value,Color Cl,clay,1,#d5c36b SiCl,silty clay,2,#b96947 SaCl,sandy clay,3,#9d3706 ClLo,clay loam,4,#ae868f SiClLo,silty clay loam,5,#f86714 SaClLo,sandy clay loam,6,#46d143 Lo,loam,7,#368f20 SiLo,silt loam,8,#3e5a14 SaLo,sandy loam,9,#ffd557 Si,silt,10,#fff72e LoSa,loamy sand,11,#ff5a9d Sa,sand,12,#ff005b NODATA,,255,#ffffff </code></pre> To submit an issue or request support please visit <strong>https://isda-africa.com/isdasoil</strong>
iSDAsoil dataset soil total organic Nitrogen (N) log-transformed predicted at 30 m resolution for... more iSDAsoil dataset soil total organic Nitrogen (N) log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.n_tot_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil total N mean value, sol_log.n_tot_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil total N model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.n_tot_ncs R-square: 0.732 Fitted values sd: 0.326 RMSE: 0.197 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -1.87298 -0.09584 -0.00985 0.07613 3.14728 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 0.267429 0.493235 0.542 0.588 regr.ranger 1.128208 0.005766 195.669 &lt; 2e-16 *** regr.xgboost -0.048780 0.006108 -7.987 1.4e-15 *** regr.cubist 0.143954 0.004424 32.539 &lt; 2e-16 *** regr.nnet -0.482261 0.797938 -0.604 0.546 regr.cvglmnet -0.170889 0.004955 -34.489 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.1972 on 99249 degrees of freedom Multiple R-squared: 0.7319, Adjusted R-squared: 0.7319 F-statistic: 5.419e+04 on 5 and 99249 DF, p-value: &lt; 2.2e-16</code></pre> To back-transform values (y) to g/kg use the following formula: <pre><code>g/kg = expm1( y / 100 )</code></pre> To submit an issue or request support please [...]
iSDAsoil dataset soil total carbon in permilles log-transformed predicted at 30 m resolution for ... more iSDAsoil dataset soil total carbon in permilles log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.c_tot_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil total Carbon mean value, sol_log.c_tot_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil total Carbon model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.c_tot R-square: 0.794 Fitted values sd: 0.571 RMSE: 0.291 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -2.70312 -0.16714 -0.00549 0.15691 3.01116 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 0.025841 0.032713 0.790 0.429570 regr.ranger 0.902240 0.008462 106.619 &lt; 2e-16 *** regr.xgboost 0.066535 0.008145 8.169 3.18e-16 *** regr.cubist 0.145730 0.006927 21.039 &lt; 2e-16 *** regr.nnet -0.048957 0.013466 -3.636 0.000278 *** regr.cvglmnet -0.075212 0.005556 -13.537 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.291 on 50140 degrees of freedom Multiple R-squared: 0.7938, Adjusted R-squared: 0.7938 F-statistic: 3.861e+04 on 5 and 50140 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request supp [...]
iSDAsoil dataset soil extractable Calcium log-transformed predicted at 30 m resolution for 0–20 a... more iSDAsoil dataset soil extractable Calcium log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.ca_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Calcium mean value, sol_log.ca_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Calcium model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.ca_mehlich3 R-square: 0.84 Fitted values sd: 1.24 RMSE: 0.543 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -6.0376 -0.2577 0.0076 0.2756 5.3825 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 5.737959 3.850998 1.490 0.136 regr.ranger 1.054018 0.003175 331.978 &lt; 2e-16 *** regr.xgboost -0.030930 0.003939 -7.853 4.1e-15 *** regr.cubist 0.061829 0.003561 17.364 &lt; 2e-16 *** regr.nnet -0.855297 0.561006 -1.525 0.127 regr.cvglmnet -0.065040 0.003225 -20.166 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.5428 on 144593 degrees of freedom Multiple R-squared: 0.8403, Adjusted R-squared: 0.8402 F-statistic: 1.521e+05 on 5 and 144593 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request support pl [...]
iSDAsoil dataset soil extractable Iron (Fe) log-transformed predicted at 30 m resolution for 0–20... more iSDAsoil dataset soil extractable Iron (Fe) log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.fe_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Iron mean value, sol_log.fe_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Iron model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.ca_mehlich3 R-square: 0.84 Fitted values sd: 1.24 RMSE: 0.543 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -6.0376 -0.2577 0.0076 0.2756 5.3825 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 5.737959 3.850998 1.490 0.136 regr.ranger 1.054018 0.003175 331.978 &lt; 2e-16 *** regr.xgboost -0.030930 0.003939 -7.853 4.1e-15 *** regr.cubist 0.061829 0.003561 17.364 &lt; 2e-16 *** regr.nnet -0.855297 0.561006 -1.525 0.127 regr.cvglmnet -0.065040 0.003225 -20.166 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.5428 on 144593 degrees of freedom Multiple R-squared: 0.8403, Adjusted R-squared: 0.8402 F-statistic: 1.521e+05 on 5 and 144593 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request support please [...]
iSDAsoil dataset soil extractable Sulphur (S) log-transformed predicted at 30 m resolution for 0–... more iSDAsoil dataset soil extractable Sulphur (S) log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.s_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Sulphur mean value, sol_log.s_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Sulphur model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.s_mehlich3 R-square: 0.548 Fitted values sd: 0.423 RMSE: 0.384 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -2.5729 -0.2102 -0.0264 0.1694 5.0049 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 1.459208 4.154229 0.351 0.725 regr.ranger 0.937179 0.016167 57.967 &lt; 2e-16 *** regr.xgboost 0.002587 0.016252 0.159 0.874 regr.cubist 0.145396 0.010890 13.351 &lt; 2e-16 *** regr.nnet -0.672062 1.796642 -0.374 0.708 regr.cvglmnet -0.045157 0.011256 -4.012 6.04e-05 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.3841 on 37530 degrees of freedom Multiple R-squared: 0.5481, Adjusted R-squared: 0.548 F-statistic: 9103 on 5 and 37530 DF, p-value: &lt; 2.2e-16</code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request support please visit <stron [...]
iSDAsoil dataset soil clay content (USDA system) in % predicted at 30 m resolution for 0–20 and 2... more iSDAsoil dataset soil clay content (USDA system) in % predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, LandPKS, and other national and regional soil datasets). Layer description: sol_clay_tot_psa_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil clay content mean value, sol_clay_tot_psa_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil clay content (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: clay_tot_psa R-square: 0.746 Fitted values sd: 16.5 RMSE: 9.63 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -75.803 -4.512 -0.178 3.748 82.146 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 4.494652 8.914671 0.504 0.61413 regr.ranger 1.076957 0.003611 298.210 &lt; 2e-16 *** regr.xgboost -0.012617 0.004678 -2.697 0.00699 ** regr.cubist 0.030730 0.003930 7.820 5.32e-15 *** regr.nnet -0.238376 0.365390 -0.652 0.51415 regr.cvglmnet -0.044547 0.004379 -10.174 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 9.629 on 122269 degrees of freedom Multiple R-squared: 0.7458, Adjusted R-squared: 0.7458 F-statistic: 7.175e+04 on 5 and 122269 DF, p-value: &lt; 2.2e-16</code></pre> To submit an issue or request support please visit <strong>https://isda-africa.com/isdasoil</strong>
iSDAsoil dataset soil extractable Potassium (K) log-transformed predicted at 30 m resolution for ... more iSDAsoil dataset soil extractable Potassium (K) log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.k_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Potassium mean value, sol_log.k_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Potassium model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.k_mehlich3 R-square: 0.773 Fitted values sd: 0.938 RMSE: 0.509 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -4.3088 -0.2648 -0.0037 0.2639 6.8136 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 10.907726 6.134422 1.778 0.0754 . regr.ranger 1.004487 0.003878 259.026 &lt;2e-16 *** regr.xgboost -0.004081 0.004739 -0.861 0.3892 regr.cubist 0.084556 0.004346 19.454 &lt;2e-16 *** regr.nnet -2.205286 1.228586 -1.795 0.0727 . regr.cvglmnet -0.064510 0.003933 -16.401 &lt;2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.5092 on 139122 degrees of freedom Multiple R-squared: 0.7725, Adjusted R-squared: 0.7725 F-statistic: 9.451e+04 on 5 and 139122 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request s [...]
iSDAsoil Africa dataset soil pH (in H2O) predicted at 30 m resolution for 0-20 and 20-50 cm depth... more iSDAsoil Africa dataset soil pH (in H2O) predicted at 30 m resolution for 0-20 and 20-50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Layer description: sol_ph_h2o_m_30m_20..50cm_2001..2017_v0.13_wgs84.tif = predicted soil pH mean value, sol_ph_h2o_md_30m_20..50cm_2001..2017_v0.13_wgs84.tif = predicted soil pH model (prediction) errors, Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (ALOS AW3D, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, LandPKS, and other national and regional soil datasets). Model errors are derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) indicates: <pre><code>Variable: ph_h2o R-square: 0.818 Fitted values sd: 0.972 RMSE: 0.459 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -5.5939 -0.2328 -0.0066 0.2222 4.7477 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 1.113440 1.164473 0.956 0.338986 regr.ranger 1.032918 0.003138 329.116 &lt; 2e-16 *** regr.xgboost -0.014201 0.004185 -3.393 0.000691 *** regr.cubist 0.049667 0.003709 13.392 &lt; 2e-16 *** regr.nnet -0.188570 0.188214 -1.002 0.316398 regr.cvglmnet -0.059763 0.003636 -16.438 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.4591 on 133378 degrees of freedom Multiple R-squared: 0.8176, Adjusted R-squared: 0.8176 F-statistic: 1.195e+05 on 5 and 133378 DF, p-value: &lt; 2.2e-16</code></pre> To submit an issue or request support please visit <strong>https://soil.isda-africa.com</strong>
iSDAsoil dataset soil extractable Phosphorus (P) log-transformed predicted at 30 m resolution for... more iSDAsoil dataset soil extractable Phosphorus (P) log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.p_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Phosphorus mean value, sol_log.p_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil extractable Phosphorus model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.p_mehlich3 R-square: 0.486 Fitted values sd: 0.687 RMSE: 0.707 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -3.2892 -0.3942 -0.0637 0.2614 4.9466 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) 3.378801 3.143200 1.075 0.282 regr.ranger 0.861655 0.011099 77.631 &lt; 2e-16 *** regr.xgboost 0.066139 0.013091 5.052 4.38e-07 *** regr.cubist 0.157674 0.008886 17.744 &lt; 2e-16 *** regr.nnet -1.649621 1.442240 -1.144 0.253 regr.cvglmnet 0.013628 0.010407 1.310 0.190 --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.7066 on 53493 degrees of freedom Multiple R-squared: 0.486, Adjusted R-squared: 0.486 F-statistic: 1.012e+04 on 5 and 53493 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to ppm use the following formula: <pre><code>ppm = expm1( y / 10 )</code></pre> To submit an issue or request support please [...]
iSDAsoil dataset soil stone content / coarse fragments log-transformed predicted at 30 m resoluti... more iSDAsoil dataset soil stone content / coarse fragments log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Layer description: sol_log.wpg2_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil stone content mean value, sol_log.wpg2_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil stone content model (prediction) errors, Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates: <pre><code>Variable: log.wpg2 R-square: 0.709 Fitted values sd: 1.25 RMSE: 0.803 Random forest model: Call: stats::lm(formula = f, data = d) Residuals: Min 1Q Median 3Q Max -4.0555 -0.3113 -0.0222 0.2378 4.5794 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) -0.008606 1.361982 -0.006 0.995 regr.ranger 0.972265 0.004443 218.854 &lt; 2e-16 *** regr.xgboost 0.034649 0.006404 5.411 6.3e-08 *** regr.cubist 0.069589 0.005229 13.308 &lt; 2e-16 *** regr.nnet -0.012756 0.796535 -0.016 0.987 regr.cvglmnet -0.056645 0.005509 -10.283 &lt; 2e-16 *** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.8032 on 92785 degrees of freedom Multiple R-squared: 0.7092, Adjusted R-squared: 0.7092 F-statistic: 4.525e+04 on 5 and 92785 DF, p-value: &lt; 2.2e-16 </code></pre> To back-transform values (y) to % use the following formula: <pre><code>% = expm1( y / 10 )</code></pre> To submit an issue or request support please v [...]
Digital Terrain Model for Continental Europe based on the three publicly available Digital Surfac... more Digital Terrain Model for Continental Europe based on the three publicly available Digital Surface Models and predicted using an Ensemble Machine Learning (EML). EML was trainined using GEDI level 2B points (Level 2A; "elev_lowestmode") and ICESat-2 (ATL08; "h_te_mean"): about 9 million points were overlaid vs MERITDEM, AW3D30, GLO-30, EU DEM, GLAD canopy height, tree cover and surface water cover maps, then an ensemble prediction model (mlr package in R) was fitted using random forest, Cubist and GLM, and used to predict most probable terrain height (bare earth). Input layers used to train the EML include: "lcv_bare.earth_glcf.landsat": UMD GLAD bare earth estimate for year 2010 based on Landsat time series, "dtm_elev.dsm_alos.aw3d": Digital Surface Model based on ALOS AW3D, "dtm_canopy.height_glad.umd": UMD GLAD canopy height for 2019 based on GEDI data, "dtm_elev.dsm_eudem.eea": Copernicus EU DEM based on the SRTM and ASTER DEMs, "hyd_surface.water_jrc.gswe": JRC Global Surface Water Explorer surface water probability based on the Landsat time-series, "lcv_landcover.12_pflugmacher2019": land cover map of Europe at 30 based on Pflugmacher et al. (2019), "lcv_tree.cover_umd.landsat_2000": forest tree cover for year 2000 based on the Global Forest Change data, "lcv_tree.cover_umd.landsat_2010": forest tree cover for year 2010 based on the Global Forest Change data, Detailed processing steps can be found <strong>here</strong>. Read more about the processing steps <strong>here</strong>. Training data set can be obtained in the file "gedi_elev.lowestmode_2019_eumap.RDS". The initial linear model fitted using the four independent Digital Surface / Digital Terrain models shows: <pre><code>Residuals: Min 1Q Median 3Q Max -124.627 -1.097 0.973 2.544 59.324 Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) -1.6220640 0.0032415 -500.4 &lt;2e-16 *** eu_dem25m_ -0.1092988 0.0005531 -197.6 &lt;2e-16 *** eu_AW3Dv2012_30m_ 0.0933774 0.0005957 156.7 &lt;2e-16 *** eu_GLO30_30m_ 0.2637153 0.0006062 435.1 &lt;2e-16 [...]
Uploads
Papers by josip krizan