Tarin Paz-Kagan

Comparing the Effect of Preprocessing Transformations on Methods of Land-Use Classification Derived From Spectral Soil Measurements

Advanced classifiers, e.g., partial least squares discriminant analysis (PLS-DA) and random forests (RF), have been recently used to model reflectance spectral data in general, and of soil properties in particular, since their spectra are multivariate and highly collinear. Preprocessing transformations (PPTs) can improve the classification accuracy by increasing the variability between classes while decreasing the variability within classes. Such PPTs are common practice prior to a PLS-DA, but are rarely used for RF. The objectives of this paper are twofold: to compare the performances of PLS-DA and RF for modeling the spectral reflectance of soil in changed land-uses with different treatments and to compare the effects of nine different PPTs on the prediction accuracy of each of these classification methods. Differences in six physical, biological, and chemical soil properties of changed land-uses from the northern Negev Desert in Israel were evaluated. Significant differences were found between soil properties, which are used to classify land-uses and treatments. Depending on the dataset, different PPTs improved the classification accuracy by 11%-24% and 32%-42% for PLS-DA and RF, respectively, in comparison to the spectra without PPT. Out of the PPTs tested, the generalized least squares weighting (GLSW)-based transformations were found to be the most effective for most classifications using both PLS-DA and RF. Our results show that both PLS-DA and RF are suitable classifiers for spectral data, provided that an appropriate PPT is applied.

Structural Changes of Desertified and Managed Shrubland Landscapes in Response to Drought: Spectral, Spatial and Temporal Analyses

Drought events cause changes in ecosystem function and structure by reducing the shrub abundance and expanding the biological soil crusts (biocrusts). This change increases the leakage of nutrient resources and water into the river streams in semi-arid areas. A common management solution for decreasing this loss of resources is to create a runoff-harvesting system (RHS). The objective of the current research is to apply geo-information techniques, including remote sensing and geographic information systems (GIS), on the watershed scale, to monitor and analyze the spatial and temporal changes in response to drought of two source-sink systems, the natural shrubland and the human-made RHSs in the semi-arid area of the northern Negev Desert, Israel. This was done by evaluating the changes in soil, vegetation and landscape cover. The spatial changes were evaluated by three spectral indices: Normalized Difference Vegetation Index (NDVI), Crust Index (CI) and landscape classification change between 2003 and 2010. In addition, we examined the effects of environmental factors on NDVI, CI and their clustering after successive drought years. The results show that vegetation cover indicates a negative ∆NDVI change due to a reduction in the abundance of woody vegetation. On the other hand, the soil cover change data indicate a positive ∆CI change due to the expansion of the biocrusts. These two trends are evidence for degradation processes in terms of resource conservation and bio-production. A considerable part of the changed area (39%) represents transitions between redistribution processes of resources, such as water, sediments, nutrients and seeds, on the watershed scale. In the pre-drought period, resource redistribution mainly occurred on the slope scale, while in the post-drought period, resource redistribution occurred on the whole watershed scale. However, the RHS management is effective in reducing leakage, since these systems are located on the slopes where the magnitude of runoff pulses is low.

Mapping the Spectral Soil Quality Index (SSQI) Using Airborne Imaging Spectroscopy

Soil quality (SQ) assessment has numerous applications for managing sustainable soil function. Airborne imaging spectroscopy (IS) is an advanced tool for studying natural and artificial materials, in general, and soil properties, in particular. The primary goal of this research was to prove and demonstrate the ability of IS to evaluate soil properties and quality across anthropogenically induced land-use changes. This aim was fulfilled by developing and implementing a spectral soil quality index (SSQI) using IS obtained by a laboratory and field spectrometer (point scale) as well as by airborne hyperspectral imaging (local scale), in two experimental sites located in Israel and Germany. In this regard, 13 soil physical, biological, and chemical properties and their derived soil quality index (SQI) were measured. Several mathematical/statistical procedures, consisting of a series of operations, including a principal component analysis (PCA), a partial least squares-regression (PLS-R), and a partial least squares-discriminate analysis (PLS-DA), were used. Correlations between the laboratory spectral values and the calculated SQI coefficient of determination (R2) and ratio of performance to deviation (RPD) were R2 = 0.84; RPD = 2.43 and R2 = 0.78; RPD = 2.10 in the Israeli and the German study sites, respectively. The PLS-DA model that was used to develop the SSQI showed high classification accuracy in both sites (from laboratory, field, and imaging spectroscopy). The correlations between the SSQI and the SQI were R2 = 0.71 and R2 = 0.7, in the Israeli and the German study sites, respectively. It is concluded that soil quality can be effectively monitored using the spectral-spatial information provided by the IS technology. IS-based classification of soils can provide the basis for a spatially explicit and quantitative approach for monitoring SQ and function at a local scale.

Linear Multi-Task Learning for Predicting Soil Properties Using Field Spectroscopy

Field spectroscopy has been suggested to be an efficient method for predicting soil properties using quantitative mathematical models in a rapid and non-destructive manner. Traditional multivariate regression algorithms usually regard the modeling of each soil property as a single task, which means only one response variable is considered as the output during modeling. Therefore, these algorithms are less suitable for the prediction of several key soil properties with low concentrations or unobvious spectral absorption signals. In the current study, we investigated the performance of a linear multi-task learning (LMTL) algorithm based on a regularized dirty model for modeling and predicting several key soil properties using field spectroscopy (350–2500 nm) as an integrated approach. We tested seven key soil properties including available nitrogen (N), phosphorus (P) and potassium (K), pH, water content (WC), organic matter (OM), and electrical conductivity (EC) in drylands. The model performances of LMTL models were compared with the commonly used single-task algorithm of the partial least squares regression (PLS-R). Our results show that the LMTL models outperformed the PLS-R models with the advantage of shared features; the ratio of performance to deviation (RPD) values in the validation set improved by 10.24%, 4.93%, 25.77%, 11.76%, 6.74%, 53.13%, and 3.15% for N, P, K, pH, WC, OM, and EC, respectively. The best prediction was obtained for OM with RPD = 2.29, indicating high accuracy (RPD > 2). The prediction results of N, P, WC, and pH were categorized as of moderate accuracy (1.4 < RPD < 2), while K and EC were categorized as of poor accuracy (RPD < 1.4). However, the explanatory power of the LMTL models was moderate due to fewer features being selected by the regularization algorithm of the LMTL approach, which should be further studied in the soil spectral analysis. Our results highlight the use of LMTL in field spectroscopy analysis that can improve the generalization performance of regression models for predicting soil properties.

Time series analysis of vegetation-cover response to environmental factors and residential development in a dryland region

Land-use changes as a result of residential development often lead to degradation and alter vegetation cover (VC). Although these are worldwide phenomena, sufficient knowledge about anthropogenic effects caused by various populated areas in dryland ecosystems is lacking. This study explored anthropogenic development in rural areas and its effects on the conservation of protected areas in drylands, focusing on the change in VC, the reasons, extent, and the drivers of change. We propose a novel framework for exploring VC change (VCC) as a function of environmental and human-driven factors including different types of populated areas in drylands. As a case study, we used a 30-year time series of Landsat satellite images over the arid region of Israel to analyze spatiotemporal VCC. The temporal analysis involved the Contextual Mann-Kendall significance test and spatial analysis to model clustering of VCC. A Gradient Boosted Regression machine learning algorithm was applied to study the relative influence of environmental and human-driven factors on VCC. In addition, we used ANOVA to examine differences between the effects of three types of populated areas on the spatiotemporal trends of VC. The results show that the most influential environmental variable on VCC was elevation (relative contribution of 17%), followed by slope (14.8%) and distance from populated areas (14.6%). Moreover, different types of populated areas affected VC differently with varying distances from residential centroids. The nature reserves increased VC positively and significantly, while livestock settlements had a negative effect. Change in vegetation was mostly confined to the stream network and occurred in lower elevations. The study demonstrates how different land-use practices alter the landscape in terms of VC and differ in their extents, patterns, and effects. With the expected growth in population and residential development worldwide, the proposed framework may assist conservation managements and policy makers in minimizing environmental degradation in drylands.

Forest composition effect on wildfire pattern and run-off regime in a Mediterranean watershed

The Remote Sensing Laboratory, Jacob Blaustein Institutes for Desert Research, Ben‐Gurion University of the Negev, Sede Boker Campus, Beersheba 84990, Israel 2 Israeli Hydrological Service, Israel Water Authority, Jerusalem 91360, Israel Correspondence Arnon Karnieli, The Remote Sensing Laboratory, Jacob Blaustein Institutes for Desert Research, Ben‐Gurion University of the Negev, Sede Boker Campus, Beersheba 84990, Israel. Email: karnieli@bgu.ac.il Funding information European Unions Horizon 2020 Research and Innovation Programme, Grant/Award Numbers: 654359 and 641762; Ministry of Science and Technology via the Ramon Foundation, Israel, Grant/Award Number: 3‐10673; Israel Water Authority, Grant/Award Number: 4500686906