Relationships between spectral features, iron oxides and colours of surface soils in northern Jordan

Ground data on spectral characteristics of Jordan's soils remain sparse, which makes the interpretation of remote sensing datasets challenging. These are, however, very useful for predicting soil properties and agricultural suitability. Previous studies have shown that soil colours correlate well with degrees of weathering intensity, as indicated by magnetic parameters and dithionite‐extractable iron (Fed) contents along a climosequence in northern Jordan. This study enlarges the database by the results of 160 bulk samples that were collected systematically from the soil surface at 40 locations. In addition to soil colour and contents of Fed, we explore mean soil reflectance spectra (MSRS) measured by analytical spectral devices (ASD) and analyse the morphological conditions of the spectra referring to the effects of iron oxides on spectral behaviours. Results show a high correlation of spectral behaviours and colour variations with Fe oxides, and no correlation with soil organic carbon Corg. The influence of the Fe oxide contents is clearly apparent in the visible range (VIS). The presence of CaCO3 increases the reflectance in all spectral ranges. Six soil groups (Gr. I ‐ Gr. VI) were discerned qualitatively and quantitatively in the study area, which largely mirror the intensity of red colour expressed by redness indices.


| INTRODUCTION
The north-western part of Jordan is located in the Mediterranean climate zone. Soil cover is limited in extent and essential for agricultural production, as climatic conditions are not favourable for agriculture in most parts of Jordan. Sustainable soil management and development of the agricultural and pastoral land-use sectors are of major concern in the region and require knowledge on the spatial variations of soils and their properties. Apart from various case studies on soil development, erosionsedimentation and land-use history of north-western Jordan (Lucke, 2008(Lucke, , 2017Lucke et al., 2008;Lucke, Kemnitz, & Bäumler, 2012;Lucke, Kemnitz, Bäumler, & Schmidt, 2014;Lucke, Schmidt, Bens, & Hüttl, 2005;Schmidt et al., 2006), several soil classification and mapping campaigns were executed in the country to support soil management and land-use planning at the national level (see summary by Lucke, Ziadat, & Taimeh, 2013). The first significant work in this field started in 1950 and recognized 12 great soil groups at a scale of 1:1,000,000 (Moormann, 1959). The resulting report applied older versions of the US Soil Taxonomy (USDA, 1938), with special reference to soil colour. Another detailed soil survey in the 1970s was restricted to the regions around Irbid (NW Jordan) and Karak (W Jordan) (Al-Qudah, 2001). However, the only systematic, nationwide soil survey and land classification was conducted in 1989-1993, resulting in soil maps at three different levels of detail (NSM and LUP, 1993). The initial survey of this project was a reconnaissance study using the visual analysis of Landsat and aerial photography (Al-Qudah, 2001). Due to poor information on spectral characteristics of the soils, there was no possibility to seriously re-evaluate previous soil mappings or to support spatially oriented soil studies in Jordan with newly available remote sensing data. Consequently, there is a need for ground-based imaging spectroscopy to study soil spectral reflectance features, in order to understand their relation to soil properties and as a reference for the application of remote sensing in soil mapping of Jordan (Chabrillat et al., 2019;Huete, 1996;Viscarra Rossel et al., 2016).
This study aims to: • provide a basic characterization of spectral reflectance curve behaviours of soils in northern Jordan, focusing on soil colour and geochemical components; • investigate the effect of Fe oxides on the spectral curve behaviour; and • contribute to the objective of establishing a spectral library of soils in Jordan and to include it in the global spectral library of the world's soil.
Morphological characteristics of the spectra were combined with statistical analysis. Soil reflectance spectra permitted a qualitative and statistical grouping of the soils in terms of spectral reflectance values (SRV) and colour.

| MATERIALS AND METHODS
The study area is situated in northern Jordan (longitude 35.90 to 36.29 E; latitude 32.34 to 32.52 N) and covers 693.77 km 2 . The elevation ranges between 1,020 (in the SW) and 500 m a.s.l (in the N and NE). The main soil types in the area are Chromoxererts, Calcixerollic and Xerochreptic Calciorthids (MOA, 1995). In older classifications, these soils belong to the groups of the Red and Yellow Mediterranean soils (Moormann, 1959) (Figure 1). They cover mainly flat to undulating landscapes. These include the Quaternary plain (fluvial deposits, sand and loess) of the Irbid-Ramtha basin in the northwest, and the dissected limestone plateau of Ajloun in the southwest. Hilly terrain in the central part consists of Tertiary calcareous rocks. Between the hilly terrain and the plain of Irbid, a considerable area is covered by alluvial fans from the limestone plateau. To the east, an undulating to rolling topography and a flat to gently sloping lava plateau are present. According to the long-term records of the average annual rainfall, precipitation falls from more than 600 mm in the west to less than 100 mm in the east. Soil colours change from dark red and brown in the west to lighter colours (more or less yellow) towards the east (Moormann, F I G U R E 1 Study area, sample locations (labelling names of samples mentioned in text); soil groups from this study according to RI LAB-BT ; old USDA soil types referring to colour are given in the background of the map (modified after Moorman, 1959). R= red; M= Mediterranean; S= soils; Y= yellow. RI LAB-BT : redness index based on CIELAB colour system after Barrón and Torrent (1986) [Color figure can be viewed at wileyonlinelibrary.com] 1959; Sahwan, Lucke, Kappas, & Bäumler, 2018). Lucke and Sprafke (2015) suggested that the haematite content increases towards the west, following the current climatic gradient and leading to redder and darker soil colours.

| Soil sampling
The soil sampling sites were located along the different landscape units as well as on transects of soil colour (40 locations covering yellow to red soil colours; see Moormann, 1959;Lucke & Sprafke, 2015;Sahwan et al., 2018). Every site was georeferenced using a global positioning system (GPS). The samples were taken from a depth of 0-10 cm, mainly on fields used as arable land.
A systematic sample collection with 30-m squares was applied in each location. The samples are from four corners of these squares (labeled A, B, C and D). The collected material (152 samples) was stored in freezer bags. Additional control samples (mixture of soil and quarry dust/aggregates) were taken from a limestone quarry and from the top of the highest archaeological tell in the area (Tell Khanasreh). On this hill, a thin soil cover apparently represents settled dust (Lucke et al., 2019). To minimize the effects of soil moisture and particle size on the spectral measurements, all samples were sieved to ≤2 mm and oven-dried at 40 C for 72 hr. Equal quantities of the samples (A, B, C and D) from each location were mixed in the laboratory. The produced mixtures were used to define the chemical components. A detailed list of sampling locations, Munsell dry colours and general description of the geomorphology is provided in Appendix 1, as well as descriptive statistics of selected chemical parameters of the soil samples in Appendix 2.

| Spectral measurement and preprocessing
Laboratory soil reflectance spectra were obtained at the Department of Land Surface Dynamics in DLR (German Aerospace Center) using an analytical spectral device (ASD Field Spec® 3 Hi-Res). The ASD measures reflectance in the VIS-NIR and SWIR (0.4-2.5 μm) spectral range, with a spectral resolution of 3 nm in the 0.4-1.0 μm range and of 10-12 nm in the 1.0-2.5 μm range. A high-intensity contact probe (HICP) with a built-in light source (6.5 W halogen lamp) and a measurement surface area equivalent to a 10-mm diameter circle was used to acquire the soil spectra. The HICP was applied to minimize errors associated with stray light. Each sample was placed in a black petri dish and subsequently scanned four times. The sensor was calibrated with a Spectralon panel as white reference once for every sample.
Possible steps in the spectrum were eliminated using the splice correction function of the ASD ViewSpec Pro software. The mean soil reflectance spectra (MSRSs) of each location were calculated from 16 measurements (four samples of each location and four-times repeated measurement). All MSRSs were normalized using the smoothing filter function sgolayfilt in R (Savitzky & Golay, 1964), which reduces high-frequency noise without demeaning the shape of the original spectra (Sellitto, Fernandes, Barrón, & Colombo, 2009). Furthermore, the normalization procedure of continuum removal (CR) was applied to evaluate the absorption features (Clark et al., 2003. This approach has been used successfully in soil science and mineralogy (e.g., Clark et al., 2003;Gomez, Lagacherie, & Coulouma, 2008;Lagacherie, Baret, Feret, Netto, & Robbez-Masson, 2008;Viscarra Rossel, Cattle, Ortega, & Fouad, 2009). In our work, the calculation of CR was achieved using the DIS-PEC software in ENVI. The spectrum was normalized by setting the hull value to 100% reflection (Clark & Roush, 1984). Two methods can be used to normalize the spectrum: subtraction (additive) or division (multiplicative). The additive technique of CR was applied in our study (Clark et al., 2003).

| Soil colour and chemical measurements
A ColorLite spectrophotometer sph850 (Katlenburg-Lindau, Germany) was used to measure the soil colours of the oven-dried fine earth. The measuring field of the spectrophotometer has a diameter of 3.5 mm; the light source is D65 and the observer angle 10 . The spectral resolution is 3.5 nm in a range from 400 to 700 nm. The measurement values are the average from two times three replicates with stirring of soil material in between the measurements. The ColorDaTra 1.0.181.5912 software was used to transfer the spectral values and colour parameters into an Excel file, in which the redness index RI LAB-BT of Barrón and Torrent (1986) was calculated. This parameter uses the L* (luminance; black (0), white (100)), a* (intensity of green (<0) or red (>0) component) and b* (intensity of blue (<0) or yellow (>0) component) values of the CIELAB colour space, which is widely used in soil science (Lucke & Sprafke, 2015;Sprafke, 2016;Viscarra Rossel et al., 2009): Selected analytical properties and redness index, RI LAB-BT , after CIELAB according to Barrón and Torrent (1986) For comparison and general soil description, soil colours were determined with the Munsell Soil Colour chart (Munsell Color Company, 1994) under constant laboratory light by one observer only. This study uses colour measurements of the CIELAB system for statistical and cluster analysis as it is more suitable than the descriptions of Munsell (Barrón & Torrent, 1986;Melville & Atkinson, 1985;Viscarra Rossel et al., 2009). The pH value was determined in distilled water with a glass electrode (pH-meter 530 by WTW, with electrode InLab 423 by Mettler-Toledo) in a soil:water solution of 1:2.5. Contents of soil organic carbon SOC (C org ) that is directly related to soil organic matter (SOM) contents and contents of calcium carbonate (CaCO 3 ) were determined using an Elementar vario EL cube C/N-analyser, measuring samples twice before and after ignition for 2 hr at 500 C based on the assumptions that organic matter is ignited and all remaining carbon is inorganic carbon. The free iron oxides, Fe d (pedogenic oxides), were extracted with sodium dithionite at room temperature according to Holmgren (Schlichting, Blume, & Stahr, 1995). Fe d quantifies the presence of Fe 3+ minerals in soils and is an indicator for goethite and haematite (Parfitt & Childs, 1988). For total element analysis and calculation of total iron content (Fe t ), samples were pulverized in an agate ring mill, type Retsch RS 200. Then the loss on ignition was determined by weighing the powder before and after drying for (a) 12 hr at 105 C in a cabinet dryer and (b) 12 hr at 1,030 C in a muffle furnace. Major element oxides such as CaO (used in this study) were measured with an energy-dispersive XRF (Spectro XEPOS plus) at the GeoZentrum Nordbayern (Erlangen, Germany). Precision and accuracy are generally higher than 0.9% (main elements) and 5% (trace elements).

| Statistical analysis
We applied simple linear regression using the R-function lm (R Core Team, 2018) to examine the relationship between different variables. In particular, we evaluated the performance of reflectance values of MSRSs in all spectral wavelength ranges (SWR) as an independent variable to predict RI LAB-BT , Fe d and C org . Furthermore, we assessed the relationships between the redness index RI LAB-BT and chemical soil components such as C org , CaO (calculated from total elements measured by XRF) and Fe d . The linear model was also used to determine the slopes of MSRSs in the ranges 450-600 nm (SCS450) and 600-750 nm (SCS600). A non-linear model (exponential) evaluated the relationship of the coefficients in these two cases (SCS450 and SCS600) with the RI LAB-BT . To classify the samples according to their L*, a* and b* values and reflectance values, the agglomerative hierarchical cluster analysis with Euclidean distance and complete linkage was applied using the R-function hclust (Maechler, Rousseeuw, Struyf, Hubert, & Hornik, 2019). The results are presented as dendrograms, which were drawn with the R-function as.dendrogram (R Core Team, 2018).

| Soil chemical and colour properties
The samples represent alkaline soils with pH w from 7.7 to 9.4. The calcium carbonate contents vary widely between 1 and 48%. Also Fe d (4.1-20.4 mg/g), as well as the total iron content Fe t (1.1-6.1%), shows considerable variation. Most of the soils contain less than 1.9% organic carbon (C org ). In the sample from the area of the limestone quarry, the calcium carbonate content is 78%. The RI LAB-BT values range from 1.68 to 48.58 (Table 1) (Figure 2). The high RI LAB-BT of some samples can probably be attributed to high haematite contents (Barrón & Torrent, 1986;Lucke & Sprafke, 2015). Therefore, it seems likely that the soil of light brown or yellow colours with low RI LAB-BT indicates the minor presence of haematite (Table 1 and Figure 2).

| Spectra of surface soils
The mean soil reflectance spectra MSRSs (Figure 3) have approximately the same behaviour and rather comparable absorption features in the region of hygroscopic water (SWIR: 1400, 1900 nm) and in the region of clay minerals/ F I G U R E 3 (a) Representative spectra of selected soil samples and (b) illustration of morphological conditions of curves organic matter (SWIR > 2,200 nm) (Baumgardner et al., 1986;Ben-Dor et al., 2008). However, they are characterized by decreasing slope steepness in the VIS range 350-780 nm with sigmoidal form, followed by slight absorption features from 780 to 980 nm (NIR) (Figure 3). Accordingly, all MSRSs represent types of spectral soil curves that are affected by Fe oxides (Bowker, Davis, Myrick, Stacy, & Jones, 1985;Condit, 1970;Escadafal & Huete, 1992;Stoner & Baumgardner, 1981). The effects of Fe oxides on spectra can be illustrated using four morphological distinctions (Demattę, Bellinaso, Romero, & Fongaro, 2014;Poppiel et al., 2018), as in Figure 3: 1 reflectance intensity value (RIV); 2 distance between concave sigmoid and y-axis (DCS); 3 steepness of convex sigmoid (SCS); and 4 iron absorption feature (IAF).
Even though the RIV represents a cumulative condition of all soil properties in the entire spectral range, it may show an overall decrease when the Fe content increases, especially if the Fe oxides occur predominantly in the form of haematite (Demattę et al., 2014). In contrast to RIV, the DCS, SCS and IAF elucidate the effects of Fe oxides and SOM in the visible and near-infrared range  (VIS-NIR, 400-1,100 nm). This also depends on the effects of soil moisture and particle size, which are minimized by oven drying and sieving the samples in this study (Escadafal & Huete, 1992;Hoffer, 1978;Hunt, 1980). The continuum removal (CR) manifests the slight absorption features in the range 780-980 nm. Furthermore, it shows strong absorption features in the range from 350 nm to approximately 700 nm, which was not observable in the original spectra (Figure 4) (Table 2). In these ranges the deepest absorption features show diverse shifts, which can be related to the presence of different Fe minerals (Hunt, 1977;Hunt & Ashley, 1979).

| Effects of chemical components on soil spectra
The soil spectra are from the type of soil affected by Fe oxides (Stoner & Baumgardner, 1981). However, the effect of SOM (or SOC) can mask the effects of Fe oxides and cause a smoothing of their absorption features (Baumgardner et al., 1986;Ben-Dor, 2002;Viscarra Rossel et al., 2016). This masking is noticeable solely in the sample from the summit of Tell Khanasri (Kh-Tell-01, C org 2.7% and Fe d 5.6 mg/g) and in the sample Jar-St-02 (C org 2.6% and Fe d 17.1 mg/g) ( Figure 5). In contrast to these two samples, the contents of C org are less than 2.1% in all other samples (Table 1). According to Baumgardner, Kristof, Johannsen, and Zachary (1969), the masking effects become less effective if the SOM contents drop below 2.0%. The statistical analysis using linear models leads to the same result. In this context, the correlations of reflectance values of MSRSs in all spectral wavelength ranges (SWR) with Fe d and C org were investigated. The coefficient of determination (R 2 ) was fairly low (lower than 0.03, p > .5) in the case of C org (R 2 average value = 0.013). In contrast, it was considerable in the case of Fe d , for which the average value of R 2 was 0.53.
Concerning the above-mentioned conditions, the effects of Fe oxides on MSRS properties dominate, whereas the effects of SOC are generally negligible (Figures 3 and  5). In this regard, the RIV of MSRSs increases from dark red-brown soils to yellow soils. The samples of low RIV have a high content of Fe as well as a low content of CaCO 3 (or CaO) and vice-versa ( Figure 3 and Table 1). Moreover, the samples with high RIV are closer to the RIV of the sample from the limestone quarry and show considerable absorption in the range of the calcite effect (Sama-02 in Figure 3). Some samples (e.g., IR-E-02 and Dujaina-02) have diverse RIV, despite comparable contents of Fe and CaCO 3 (Figure 3). This could be the result of increasing haematite contents in comparison to goethite and calcite (Cañasveras Sánchez, Barrón, Del Campillo, & Viscarra Rossel, 2012;Curcio, Ciraolo, D'Asaro, & Minacapilli, 2013;Fernandes, Barrón, Torrent, & Fontes, 2004). It is consistent with the trend of RI LAB-BT values, which are high in IR-E-02 and low in Dujaina-02 (Table 1).

T A B L E 3
The spectral ranges with R 2 values (Fe d content -RVs) higher than average In general, soils rich in CaCO 3 and relatively poor in Fe show higher reflectance than soil with less or negligible CaCO 3 content, or with high Fe content (Gaur, Pandey, & Goyal, 2016). The morphological conditions DCS, SCS and IAF show the following trends: the DCS decreases from red soils to yellow soils, and the steepness (SCS) changes from relatively low (red soils) to high (yellow soils). Taking into consideration the negligible effects of SOC, the low SCS is related to an increased proportion of haematite in comparison to goethite (Demattę et al., 2014;Poppiel et al., 2018). In our study, this finding is considerable because of the relationship between the morphological condition SCS and RI LAB-BT , which can be used as an indicator of the proportion of haematite/goethite (Barrón & Torrent, 1986). This relation was statistically highlighted using the linear correlation between RI LAB-BT and SCS in the ranges of 450-600 nm (SCS 450 ) and 600-750 nm (SCS 600 ). To facilitate this correlation, the slopes of MSRSs in the ranges SCS 450 and SCS 600 were extracted using linear regression. The coefficients of these two cases (SCS 450 and SCS 600 ) show strong exponential correlations with the RI LAB-BT ( Figure 6). The R-squared values were 0.95 (p < 0.001) and 0.86 (p < 0.001), respectively. Furthermore, the SCS 450 and SCS 600 show a considerable linear correlation with Fe d (R 2 = 0.60, p < 0.001). In contrast, there is no significant relationship with other chemical components (e.g., with C org : R 2 = 0.004, p > .5).
According to the statistical analysis of linear correlation between reflectance values of MSRSs in all spectral wavelength ranges (SWR) and Fe d , the highest R 2 value (0.60, p < 0.001) is recorded in the spectral range of 734-794 nm. Table 3 shows that the coefficients of determination (R 2 ) are higher than average (R 2 = 0.54) in several SWRs, which correspond to the absorption features of Fe oxides in the form of haematite or goethite (Grove, Hook, & Paylor, 1992;Hunt & Ashley, 1979;Morris et al. 1985). According to Grove et al. (1992), for example, haematite shows absorption maxima at 535 nm and goethite at 503 nm.
In this regard, the continuum removal (CR) curves evaluate the possible presence of haematite or goethite. For example, the deepest absorption demonstrates a shift from 479 nm (Dujaina-01) to 424 nm (Sama-02) in the visible range (Figure 4). Further shifting occurs in the range of near infrared from 895 nm (Dujaina-01) to 920 nm (Sama-02). These two shifts point to a decrease of haematite and increase of goethite from Dujaina-01 to Sama-02 (Hunt, 1977;Hunt and Ashley, 1979;Morris et al., 1985;Vilas, Jarvis, & Gaffey, 1994). As a general trend, there is a relatively high presence of haematite in Comparison of the cluster ranking sorted by redness index, RI LAB-BT , according to Barrón and Torrent (1986)  samples from western parts of the study area and a relatively low presence of haematite in samples from eastern parts. This result is consistent with studies on surface soil weathering intensity in the context of the current climatic gradients in this area (Lucke & Sprafke, 2015;Sahwan et al., 2018).

| Correlation of soil colours with chemical components
Considering the relatively low contents of C org (SOC), the Fe oxides (probably haematite/goethite) are the dominant pigmenting agents. This suggests that the redness index RI LAB-BT is an adequate indicator of haematite and goethite contents in soils, as proposed by Barrón and Torrent (1986). The relationships between RI LAB-BT and chemical soil components such as C org , CaO (calculated using XRF) and Fe d were insignificant, significant and highly significant, respectively (Figure 7). This indicates a negligible effect of SOC on the colours of the samples. The low correlation with the CaO content indicates that calcium carbonate does not play a decisive role. In contrast, the considerable correlation with Fe d (R 2 = 0.58, p < 0.001) reflects the effect of Fe oxides on soil colour.

| Soil grouping
Depending on the morphological conditions, the mean soil reflectance spectra MSRSs were arranged qualitatively in six groups: Group I to Group VI (Table 4). Figure 3 illustrates representative reflectance curves of these groups. A hierarchical cluster analysis of all MSRSs reflectance values and soil colour values (L*, a* and b*) was conducted (Table 2 and Figure 8). This qualitative grouping and clustering of spectral values categorized all samples to matching groups with the exception of some outliers, which showed strongly divergent properties (Table 4). The ranking corresponds to the decrease of haematite effects and increase of CaCO 3 effects from the first group (Group I) to the sixth group (Group VI). It more or less reflects a categorization of RI LAB-BT values in the following groups: RI LAB-BT > 30, 30-20, 20-10, 10-4, 4-2 and Compared with the map of Moormann (1959), the samples of Group I, Group II and Group III are mostly located in the areas of Red Mediterranean Soils, and Red Mediterranean Soils and Lithosols (see Figure 1). A few samples, such as MBH-03 and MBH-02 (Group II), are located in the area of Yellow Mediterranean Soils and Basalt. These samples show high effects of haematite on the spectral curves and colours. Another deviation in classification compared to the map of Moormann is the case of sample Kr-Er-02 in the transitional zone between Red Mediterranean Soils and Yellow Mediterranean Soils. This area is the geomorphologic boundary between the hilly terrain of Tertiary calcareous rocks in the central part and the level area to the west, where water surface flow and limestone debris deposition processes are active. Due to these local geomorphic influences, this sample has relatively low haematite and high CaCO 3 contents. In the clustering of this study, this sample is classified in Group V, which has a low value of RI LAB-BT and shows minor effects of Fe oxides. Such local variation cannot be detected on large-scale maps. Their mapping and monitoring need a precise evaluation of remotely sensed satellite data. However, such studies require a site-specific spectral library, which can be used as ground truthing data to facilitate the use of current and future multispectral and hyperspectral satellite data. In this regard, spectra from spectral libraries facilitate a highly precise classification of remotely sensed satellite data, in order to evaluate the spatial distribution of red soils in the area (Sahwan et al., 2018).

| CONCLUSIONS
This study combined the evaluation of laboratory reflectance spectra of surface soils in northern Jordan with systematic ground observations and laboratory measurements of soil properties. A significant amount of surface soil samples that represent an initial database of a soil spectral library of Jordan was collected. The extracted representative MSRSs of this study confirm that spectra of soils of similar landscape setting in northern Jordan depend on the iron oxides. This refers mainly to the pigmenting effects of haematite and goethite, which seem closely connected with the intensity of in-situ pedogenesis of surface soils in the study area, mirroring the current climatic gradients. To a more limited degree, contents of calcium carbonate play a role, too, in particular with regard to soil brightness. Our study discriminated morphological conditions of spectral curves that permit distinguishing between these spectra. Statistically, the study demonstrated a clear relationship between spectra, chemical components and colour characteristics of the soils. It shows that the significant contrast between the spectra results from Fe oxide effects connected to the proportion of hematite and goethite, which are the dominant soil colour agents.
Even though the spectra show partial similarity, they can be classified into six spectral groups, which highly coincide with soil colour groups. Consequently, our results can be used as a basis to assess soil colour with multispectral or hyperspectral data and thereby analyse the chemical and mineralogical properties of soils in Jordan. Such studies are essential to avoid unreliable classification of remote sensing data because of behaviour similarity of spectral curves. In summary, spectral characteristics can provide an alternative method to the traditional and highly expensive methods for assessing soil properties in Jordan. However, further studies on the prediction of constituents such as iron, organic matter and clay fraction should be conducted in order to enlarge the available data and their robustness, which will facilitate soil surveys and management practices in Jordan.