Abstract
As a means of assisting the selection of promising soil classification systems, a set of criteria were presented and tested. Inside the studied slightly saline plot World Reference Base (WRB) and Hungarian soil classification (HU) were compared at all four levels in terms of class separability, correlation to biomass, parsimony and homogeneity of classes. WRB surpassed HU in terms of the very important homogeneity of classes only, but HU performed better in terms of class separability, correlation to biomass and parsimony of classes. With many possible classification units WRB categorized the soil into a large number of classes, but 67% and 78% of them were single-profile classes at levels 3 and 4, respectively inside the ca 0.9 km2 area.
Introduction
Soil classification has not lost its relevance for modern soil research and practice, because soil classes provide a summary of many soil features (KUBIËNA, 1953). But not like in other disciplines, in soil science there are many classification systems coexisting (KRASILNIKOV et al., 2010) influenced by tradition, legal actions and other reasons.
The most widespread classification systems, USDA Soil Taxonomy and the World Reference Base for Soil Resources (WRB) (ROSSITER et al., 2017; ESFANDIARPOUR et al., 2018; SALEHI, 2018) were compared according to parent material (SOROKIN et al., 2021), classification levels, physical and chemical properties, and other features by many researchers. SHRADER et al. (1960) WEBSTER et al. (1977), ALLGOOD & GRAY (1978), OGUNKULE & BECKETT (1988), BUOL et al. (2011) studied the utility of soil classification systems for predicting selected properties and productivity, and our work follows this tradition.
We used Normalized Difference Vegetation Index (NDVI) [the ratio between the near-infrared and red reflectance difference and the sum of the same two parameters] as a universally applied remote sensing indicator/proxy of aboveground biomass (MCBRATNEY et al., 2003; TEAL et al., 2006; PETTORELLI, 2013).
We tested two soil classification systems in the present study. The Hungarian soil classification system, (HU) is a genetic and hierarchical classification system that was developed in the 1960s (SZABOLCS, 1966) and was updated later (JASSÓ et al., 1989) for mapping soils at a detailed scale and it is currently used on maps at scales of 1:10,000 to 1:1,000,000. HU has four levels, such as main type, type, subtype, variety, but it does not have a taxonomic key.
The World Reference Base for Soil Resources (WRB, 2015) does not have a declared hierarchy, but has several hierarchical levels. Its use is promoted by FAO as an "international classification" and it is also suggested reference inside the European Union (TÓTH et al., 2008). Nevertheless, its use is more common at less detailed scales, such as 1:1,000,000, and it is now being introduced for mapping of smaller areas (SCHULER et al., 2006). Its use is facilitated by a key which is based on diagnostic horizons and other features.
Our paper shows how classifications can be compared with the use of four practical criteria, such as class separability, class homogeneity, correlation with environmental parameters, and parsimony of classes, ranked in their order of importance. Such comparisons can help to select optimal classification to be used in an area. The work was motivated by the recent dispute on a renewed Hungarian soil classification that was suggested by MICHÉLI et al. (2018) and the subsequent debate articles of BIDLÓ (2019), MAKÓ (2019) and TÓTH (2019a). In his debate article, TÓTH (2019b) wrote "My specific suggestion for authors is to map appropriate sample areas based on current Hungarian soil classification and the suggested approach, using both classifications. With the map, predict the most important soil ecosystem services, and then quantify the benefits of the suggested approach by comparing these with the services determined by an independent method." This current paper shows a possible method to do what was suggested in 2019.
This report extends the depth of analysis of the TÓTH et al. (2022) publication for the Hungarian readership by providing more details of the WRB and Hungarian classifications, which are of great importance for the area. Data presented in Figures 5, 7, 11–13 show overlap with the mentioned paper.
Materials and Methods
The study arable plot used for growing cereals ( Figure 1 ) is located on the outskirts of the village of Dunavecse, on the former floodplain of the Danube River. The soils are slightly saline and have a sandy-loamy texture, with increasing average particle size along the depth of the profile. The water table is shallow and saline. Local depressions, formerly densely vegetated, are characterized by a higher fraction of silt, organic matter and salt content and a lower concentration of carbonates. Elevation dominates most soil properties. With increasing elevation, the mean salt concentration, pH, sodicity, clay content and organic carbon content decreases, as well as CaCO3 content increases in the thickness of 0 to 100 cm. The climate is semi-humid with annual temperature of ca 10.5°C.
A 4 cm resolution digital elevation model was obtained with UAV surveys using ground control points of known coordinates. NDVI values were calculated with NASA Landsat data. Annual maximum values between 2010 and 2019 were averaged for the 85 profiles, while NDVI ranges showed the difference between maximum and minimum values for the years considered.
The selected plot of 0.9 km2 is rectangular (corner coordinates: (46o 55' 16 " N 19o 01' 37" E, 46o 55' 17" N 19o 02' 12" E, 46o 55' 55" N 19o 01' 41" E, 46o 55' 49" N 19o 02' 12" E). Within the plot 85 tubular profiles of 1 m depth were taken ( Figure 1 ) which were described according to SZABOLCS (1966). There were also four digged profiles that were described, sampled and analysed. Parameters used for classification were obtained by analysing one-third of the samples, while others were estimated using measured morphological and instrumental (X-ray fluorescence spectroscopy analysis, EC and soil moisture) data, including EC, pH, Na, SOC, CaCO3 content, hygroscopicity ( Table 1 , columns 9–14) with multivariate regression equations; after which profiles were classified according to WRB and HU independently in multiple iterations.
The WRB classification was performed at four levels ( Table 1 , column 5). Reference soil groups (RSG) were determined and all possible qualifiers were added. The number of applicable principal ("princ" below) and supplementary ("suppl" below) qualifiers ranged from 1 to 4, and 1 to 5, respectively. The number of all qualifiers ranged from 2 to 6. Despite its non-hierarchical structure, soil classification was performed at levels 1, 2, 3 and 4. As our main objective was soil productivity assessment, qualifiers were added according to the fixed order of nomenclature of the WRB principles. Optional qualifiers were added according to the following approach:
-
- WRB1 level RSG
-
- WRB2 level RSG+1princ
-
- WRB3 level RSG+2princ or RSG+1princ+1suppl
-
- WRB4 level RSG+3princ or RSG+2princ+1suppl or RSG+1princ+2suppl
The application of qualifiers according to the above principle is in harmony with the principles of WRB name generation (WRB, 2015, p. 14–15, 3d–5.#) used for soil mapping. Qualifiers that cannot be directly associated with yield (supplementary texture qualifiers) were ignored.
All HU levels were used for classification ( Table 1 , column 6) and then the soil evaluation index proposed by IZSÓ (1986) was determined.
Evidently the classes ( Table 1 ) reflect the rules of both classification systems and where there was for example alluvial/hydromorphic feature noticed and expressed in HU it often was not expressed in WRB due to the strict limits of WRB regarding strength of the feature, depth of occurrence and thickness of relevant layer.
Classification of the soil profiles applying Hungarian Soil Classification (HU) and World Reference Base for Soil Resources (WRB) at four levels and important average values of profiles as ten-year-average NDVI values, range of NDVI values during the ten years (R_NDVI), 0–0.3 m pH2.5, Na2.5, CaCO3 %, Soil organic carbon % (SOC), hygroscopicity (%), ECe (µS cm–1) . Classification first level is indicated by CAPITAL, second level by Italic, third level by underscoring and fourth level by superscript. If not more than three levels are indicated, fourth level is same as third level for WRB. The digits following class names indicate the codes that identify the classes of Figures 4–10. Thousands show first level, hundreds second level, tens third level and ones fourth level of classification, but several combined features result in more than four digits in HU at the fourth level.
Layout of profiles inside the plot. Graph is not to scale. Standard distance of profiles in rows and columns was 100 m, except for A1, A2, F8 and Y1, see Table 1 for coordinates (sunfleck of H3 and missing sample in a few profiles, F6, I6 etc are artefacts due to difficult sampling/photography)
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Layout of profiles inside the plot. Graph is not to scale. Standard distance of profiles in rows and columns was 100 m, except for A1, A2, F8 and Y1, see Table 1 for coordinates (sunfleck of H3 and missing sample in a few profiles, F6, I6 etc are artefacts due to difficult sampling/photography)
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Layout of profiles inside the plot. Graph is not to scale. Standard distance of profiles in rows and columns was 100 m, except for A1, A2, F8 and Y1, see Table 1 for coordinates (sunfleck of H3 and missing sample in a few profiles, F6, I6 etc are artefacts due to difficult sampling/photography)
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Scattergram of soil evaluation index and mean NDVI values
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Scattergram of soil evaluation index and mean NDVI values
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Scattergram of soil evaluation index and mean NDVI values
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
The class separability was assessed by the number of classes that showed significant differences in NDVI values pairwise, using ANOVA. The larger the ratio of significantly different class pairs, the better is the classification.
Class homogeneity was analysed for mean NDVI and, as a reference, for elevation, at the four classification levels by the value of “1-RV”, where RV is the pooled within-class variance/total variance. The higher this values, the more precise is the classification.
Parsimony of classes was determined by the number of distinguished classes, with special consideration of single-profile classes. Greater number of classes, especially single-profile classes might cause difficulties for mapping (VAN HUYSSTEEN et al., 2013).
Correlation of classes to environmental parameters was tested by calculating Pearson correlation between mean NDVI and elevation values of distinguished classes at all four levels. Stronger correlation means easier use for predicting productivity.
A series of boxplots shows the NDVI and elevation values of distinguished classes. Width of boxes indicates number of cases in the class.
Results and Discussion
As Figure 1 and Table 1 show there was great lateral and depth variation of soil morphology as well as quantitative soil properties, but the spatial distribution of properties will be described in another publication in detail. With increasing depth, SOC% and clayiness decreased, but salinity related properties and CaCO3% increased. The variation of CaCO3 was much less in the full one-meter profile than in the 0–30 layer, but clayiness was more heterogeneous in the full profile length.
Although the soil evaluation index did not highly correlate with mean NDVI, it still indicated a significant correlation of r = 0.231* ( Table 2, Figure 2 ). This finding has corroborated the findings of TÓTH et al. (2009) who found moderate performance of this index for yield evaluation. The stronger negative correlations with the range of NDVI and salinity indicate the profound base and suitability of the approach.
Correlation coefficient of the major variables with soil evaluation index (n = 85)
| Soil evaluation index | ||
|---|---|---|
| 10 years average NDVI value | Pearson Correlation | 0.231* |
| Sig. (2-tailed) | 0.033 | |
| 10 years NDVI range | Pearson Correlation | –0.255* |
| Sig. (2-tailed) | 0.019 | |
| Elevation above sea level (m) | Pearson Correlation | 0.140 |
| Sig. (2-tailed) | 0.201 | |
| ECe (0–30 cm) µS cm–1 | Pearson Correlation | –0.334** |
| Sig. (2-tailed) | 0.002 | |
| ECe (0–100 cm) µS cm–1 | Pearson Correlation | –0.588** |
| Sig. (2-tailed) | 0.000 |
Mean values of NDVI (top) and elevation (bottom) at level 1 of WRB (Reference soil groups). See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean values of NDVI (top) and elevation (bottom) at level 1 of WRB (Reference soil groups). See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean values of NDVI (top) and elevation (bottom) at level 1 of WRB (Reference soil groups). See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Evaluation of the two classification systems in terms of NDVI and elevation
The highest NDVI values can be attributed to the Chernozem, Kastanozem, and Phaeozem reference groups, but these three reference groups do not differ significantly ( Figure 3 ). The NDVI values of the Calcisol and Regosol RSGs were the lowest, however, the latter had only one profile, so the difference could not be interpreted statistically. According to the elevation, the Gleysols are in the lowest and the Kastanozem and Chernozem are separated at the highest position, but the other reference soil groups are not clearly differentiated. The profiles classified according to the RSG therefore do not represent homogeneous and not clearly distinct groups according to either NDVI or elevation. The distribution of NDVI values and the elevation of the RSGs was broadly similar.
Mean values of NDVI (top) and elevation (bottom) at level 2 of WRB. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean values of NDVI (top) and elevation (bottom) at level 2 of WRB. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean values of NDVI (top) and elevation (bottom) at level 2 of WRB. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
The highest mean NDVI value was shown by Endocalcic, Amphicalcic and Pantocalcaric grade Chernozems, Kastanozems and Phaeozems, but same classes showed also low NDVI values for some profiles ( Figure 4 ). This is a good indication that the homogeneity of classes obtained by a second-level classification of a reference group and a qualifier is low and significantly dispersed according to NDVI. In some cases, the second classification level is well differentiated within the RSG based on NDVI, such as between Amphicalcic and Endocalcic Kastanozems and Chernic and Pantocalcaric Phaeozems. Based on the qualifiers, there is no such distinction in the value of NDVI or elevation inside Chernozems.
The profiles at highest elevations have been classified as Endocalic and Amphicalcic at the second level, and Haplic and Chernic at the lowest elevations, but these properties cannot really be related to their topographic position. The distribution of NDVI values by second level classes is only broadly similar to the distribution of elevation.
In classes containing higher number of profiles, the lowest NDVI values were showed by classes with Alcalic, Gleyic and Protosalic qualifiers based on the second qualifier added at the third level of the classification, almost independently of RSG and first qualifier, which are associated with poorer productivity ( Figure 5 ). Third-level qualifiers (Cambic, Endoprotosalic, Endofluvic) do not clearly indicate favourable soil conditions. The homogeneity of the classes is low according to the elevation, and the standard deviation of the elevation values is large even within the third classification level. The added qualifiers of the profiles in lowest elevation at the third level are varied (Endogleyic, Amphigleyic, Endoprotosalic, Katoprotosalic, Katofluvic), but partly refer to the low topographic position; this cannot be stated for the profiles at highest elevation (Cambic, Katofluvic, Amphicalcic). At this level, the distribution of classes by NDVI and elevation showed no similarity. The statistical evaluation of the differences is complicated by the fact that the number of different classes increases remarkable with the level of classification, so the number of single-profile classes increased as well.
In many cases, there were no additional added classifiers at the fourth level, so they are identical with the third level classification (see Table 1 for details). For the profiles showing highest NDVI values, Cambic is added as a fourth-level qualifier, which cannot be causally related to the higher NDVI value, while the profiles with the lowest NDVI value were classified as Endosalic, Endoprotosalic, or Katoprotosalic at the fourth level ( Figure 6 ). Here, low NDVI is associated with salt accumulation in the profile. These profiles are simultaneously located in the lowest topographic position, so the distribution of NDVI and elevation is similar in this relationship, but this is not typical for the other qualifiers added at the fourth level. The heterogeneity of the individual classes and the standard deviation of the values in terms of NDVI and elevation are also typically highest where the fourth level was identical with the third classification level, i.e., no further qualifier could be given.
The NDVI values and elevation values of Chernozem and Meadow main soil types appear to be well separated (significantly different) at the first classification level ( Figure 7 ). The NDVI and elevation values for Alluvial soils fall between the two previous groups and are not significantly different. In general, higher elevation values are associated with higher productivity values, presumably because at higher elevations productivity is not inhibited by damaging surplus water. The thickness of the boxplots also clearly shows the relative number of soil profiles belonging to each main soil type, the sample contains mostly Chernozem profiles and few Alluvial soil profiles.
Mean values of NDVI (top) and elevation (bottom) at level 3 of WRB*
*Codes and names of the classes shown in the graphs from left to right are the following 100-Amphiprotosalic Endogleyic Regosol, 210-Pantocalcaric Katofluvic Cambisol, 220-Pantocalcaric Cambisol (Endoprotosalic), 300-Amphicalcic Chernic Gleysol, 301-Endocalcic Chernic Gleysol, 410-Haplic Calcisol (Alcalic), 421-Cambic Calcisol (Katofluvic), 422-Cambic Calcisol (Endogleyic), 423-Cambic Calcisol (Protosodic), 510-Amphigleyic Phaezoem (Protosodic), 520-Chernic Phaeozem, 521-Pantocalcaric Chernic Phaeozem, 522-Katofluvic Chernic Phaeozem, 523-Amphifluvic Chernic Phaeozem, 524-Endofluvic Chernic Phaeozem, 525-Amphigleyic Chernic Phaeozem, 530-Pantocalcaric Phaeozem, 531-Pantocalcaric Phaeozem (Alcalic), 610-Endogleyic Amphicalcic Kastanozem, 611-Amphicalcic Kastanozem (Endoprotosalic), 622-Endocalcic Kastanozem (Cambic), 623-Endogleyic Endocalcic Kastanozem, 624-Amphigleyic Epicalcic Kastanozem, 710-Haplic Chernozem, 711-Haplic Chernozem (Pachic), 712-Haplic Chernozem (Katoprotosalic), 713-Haplic Chernozem (Endoprotosalic), 720-Amphicalcic Chernozem, 721-Katofluvic Amphicalcic Chernozem, 722-Endofluvic Amphicalcic Chernozem, 723-Endogleyic Amphicalcic Chernozem, 724-Amphicalcic Chernozem (Alcalic), 725-Amphicalcic Chernozem (Endoprotosalic), 726-Amphigleyic Chernozem (Katoprotosalic), 730-Epicalcic Chernozem, 731-Epicalcic Chernozem (Cambic), 732-Katofluvic Epicalcic Chernozem, 733-Katogleyic Epicalcic Chernozem, 734-Amphigleyic Epicalcic Chernozem, 735-Endogleyic Epicalcic Chernozem, 736-Epicalcic Chernozem (Pachic), 737-Epicalcic Chernozem (Endoprotosalic), 741-Endofluvic Endocalcic Chernozem, 742-Endogleyic Endocalcic Chernozem, 743-Endocalcic Chernozem (Cambic), 744-Endocalcic Chernozem (Pachic), 750-Katocalcic Chernozem, 761-Endogleyic Chernozem (Cambic), 762-Katofluvic Endogleyic Chernozem
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean values of NDVI (top) and elevation (bottom) at level 3 of WRB*
*Codes and names of the classes shown in the graphs from left to right are the following 100-Amphiprotosalic Endogleyic Regosol, 210-Pantocalcaric Katofluvic Cambisol, 220-Pantocalcaric Cambisol (Endoprotosalic), 300-Amphicalcic Chernic Gleysol, 301-Endocalcic Chernic Gleysol, 410-Haplic Calcisol (Alcalic), 421-Cambic Calcisol (Katofluvic), 422-Cambic Calcisol (Endogleyic), 423-Cambic Calcisol (Protosodic), 510-Amphigleyic Phaezoem (Protosodic), 520-Chernic Phaeozem, 521-Pantocalcaric Chernic Phaeozem, 522-Katofluvic Chernic Phaeozem, 523-Amphifluvic Chernic Phaeozem, 524-Endofluvic Chernic Phaeozem, 525-Amphigleyic Chernic Phaeozem, 530-Pantocalcaric Phaeozem, 531-Pantocalcaric Phaeozem (Alcalic), 610-Endogleyic Amphicalcic Kastanozem, 611-Amphicalcic Kastanozem (Endoprotosalic), 622-Endocalcic Kastanozem (Cambic), 623-Endogleyic Endocalcic Kastanozem, 624-Amphigleyic Epicalcic Kastanozem, 710-Haplic Chernozem, 711-Haplic Chernozem (Pachic), 712-Haplic Chernozem (Katoprotosalic), 713-Haplic Chernozem (Endoprotosalic), 720-Amphicalcic Chernozem, 721-Katofluvic Amphicalcic Chernozem, 722-Endofluvic Amphicalcic Chernozem, 723-Endogleyic Amphicalcic Chernozem, 724-Amphicalcic Chernozem (Alcalic), 725-Amphicalcic Chernozem (Endoprotosalic), 726-Amphigleyic Chernozem (Katoprotosalic), 730-Epicalcic Chernozem, 731-Epicalcic Chernozem (Cambic), 732-Katofluvic Epicalcic Chernozem, 733-Katogleyic Epicalcic Chernozem, 734-Amphigleyic Epicalcic Chernozem, 735-Endogleyic Epicalcic Chernozem, 736-Epicalcic Chernozem (Pachic), 737-Epicalcic Chernozem (Endoprotosalic), 741-Endofluvic Endocalcic Chernozem, 742-Endogleyic Endocalcic Chernozem, 743-Endocalcic Chernozem (Cambic), 744-Endocalcic Chernozem (Pachic), 750-Katocalcic Chernozem, 761-Endogleyic Chernozem (Cambic), 762-Katofluvic Endogleyic Chernozem
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean values of NDVI (top) and elevation (bottom) at level 3 of WRB*
*Codes and names of the classes shown in the graphs from left to right are the following 100-Amphiprotosalic Endogleyic Regosol, 210-Pantocalcaric Katofluvic Cambisol, 220-Pantocalcaric Cambisol (Endoprotosalic), 300-Amphicalcic Chernic Gleysol, 301-Endocalcic Chernic Gleysol, 410-Haplic Calcisol (Alcalic), 421-Cambic Calcisol (Katofluvic), 422-Cambic Calcisol (Endogleyic), 423-Cambic Calcisol (Protosodic), 510-Amphigleyic Phaezoem (Protosodic), 520-Chernic Phaeozem, 521-Pantocalcaric Chernic Phaeozem, 522-Katofluvic Chernic Phaeozem, 523-Amphifluvic Chernic Phaeozem, 524-Endofluvic Chernic Phaeozem, 525-Amphigleyic Chernic Phaeozem, 530-Pantocalcaric Phaeozem, 531-Pantocalcaric Phaeozem (Alcalic), 610-Endogleyic Amphicalcic Kastanozem, 611-Amphicalcic Kastanozem (Endoprotosalic), 622-Endocalcic Kastanozem (Cambic), 623-Endogleyic Endocalcic Kastanozem, 624-Amphigleyic Epicalcic Kastanozem, 710-Haplic Chernozem, 711-Haplic Chernozem (Pachic), 712-Haplic Chernozem (Katoprotosalic), 713-Haplic Chernozem (Endoprotosalic), 720-Amphicalcic Chernozem, 721-Katofluvic Amphicalcic Chernozem, 722-Endofluvic Amphicalcic Chernozem, 723-Endogleyic Amphicalcic Chernozem, 724-Amphicalcic Chernozem (Alcalic), 725-Amphicalcic Chernozem (Endoprotosalic), 726-Amphigleyic Chernozem (Katoprotosalic), 730-Epicalcic Chernozem, 731-Epicalcic Chernozem (Cambic), 732-Katofluvic Epicalcic Chernozem, 733-Katogleyic Epicalcic Chernozem, 734-Amphigleyic Epicalcic Chernozem, 735-Endogleyic Epicalcic Chernozem, 736-Epicalcic Chernozem (Pachic), 737-Epicalcic Chernozem (Endoprotosalic), 741-Endofluvic Endocalcic Chernozem, 742-Endogleyic Endocalcic Chernozem, 743-Endocalcic Chernozem (Cambic), 744-Endocalcic Chernozem (Pachic), 750-Katocalcic Chernozem, 761-Endogleyic Chernozem (Cambic), 762-Katofluvic Endogleyic Chernozem
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
At the second level of classification (soil types), the basic Chernozem soil type (190) and its transition to Meadow and Alluvial soils (200; 210) are clearly distinguished ( Figure 8 ). No significant differences in productivity or elevation are found between them, but the trend for both parameters is 190 > 200 > 210. For types 200 and 210, some NDVI values are very low (many outliers), suggesting effect of other soil problem (e.g. erosion). The elevation values for soil types 190, 200 and 210 showed the widest range. It is interesting that for the type 200 (meadow Chernozem), the lowest elevation is associated with the lowest (outlier) NDVI value. Within the main type of Meadow soils, three types can be distinguished, the basic type of Meadow soil (300) and the transitions towards Chernozem soils (330) and Alluvial soils (310). There is no significant difference between the NDVI values for these classes, but as expected the order is 330 > 310 > 300; where the order of productivity presumably decreases with the adverse effect of surplus water. The difference between the elevation values of each type is more significant, the alluvial Meadow soil (310) shows significantly smaller value, while the Chernozem Meadow soils (330) lie slightly higher than the Meadow soils (300). Overall, NDVI values by soil type generally reflect the productivity-inhibiting adverse effect of surplus water.
The third level shows the distribution of the two parameters according to the soil subtypes ( Figure 9 ). Subtype for which we do not see boxplot diagrams also appear here (301), as it only has a single soil profile. Compared to the previous ones, the subtypes provide much additional information, since they also display the salinity effect. In the case of Meadow Chernozem soils, productivity visibly decreases in the direction of salt-affected subtypes (201 > 203 > 204) and the elevation decreases similarly. A similar observation can be made for the subtypes of Chernozem Meadow soils (330), the subtype salty in deeper horizons (333) lies at a lower elevation and is less productive than the subtype free from salt effects (331).
We don't really get any extra information from the level 4 boxplot diagrams, as there are a lot of soil variety with a single soil profile here ( Figure 10 ). These soil varieties do not have a boxplot diagram, so the differences in NDVI and elevation between the varieties are not very easily comparable. Wherever this is possible (e.g. 203110 - 203210 or 211200 - 211202), the differences are not very clear either.
Mean values of NDVI (a) and elevation (b) at level 4 of WRB. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean values of NDVI (a) and elevation (b) at level 4 of WRB. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean values of NDVI (a) and elevation (b) at level 4 of WRB. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean values of NDVI (a) and elevation (b) at level 4 of WRB. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean values of NDVI (a) and elevation (b) at level 4 of WRB. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Comparative evaluation of the classification systems
Compared to the ideal case of complete separability (ARNOLD, 2001), only a fraction of the classes was separated ( Figure 11 ). At levels 1, 3 and 4, HU demonstrated better differentiation, but differences were not great.
As shown by Figure 12 the homogeneity of the classes, calculated according to BECKETT & BURROUGH (1971), was greater for WRB, the best at the more detailed levels of 3 and 4. This is explained by the flexibility provided by the large number of principal and supplementary qualifiers. The 1-RV of the WRB was about 2 times higher than the corresponding value of HU ( Figure 12 ).
Mean NDVI (top) and elevation (bottom) of Chernozem (n = 59), Meadow (n = 22) and Alluvial soils (n = 4) at the main type level (HU1) of the Hungarian Classification System
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean NDVI (top) and elevation (bottom) of Chernozem (n = 59), Meadow (n = 22) and Alluvial soils (n = 4) at the main type level (HU1) of the Hungarian Classification System
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean NDVI (top) and elevation (bottom) of Chernozem (n = 59), Meadow (n = 22) and Alluvial soils (n = 4) at the main type level (HU1) of the Hungarian Classification System
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
WRB1 was separated into two, the WRB3 and the WRB4 to four times as many classes as HU ( Table 3 ). The number of HU4 classes significantly increased compared to HU3, and the number of WRB4 classes was twofold of HU4. Statistical evaluation was challenging due to the large number of single-profile classes. HU had 0, 0, 8 and 54% and WRB had 14, 33, 67 and 78% such classes at levels 1, 2, 3 and 4, respectively. At level 4 both systems had a large number of single-profile classes. HU had lower number of single-profile classes, while the WRB was less manageable with higher number. On the other hand Figure 3 shows that the number of classes with more than one profile showed much less difference.
Mean NDVI (top) and elevation (bottom) of the classes of the Hungarian Classification System at level two. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean NDVI (top) and elevation (bottom) of the classes of the Hungarian Classification System at level two. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean NDVI (top) and elevation (bottom) of the classes of the Hungarian Classification System at level two. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean NDVI (top) and elevation (bottom) of the classes of the Hungarian Classification System at level three. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean NDVI (top) and elevation (bottom) of the classes of the Hungarian Classification System at level three. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean NDVI (top) and elevation (bottom) of the classes of the Hungarian Classification System at level three. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Pearson correlation coefficient between ten-year average NDVI values and mean elevation of the distinguished classes at four levels. Number of classes is indicated in brackets
| Level | WRB classification | HU classification |
|---|---|---|
| 1 | 0.388 (7) | 0.561 (3) |
| 2 | 0.763** (18) | 0.763* (7) |
| 3 | 0.574** (49) | 0.821** (12) |
| 4 | 0.562** (59) | 0.707** (26) |
Correlation is significant at the 0.01 level (2-tailed).
Correlation is significant at the 0.05 level (2-tailed).
Correlation of NDVI values with elevation is shown in Table 3 . In case of detailed levels HU3 (r = 0.821**) and WRB3 (r = 0.574**) were found to be suitable for productivity and yield estimation. At level 4, HU also performed better (r = 0.707**) than WRB (r = 0.562**).
Mean NDVI (top) and elevation (bottom) of the classes of the Hungarian Classification System at level four. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean NDVI (top) and elevation (bottom) of the classes of the Hungarian Classification System at level four. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Mean NDVI (top) and elevation (bottom) of the classes of the Hungarian Classification System at level four. See profile, classification and code list in Table 1
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
The ratio of significantly different classes compared to the total number of classes at the four levels of Hungarian Classification and World Reference Base
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
The ratio of significantly different classes compared to the total number of classes at the four levels of Hungarian Classification and World Reference Base
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
The ratio of significantly different classes compared to the total number of classes at the four levels of Hungarian Classification and World Reference Base
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
1-RV values (the fraction of within-class variance/total variance) calculated with NDVI for the four levels of Hungarian Classification and World Reference Base
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
1-RV values (the fraction of within-class variance/total variance) calculated with NDVI for the four levels of Hungarian Classification and World Reference Base
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
1-RV values (the fraction of within-class variance/total variance) calculated with NDVI for the four levels of Hungarian Classification and World Reference Base
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Consistent with the results of SCHULER et al. (2006), the WRB had a greater number of classes ( Table 3, Figure 13 ). However, due to the greater number of environmental factors covered, any global classification system is likely to have a greater number of classes than local systems. In HU, the environmental factors are closely related to the specific morphological, sedimentological and climatic conditions of the Pannonian basin, which are reflected in the specific soil development characteristics. These particularities have determined the intensity of soil-forming factors and processes, which is reflected in the local organic matter and CaCO3 accumulation, water balance and leaching. Such pedogenic processes indicate an increase of the thickness of the profile during the Quaternary, when loess deposition and thus the widespread presence of CaCO3 (STEFANOVITS, 1963), together with the alluvial character of the landscape and the ubiquitous shallow water table, significantly influenced the physical and chemical properties of the soils in the area (ARANY, 1956).
Number of classes with more than one profile at the four levels of Hungarian Classification and World Reference Base
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Number of classes with more than one profile at the four levels of Hungarian Classification and World Reference Base
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Number of classes with more than one profile at the four levels of Hungarian Classification and World Reference Base
Citation: Agrokémia és Talajtan 71, 1; 10.1556/0088.2022.00121
Our results show that none of the classification systems performed excessively poorly or outstandingly when only levels 3 and 4 are considered. An advantage and at the same time a disadvantage of WRB is that it considers many aspects using a large number of physical and chemical parameters (KRASILNIKOV et al., 2009). The good performance of HU may be due to extensive experience with alluvial, floodplain and saline soils. This knowledge was integrated from earlier Hungarian classification systems (TREITZ, 1924, DE SIGMOND, 1927, 1938) into the current soil classification. More details of technical evaluation are provided in TÓTH et al. (2022).
Because transitioning to a new system involves significant changes in all databases, including GIS datasets, which may lead to disputes (BIDLÓ, 2019; MAKÓ, 2019; TÓTH, 2019a, b), such transitions should ideally be preceded by a thorough discussion highlighting the advantages and disadvantages of both the old and new systems in terms of land use management and mapping.
Acknowledgement
This research was financed by the Hungarian National Research, Development and Innovation Office Foundation (Grant No. K 124290). We are deeply indebted to the following persons Péter Horváth for facilitating the field work in the firm where the plot is located; Nóra Szűcs-Vásárhelyi and Márton Tóth for helping in the description of profiles; Zsófia Adrienn Kovács for assisting in the field and laboratory and Gyöngyi Barna for encouragement and motivation.
References
ALLGOOD, F.R. & GRAY, F ., 1978. Utilization of soil characteristics in computing productivity ratings of Oklahoma soils. Soil Science. 125. 359–366.
ARANY, S ., 1956. Salt affected soils and their reclamation. Mezőgazda, Budapest. (in Hungarian)
ARNOLD, R. W ., 2001. Soil classification principles. In: Soil сlassification 2001: contributions to the International symposium “Soil Classification 2001”, 8–12 October 2001, Velence, Hungary. pp. 3–8.
BECKETT, P.H.T. & BURROUGH, P.A ., 1971. The relation between cost and utility in soil survey: IV. Comparison of the utilities of soil maps produced by different survey procedures, and to different scales. Journal of Soil Science. 22. (4) 466–480.
BIDLÓ, A ., 2019. Invited commentaries to „Michéli Erika, Fuchs Márta, Szegi Tamás, Csorba Ádám, Dobos Endre, Szabóné Kele Gabriella: “Proposal for the modernization of the national soil classification system. Principles, structure and classification rules” (2018.10.10.). Agrokémia és Talajtan. 68. (2) 345–354. (in Hungarian)
BUOL, S.W ., SOUTHARD, R.J ., GRAHAM, R.C. & MCDANIEL, P.A ., 2011. Soil genesis and classification. John Wiley & Sons.
DE SIGMOND, A . 1938. The principles of soil science. Thomas Murby & Co. London.
DE SIGMOND, A . 1927. Hungarian alkali soils and methods of their reclamation. Special publication issued by the California Agricultural Experiment Station. University of California. Berkeley.
ESFANDIARPOUR, I ., MOSLEH, Z. & FARPOOR, M.H ., 2018. Comparing soil taxonomy and WRB systems to classify soils with clay-enriched horizons (A case study: arid and semi-arid regions of Iran). Desert. 23. (2) 315–325.
IZSÓ I . (ed.), 1986. Táblázatok a termőföld értékeléséhez. MÉM Földügyi és Térké- pészeti Hivatal. Budapest. (in Hungarian)
JASSÓ F ., HORVÁTH B ., IZSÓ I ., KIRÁLY L ., PARÁSZKA L ., KELE G ., 1989. Guidelines to large-scale soil mapping. 2nd ed. Agroinform, Budapest. (in Hungarian)
KRASILNIKOV, P ., ARNOLD, R.W. & IBÁÑEZ, J.J ., 2010. Soil classifications: their origin, the state-of-the-art and perspectives. In: Proceedings of the 19th World Congress of Soil Science: Soil solutions for a changing world, Brisbane, Australia, 1–6 August 2010. Symposium 1.4. 2 Soil classification benefits and constraints to pedology. International Union of Soil Sciences (IUSS), c/o Institut für Bodenforschung, Universität für Bodenkultur. pp. 19–22.
KRASILNIKOV, P ., MARTI, J.J.I ., ARNOLD, R. & SHOBA, S. (eds.)., 2009. A handbook of soil terminology, correlation and classification. Routledge, London.
KUBIËNA, W.L ., 1953. Bestimmungsbuch und Systematik der Böden Europas. Ferdinand Enke Verlag, Stuttgart.
MAKÓ, A ., 2019. Invited commentaries to „Michéli Erika, Fuchs Márta, Szegi Tamás, Csorba Ádám, Dobos Endre, Szabóné Kele Gabriella: “Proposal for the modernization of the national soil classification system. Principles, structure and classification rules” (2018.10.10.). Agrokémia és Talajtan. 68. (2) 323–332. (in Hungarian)
MCBRATNEY, A. B ., SANTOS, M. M. & MINASNY, B ., 2003. On digital soil mapping. Geoderma. 117. (1–2) 3–52.
MICHÉLI E ., FUCHS M ., SZEGI T ., CSORBA, Á ., DOBOS E ., SZABÓNÉ KELE G ., 2018. Proposal for the modernization of the national soil classification system. Principles, structure and classification rules. Discussion material. Szent István University, Gödöllő. (in Hungarian)
OGUNKUNLE, A.O. & BECKETT, P.H.T ., 1988. Combining soil map and soil analysis for improved yield prediction. Catena. 15. (6) 529–538.
PETTORELLI, N ., 2013. The normalized difference vegetation index. Oxford University Press, Oxford.
ROSSITER, D.G ., ZENG, R. & ZHANG, G.L ., 2017. Accounting for taxonomic distance in accuracy assessment of soil class predictions. Geoderma. 292. 118–127.
SALEHI, M.H ., 2018. Challenges of Soil Taxonomy and WRB in classifying soils: Some examples from Iranian soils. Bulletin of Geography. Physical Geography Series. 14. (1) 63–70.
SCHULER, U ., CHOOCHAROEN, C ., ELSTNER, P ., NEEF, A ., STAHR, K ., ZAREI, M. & HERRMANN, L ., 2006. Soil mapping for land‐use planning in a karst area of N Thailand with due consideration of local knowledge. Journal of Plant Nutrition and Soil Science. 169. (3) 444–452.
SHRADER, W.D ., SCHALLER, F.W ., PESEK, J.T ., SLUSHER, D.F. & RIECKEN, F.F ., 1960. Estimated crop yields on Iowa soils. U. S. Deptartment of Agriculture and Iowa Agricultural. Experiment. Station. Special Report. 25, April 1960.
SOROKIN, A ., OWENS, P ., LÁNG, V ., JIANG, Z.D ., MICHÉLI, E. & KRASILNIKOV, P ., 2021. “Black soils” in the Russian Soil Classification system, the US Soil Taxonomy and the WRB: Quantitative correlation and implications for pedodiversity assessment. Catena, 196. 104824.
STEFANOVITS, P ., 1963. Soils of Hungary. Akadémiai Kiadó, Budapest, Hungary. (in Hungarian)
SZABOLCS, I . (Ed.), 1966. Handbook of the large-scale genetic soil mapping. OMMI Genetikus Talajtérképek. Ser. 1. No. 9. Budapest. (In Hungarian)
TEAL, R.K ., TUBANA, B ., GIRMA, K ., FREEMAN, K.W ., ARNALL, D.B ., WALSH, O . & RAUN, W.R . 2006. In‐season prediction of corn grain yield potential using normalized difference vegetation index. Agronomy Journal. 98. (6) 1488–1494.
TÓTH, G ., 2019a. Invited commentaries to „Michéli Erika, Fuchs Márta, Szegi Tamás, Csorba Ádám, Dobos Endre, Szabóné Kele Gabriella: “Proposal for the modernization of the national soil classification system. Principles, structure and classification rules”. Agrokémia és Talajtan. 68. (2) 333–344. (in Hungarian)
TÓTH, G ., MAKÓ, A. & MÁTÉ, F ., 2009. Designation of local varieties in the Hungarian soil classification system: Remarks from a viewpoint of land evaluation application. Eurasian Soil Science. 42. (13) 1448–1453.
TÓTH, G ., MONTANARELLA, L ., STOLBOVOY, V ., MÁTÉ, F ., BÓDIS, K ., JONES, A ., PANAGOS, P. & VAN LIEDEKERKE, M ., 2008. Soils of the European union. JRC Scientific and Technical Reports. Office for Official Publications of the European Communities, Luxembourg.
TÓTH, T ., 2019b. Invited commentaries to „Michéli Erika, Fuchs Márta, Szegi Tamás, Csorba Ádám, Dobos Endre, Szabóné Kele Gabriella: “Proposal for the modernization of the national soil classification system. Principles, structure and classification rules”. Agrokémia és Talajtan. 68. (2) 315–321. (in Hungarian)
TÓTH, T ., GALLAI, B ., NOVÁK, T ., CZIGÁNY, S ., MAKÓ, A ., KOCSIS, M ., ÁRVAI, M ., MÉSZÁROS, J ., LÁSZLÓ, P ., KOÓS, S. & BALOG, K ., 2022. Practical evaluation of four classification levels of Soil Taxonomy, Hungarian classification and WRB in terms of biomass production in a salt-affected alluvial plot. Geoderma. 410. 115666.
TREITZ, P ., 1924. The nature and properties of salt-affected soils. Budapest.
VAN HUYSSTEEN, C. W ., LE ROUX, P. A. L . & TURNER, D. P ., 2013. Principles of soil classification and the future of the South African system. South African Journal of Plant and Soil. 30. (1) 23–32.
WEBSTER, R ., HODGE, C.A.H ., DRAYCOTT, A. P. & DURRANT, M.J ., 1977. The effect of soil type and related factors on sugar beet yield. The Journal of Agricultural Science. 88. (2). 455–469.
WRB, IUSS WORKING GROUP , 2015. World Reference Base for Soil Resources 2014, update 2015 International soil classification system for naming soils and creating legends for soil maps. World Soil Resources Reports No. 106. Food and Agriculture Organization of the United Nations, Rome.
