Use of remote sensing to evaluate the effects of environmental factors on soil salinity in a semi-arid area

1 Use of remote sensing to evaluate the effects of environmental factors on 1 soil salinity in a semi-arid area 2 Francisco Pedrero Salcedo a*, Pedro Pérez Cutillas b, Juan José Alarcón Cabañero a and 3 Alessandro Gaetano Vivaldi c 4 a Irrigation Department. Centro de Edafología y Biología Aplicada del Segura (CEBAS5 CSIC) Apdo. 164, 30100 Espinardo, Murcia, Spain 6 b Department of Geography, Campus de La Merced, 30001 Murcia, Spain 7 cDipartimento di Scienze Agro-Ambientali e Territoriali, Università degli Studi di Bari 8 Aldo Moro, Via Amendola 165/A, 70126 Bari, Italy 9 10 Corresponding Author: Apdo. 164, 30100 Espinardo, Murcia, Spain. Tel: +34 11 665961041. Email: [email protected] 12 13 Keywords: Irrigated agriculture, degraded water, secondary soil salinization, remote sensing 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 2 Abstract 29 The global water crisis, driven by water scarcity and water quality deterioration, is expected to 30 continue and intensify in dry and overpopulated areas, and will play a critical role in meeting 31 future agricultural demands. Sustainability of agriculture irrigated with low quality water will 32 require a comprehensive approach to soil, water, and crop management consisting of siteand 33 situation-specific preventive measures and management strategies. Other problem related with 34 water quality deterioration is soil salinization. Around 1Bha globally are salinized and soil 35 salinization may be accelerating for several reasons including the changing climate. The 36 consequences of climate change on soil salinization need to be monitored and mapped and, in this 37 sense, remote sensing has been successfully applied to soil salinity monitoring. Although many 38 issues remain to be resolved, some as important as the imbalance between ground-based 39 measurements and satellite data. The main objective of this paper was to determine the influence 40 of environmental factors on salinity from natural causes, and its effect on irrigated agriculture 41 with degraded water. The study was developed on Campo de Cartagena, an intensive water42 efficient irrigated area which main fruit tree is citrus (30%), a sensible crop to salinity. nine 43 representative citrus farms were selected, soil samples were analysed and different remote sensing 44 indices and sets of environmental data were applied. Despite the heterogeneity between variables 45 found by the descriptive analysis of the data, the relationship between farms, soil salinity and 46 environmental data showed that applied salinity spectral indices were valid to detect soil salinity 47 in citrus trees. Also, a set of environmental characterization provided useful information to 48 determine the variables that most influence primary salinity in crops. Although the data extracted 49 from spatial analysis indicated that to apply soil salinity predictive models, other variables related 50 to agricultural management practices must be incorporated. 51 52 53 54 3 1 Introduction 55 56 Water scarcity is among the top five global risks and reaches far beyond socio-economic and 57 environmental challenges impacting the livelihoods and wellbeing of all people (World Economic 58 Forum, 2020). The global water crisis, driven by water scarcity and water quality deterioration, 59 is expected to continue and intensify in dry and overpopulated areas (UN-Water, 2020). 60 Despite growing water scarcity and water quality deterioration, irrigation has played a key role in 61 providing food for the expanding global population and is expected to play a critical role in 62 meeting future agricultural demands (World Bank, 2017). Characterized by limited freshwater 63 resources, arid and semiarid areas rely on irrigation to ensure productivity and sustainability of 64 agricultural production systems. 65 Although salt management techniques such as leaching seem straightforward, the long-term 66 sustainability of irrigated lands remains a challenge as irrigation itself impacts land and water 67 resources in ways that can lower farm productivity over time, particularly in arid and semiarid 68 areas where most irrigation takes place. Irrigation in these areas inevitably degrades the quality 69 of water in downstream reaches, as dissolved salts enter irrigation return flows, thus increasing 70 the salinity of watercourses from which other farmers and communities draw their water supplies 71 (Wichelns and Qadir, 2015). 72 Soil salinity occur under a wide range of climatic conditions, both under natural (primary 73 salinization) and human-induced conditions (secondary salinization), but is particularly 74 widespread in arid and semi-arid climates where rainfall is inadequate to leach accumulated salts 75 below the plant’s rooting zone, whether irrigated or rainfed (Hopman et al., 2021). 76 The key factors associated with soil salinity are geology and its chemistry, climate, and local 77 hydrology ((Hopman et al., 2021). Under field conditions, distribution of salts is neither uniform 78 with soil depth nor constant with time. The non-uniformity of salinity distribution is usually 79 affected by both irrigation and leaching practices designed to control salt gradients in the root 80 zone, and by the amount and patterns of rainfall (Minhas et al., 2020). Recent studies have 81 4 demonstrated the weather-related influence on crop response to salinity (Groenveld et al., 2013; 82 Tack et al., 2015; Perelman et al., 2020). 83 As agricultural use of saline, sodic, and saline-sodic waters increase, the sustainability of irrigated 84 agriculture will become a more serious issue. In the future, sustainable food and feed systems 85 should have good crop production with minimized adverse environmental and ecological impacts 86 (Qadir and Oster, 2004). Sustainability of agriculture irrigated with low quality water will require 87 a comprehensive approach to soil, water, and crop management consisting of siteand situation88 specific preventive measures and management strategies. 89 Approximately 6% of the world’s terrestrial land is believed to be salinized by primary 90 salinization. In addition, some 20% of all cropland and between 1/4 and 1/3 of irrigated land is 91 salinized by secondary salinization, totalling about 1Bha globally (Hopman et al., 2021). 92 Estimates suggest that the global annual cost of salt-induced land degradation in irrigated areas 93 could be US$ 27 billion due to lost crop production (Qadir et al., 2014). These costs are expected 94 to be even higher when other cost components such as infrastructure deterioration. In addition, 95 there could be additional environmental costs associated with salt-affected lands as these lands 96 emit more greenhouse gases thus contributing to global warming (Ivits et al., 2013). 97 Soil salinization may be accelerating for several reasons including the changing climate. The 98 consequences of climate change on soil salinization processes have been overlooked and that 99 changes the extent by which soil salinity will need to be monitored and mapped Corwin (2020). 100 He suggests that both proximal and remote sensors are the best methods to achieve this in a timely 101 manner. 102 Remote sensing has been successfully applied to soil salinity monitoring, ranging from detailed 103 detection using radiometers in the laboratory to medium resolution multispectral analyses (Zhang 104 et al., 2015; Kasim et al., 2017; Ivushkin et al., 2018; Tian et al., 2020; Xu et al., 2020; Zhu et al., 105 2021). Nowadays, multispectral remote sensing techniques have shown improvements that 106 effectively characterize salinity processes in the soil and plant canopy, in addition to offering a 107 wider range of spatial analysis. In this sense, they have proven efficient for assessing salinity at 108 regional scales, and even for estimating soil salinity (up to 20 dSm-1) from ground reflectance 109 5 spectral analysis which is within the salinity range of agricultural crop response (Scudiero et al., 110 2015). However, many issues remain to be resolved, some as important as the imbalance between 111 ground-based measurements and satellite data, because multispectral bands are not accurate 112 enough to precisely detect the fine physiological differences caused by saline stress (Masoud et 113 al., 2019). 114 In analysis at regional scale, other issues arise related to the complexity and heterogeneity of land 115 cover combined to spatial and temporal dynamics of soil salinization under semiarid conditions 116 (Scudiero et al., 2015; Sultanov et al., 2018). Modelling has allowed the addition of covariates 117 describing the environmental setting (geomorphology, climate, hydrology, land use, soils, etc.), 118 supporting the spectral reflectivity data acquired by remote sensors (Peng, 2018; Chi et al., 2019; 119 Zhang et al., 2021). Although this information from different sources, in addition to diverse spatial 120 variability, provides uncertainty in the simulation models results (Scudiero et al., 2015; Zhang et 121 al., 2021). An important issue that should be considered is the scaling effects. First, it would be 122 convenient to determine a useful balance in spatial resolution between the different layers of 123 information applied in the models (Mukundan et al., 2010; Xu et al., 2017). The dynamics 124 involved in salinity processes may be affected by scale dependence in their representation on the 125 landscape (Zhou et al., 2013; Ruiz-Navarro et al., 2012). Second, it would be valuable to establish 126 the correct spatial resolution specifically for each single variable, which allows the process to be 127 represented at the proper scale under our purposes. In addition, related to the scaling factor, the 128 interaction of salinity effects caused by human activities should not be neglected, since it may 129 become a concern of agricultural sustainability (Welle and Mauter, 2017; Newton et al., 2020). 130 These impacts tend to be arbitrary and localised in contrast to natural salinity processes, so their 131 application in modelling should be treated with caution due to adverse spatial autocorrelation 132 effects on these variables (Zhang et al., 2011). In summary, it would be recommended to apply a 133 multilevel factor system, analysing anthropogenic and biophysical drivers in the medium to long 134 term, as many factors interact and influence each other during the salinization process (Seydehmet 135 et al., 2018a). 136 6 The study area is the Region of Murcia, one of the driest places in Europe.The studied area is the 137 Region of Murcia, the driest area in Europe. In this region, water is the limiting factor, since the 138 region relies almost entirely on irrigation due the low annual rainfall of +/- 300mm, which 139 insufficient for most crops. Along with increasing water scarcity due to climate change, farmers 140 in the region are being forced to rely more and more on saline groundwater as well as treated 141 wastewater, which can produce adverse effects on irrigated agriculture (Pedrero and Alarcón, 142 2009). Unlike other regions of Spain, some aquifers in Campo de Cartagena have been abandoned 143 due to salinization (Rodriguez-Estrella et al., 2004) form sea water intrusion caused by excessive 144 groundwater pumping (Custodio et al., 2009). Even so soil salinization still exists, it is estimated 145 that around 100,000 has of irrigation, are irrigated of which 85% have a high salt content on the 146 irrigation water (Alcón et al., 2014). Treated or non-saline water is generally more expensive than 147 saline water, so financial feasibility influences the quality of water used (Bastida et al., 2017). 148 The main objective of this paper is not to provide a prediction of salinity from environmental 149 variables, which as mentioned above has been studied in the literature in abundance. Rather, it is 150 to determine the influence of environmental factors on salinity from natural causes, and its effect 151 on irrigated agriculture with degraded water in an ecosystem close to a coastal lagoon under high 152 anthropic pressure. Our hypothesis holds that in semi-arid conditions, the influence of 153 environmental factors has an important effect on agricultural soil salinity. This can lead to 154 interactions in agronomic management, which must be controlled in order not to affect the 155 productivity of the farms. Therefore, to analyse this issue, the following specific objectives are 156 proposed: i) To determine the reliability of spectral indices of salinity in citrus crops; ii) To 157 establish which environmental variables have the strongest influence on agricultural soil salinity; 158 iii) To analyse the relationship between the environmental properties of the farms under study 159 and their soil salinity values; and iv) To identify the role of environmental factors on soil salinity 160 in a semi-arid ecosystem. 161 162 7 2 Materials and methods 163 2.1 Study area 164 The study area is located on the Campo de Cartagena on the southeast portion of the Iberian 165 Peninsula (Figure 1). This natural region constitutes a basin composed of low permeability detrital 166 sediments, mainly constituted by marl, with intervals of high permeability materials such as 167 limestone, sand and conglomerates (Rodríguez-Estrella, 2004). The Campo de Cartagena covers 168 1,300 km2 of which 30% is under intensive water-efficient agriculture. This irrigated region 169 requires about 180 Mm3/year, which is supplied by 6 different water sources that irrigate around 170 50,000 Ha of mainly horticultural crops (59%) and citrus trees (30%). Citrus trees were selected 171 for this study because of their sensitivity to specific ion toxicities as comparted to most annual 172 crops. 173 174 2.2 Soil data collection and environmental information 175 2.2.1 Irrigation water management, analysis and soil sampling and design 176 In order to assess soil salinity, 9 representative citrus farms from Campo de Cartagena were 177 selected for the study (Figure 1). Farms size was between 8-100 Has and trees from ranged from 178 2 to 30 years old (Table 1). 179 The irrigation system on the different farms consisted of a double drip line laid on the soil surface 180 next to each tree row. It provided three self-pressurecompensating on-line emitters per tree 181 discharging 4 L h−1 each, placed at 0.85 m from the trunk and spaced 0.9 m apart, except on the 182 citrus farms with trees less than 5 years (1 and 5) where the irrigation was distributed with a single 183 drip line, and 2 emiters discharging 4 L h−1 each, placed at 0.3 m from the trunk and spaced 0.6 184 m apart. Trees were irrigated daily from January to December. The total amounts of water applied 185 were measured within line water flow meters and is described on table 1. Pest control was that 186 commonly used by growers in Campo de Cartagena, and no weeds were allowed to develop within 187 the orchard. 188 8 Three water samples from each farm were collected weekly during one month in order to 189 characterize irrigation water quality. The samples from each irrigation source were collected in 190 glass and plastic bottles, transported in an ice chest to the laboratory and stored at 5 ºC before 191 being processed for analysis. Electrical conductivity (EC) was determined using an electrical 192 conductivity meter as above. The concentration of cations was determined by inductively coupled 193 plasma mass spectrometry (ICP-ICAP 6500 Duo Thermo, England) and anions (chloride) were 194 analyzed by ion chromatography (Metrohm, Switzerland) with a liquid chromatograph. 195 Several sample points were selected at each farm and samples were collected from late October 196 to early November. At each sample point, soil samples were collected in the top 0.3 m of soil 197 between furrows (dry area) (A) and on the ridges (wet area) (B). Two replicates per soil sample 198 were collected. Soil samples were air dried and ground before being sieved to a particle size of 2199 mm. Afterward, soluble salts were determined in the saturated paste extract as described by the 200 method described by Rhoades (1982). The ECe was measured with a multi-range Cryson-HI8734 201 electrical conductivity meter (Crison Instruments, S.A., Barcelona, Spain), and the concentration 202 of cations on the soil solution was determined by inductively coupled plasma mass spectrometry 203 (ICP-ICAP 6500 Duo Thermo, England). 204 205 206 2.2.2 Remote sensing indices 207 The surface reflectance values were obtained from the OLI and TIRS sensors of the Landsat 8 208 satellite. The scenes of the study area (Path 199/ Row 34) were obtained from the USGS product 209 service (2021). The selection of the date of acquisition of the image (10/26/2020) was made as 210 close as possible to the capture of the ECe records in the field (late October and early November). 211 It is important to mention that no rainfall was recorded in the weeks prior to the collection of 212 salinity data in the field, a condition that did not interfere in the processes of soil moisture 213 fluctuation that would modify the actual soil salinity (Alexakis et al., 2016; Yu et al., 2019). The 214 downloaded satellite image was pre-processed at L1TP level with radiometric calibration and 215 9 orthorectification. Image conversion to TOA reflectance and brightness temperature was done 216 with a semi-automatic classification plugin (Congedo, 2020) using QGIS v.3.12.2. (QGIS, 2021). 217 A series of salinity indices were selected based on the literature, as well as some other related 218 spectral indices (Table 2). The vegetation spectral response can show the presence of salt in the 219 soils due to the increase of stress in the crops (Asfaw et al., 2018). Therefore, wavelengths in the 220 visible and near infrared spectrum, related to photosynthetic activity, are useful to determine these 221 soil salinity processes. Images with wavelengths of the green range: 𝜌𝑔 (0.53 - 0.59 µm), red: 𝜌𝑟 222 (0.64 - 0.67 µm), and near infrared: 𝜌𝑁𝐼𝑅 (0.85 - 0.88 µm) were applied in the Salinity Index 223 (𝑆𝐼 =√𝜌𝑟∗ 𝜌𝑁𝐼𝑅) (Dehni and Lounis, 2012); Normalized Difference Salinity Index (𝑁𝐷𝑆𝐼 = 224 (𝜌𝑟− 𝜌𝑁𝐼𝑅 𝜌𝑟+ 𝜌𝑁𝐼𝑅 ⁄) (Khan et al., 2005); and the Vegetation Soil Salinity Index (𝑉𝑆𝑆𝐼 = 2 ∗ 225 𝜌𝑔− 5 ∗ (𝜌𝑟+ 𝜌𝑁𝐼𝑅)) (Dehni and Lounis, 2012). 226 Similarly, other indices were obtained in two ranges of the Short Wave Infrared: 𝜌𝑆𝑊𝐼𝑅1 (1.57 - 227 1.65 µm) and 𝜌𝑆𝑊𝐼𝑅2 (2.11 - 2.29 µm) that discriminates the vegetation and the soil moisture 228 content, as well as Thermal Infrared: 𝜌𝑇𝐼𝑅1 (10.60 - 11.19 µm) which estimates the soil moisture 229 and the thermal mapping. They were applied in the Soil Moisture Index (𝑆𝑀𝐼 = 230 (𝜌𝑇𝐼𝑅1 − 𝜌𝑆𝑊𝐼𝑅1 𝜌𝑇𝐼𝑅1 + 𝜌𝑆𝑊𝐼𝑅1 ⁄) to calculate soil moisture (Hasab et al., 2020); and the Clay 231 Chemical Index (𝐶𝐶𝐼 = (𝜌𝑆𝑊𝐼𝑅1 𝜌𝑆𝑊𝐼𝑅2 ⁄) to calculate the content of clay chemicals in the soil 232 (Allbed and Kumar, 2013; Khattab and Merkel, 2014). 233 2.2.3 Others environmental data 234 To deepen the analysis of soil salinity, different sets of environmental data were applied and are 235 described on Table 2. First, the characteristics of the soils in the study area were determined from 236 digital soil mapping based on a global compilation of soil profile data (WoSIS) and environmental 237 information from soilgrids.org. SoilGrids data is not as accurate or relevant as distributed data 238 that could be produced locally, but it can be used as an influential variable in the design of models 239 at different scales (Hengl et al., 2017). The percentage of clay, silt and sand, as well as the content 240 of organic carbon within the soil was used. Another useful data was the depth to the bedrock, 241 which is important to understand the vertical development of the salinity processes. In this sense, 242 16 water and desalination). On the farms where the water quality is worst (specially the EC), is 411 because the amount of water granted from Tajo-Segura transfer canal was insufficient to 412 adequately meet the crop water requirements. 413 4.2 Remote sensing on soil salinity management on farms 414 The effects of salinity on yields of different crops are known to depend on interactions between 415 salinity, soil, water and climatic conditions (Maas and Grattan, 1999; Grattan et al., 2002). The 416 statistical correlation coefficients between the different environmental variables and the ECe 417 obtained in the field showed a similarity in both types of samples (collected in the furrows 418 between plants (ECe A) or in the ridge next to the irrigation drippers (ECe B), despite the soil 419 salinity being extremely variable across the terrain, even over close distances where salinity varies 420 with topography and other environmental factors (Masoud et al., 2019). This result suggests that 421 the spectral reflectivity captured by the sensor at pixel scale, as an average product of a 422 heterogeneous land surface emissivity (Zhao-Liang et al., 2013), provides reliable correlation 423 values for soil salinity regardless of the levels of salt accumulation that may occur in that same 424 pixel. In this case, the plant canopy covers a large part of the pixel surface, appearing as the object 425 that most influences the spectral emissivity, so leaves become the transmitter of the effects of soil 426 salinity (Scudiero et al., 2015; Romero-Trigueros et al., 2017; Tian et al., 2020). This occurs 427 because salinity around plant roots can decrease the water absorption by roots, causing a water 428 stress that influences the radiation emitted by the leaves, especially in the red-edge and near429 infrared wavelengths (Kriston-Vizi et al., 2008; Zhu et al., 2021). 430 Several studies have found leaf anatomical changes as a result of prolonged exposure to saline 431 irrigation (Romero-Aranda et al., 2002). Most of them have described an increase in leaf thickness 432 as well as a decrease in intercellular spaces in the mesophyll spongy layer (Nastou et al. 1999; 433 Contreras et al., 2014). Increases in leaf thickness have also been observed in olive trees irrigated 434 with saline water (Vigo et al., 2005; Kchaou et al., 2010). It should not be overlooked that the 435 effects of environmental stress on photosynthetic parameters cause significant gradients within 436 the leaf, because the upper layers of cells are exposed to a more efficient photosynthetic photon 437 flux density than the inner ones (Peguero-Pina et al. 2009; Morales et al., 2014), so this 438 17 heterogeneity within the leaf may pose some difficulties in the detection of spectral emissivity. 439 The age of the leaves must also be taken into account in the interpretation of remotely sensed 440 data. Older plants are less tolerant to salinity than younger plants, the former being more sensitive 441 to fluorescence decrease under salt stress (Touchette et al. 2012). In addition, it has been reported 442 in some species that older leaves suffer greater photo oxidative salinity-related stress than younger 443 leaves, despite being located lower in the canopy and therefore needing less photoprotection 444 (Tounekti et al. 2012). 445 446 The influence of saline irrigation water on changes in plant morphology allows the use of remote 447 sensors using near-infrared information. At this wavelength, the chlorophyll content of the canopy 448 can be determined, a factor that affects the normal growth and development of plants (Zhang et 449 al., 2016; Kasim et al., 2017). This could explain the significant correlation coefficients obtained 450 for the salinity indices applied in this research (Table 4), since in both cases was use near-infrared 451 spectral values in their equations. Saline stress, in contrast to water stress, produces ion toxicity 452 and damages crop cells, so reflectivity captured in the blue and red regions of the visible spectrum 453 can be used to quantify chlorophyll retention responses, photosynthetic capacity and internal leaf 454 structure (Kriston-Vizi et al., 2008; Zhang et al., 2011a). The blue wavelength range has also been 455 successfully used for detecting soil salinity, which differs from the response of crops to spectral 456 changes under other environmental stresses (Zhu et al., 2021). On the other hand, average canopy 457 temperatures on irrigated crops are known to correlate significantly with the soil salinity (Ishimwe 458 et al., 2014; Ivushkin et al., 2017; Ivushkin et al., 2018). In our case, the thermal Infrared band is 459 shown to be an important variable in the prediction of soil salinity, determining an increase in 460 canopy temperature with increasing saline concentration. However, increased thermal energy 461 dissipation from the foliar coverage can occur due to macro and micronutrient deficiencies or 462 metal toxicities in plants (Morales et al., 2014), which is important to consider. This factor may 463 be related to the negative correlation values with ECe obtained with the CCI (Table 4), as the 464 presence of chemical components such as Pb, Fe, Cu, Cr and Zn would reduce the salinity 465 response of the plants. 466 18 Likewise, soil moisture (SMI) describes a negative correlation with soil salinity (Table 4), 467 resulting in a decrease in salinity when soil moisture increases. The low correlation coefficient 468 observed for this variable may be due to a number of reasons, largely associated with salt leaching 469 processes in the soil. Firstly, it is important to identify the irrigation management on the different 470 farms analysed, due to the variation in the irrigation system protocols applied. The quality of the 471 water used is another important factor, as its salinity level will depend on the source employed 472 (Libutti and Monteleone, 2017). In our study area it is common to use groundwater extracted from 473 boreholes near the coast, desalinated water, as well as reclaimed water, therefore salinity levels 474 are significant and different between the farms studied (Table 8). In order to improve these 475 conditions, better quality water from water transfers is also used (Jimenez-Martinez et al., 2016). 476 Consequently, depending on all these factors, the irrigation water supply will cause different 477 vertical fluxes processes at depth (González-Alcaraz et al., 2014), combined with strong 478 evapotranspiration causing a heterogeneous leaching of salts into the soil. Some other research in 479 arid conditions, combining different types of irrigation with water of different quality, it was 480 observed that after the first irrigations the soluble salt content increased initially, decreasing with 481 long-term irrigation. Soil moisture did not change significantly after irrigation, but pH, electrical 482 conductivity and total soluble salt content of each soil profile layer showed general downward 483 trends (Mansouri et al., 2014). The water imbalance of the Campo de Cartagena basin has mainly 484 been caused by anthropogenic causes, leading to complex soil-water interaction processes 485 (Jimenez-Martinez et al., 2016; Fernández-García et al., 2021). The current water table, with a 486 loss of water quality, shows an upward trend mainly due to percolation irrigation returns. These 487 flows feed the highly polluted upper aquifers therefore they are not being exploited at present, 488 while groundwater extraction pressures are produced in the deeper aquifers (Contreras et al., 489 2014). Despite the rising water table, it is not high enough to affect saline percolation processes. 490 Therefore, the water table variable in our study has no interaction on ECe. This result is contrary 491 to other studies where intensive irrigation and high groundwater levels can induce a rise in the 492 water table leading to an accelerated accumulation of salts in the root zone of the soil (Ibrakhimov 493 et al., 2007; Sultanov et al., 2018). 494 19 The use of hyperspectral sensors has allowed to relate soil salinity at different soil depths through 495 the leaf emissivity response (Zhu et al., 2021). Moreover, multispectral techniques using 496 electromagnetic radiation in the microwave range can detect the soil surface through the canopy, 497 although at a depth of 2 to 5 cm from the topsoil (Li et al., 2007). Therefore, in order to identify 498 the degree of salinity stress in crops, it is necessary to monitor the salinity of the soil at depth. 499 Soil salinity along the vertical profile is usually nonuniform due to the vertical circulation of 500 water in the soil (Du et al., 2008; Yang et al., 2008). Land cover is a key factor to determine the 501 salinity disposition at depth (Le Roux et al., 2010; Yu et al., 2018; Masoud et al., 2019). Bare 502 soils tend to have higher surface salinity profiles, while the volume of vegetation cover leads to a 503 decrease in surface salinity and an increase in depth salinity. In addition, in some irrigated crops, 504 this salt accumulation at depth is increased by the absorption of saline water by the crops; as well 505 as by the additional water input for salt leaching (Yang et al., 2008). At all events, a larger surface 506 range down to the bedrock determines a higher soil volume, which allows for a lower salt content 507 when the processes described above occur. In our case, the effect of soil depth is positively related 508 to the decrease in salinity in agricultural fields (Table 4), in accordance with other researches (Du 509 et al., 2008; Le Roux et al., 2010; Masoud et al., 2019). Soil texture is a key soil property, where 510 a clear association with soil EC has been found. (Saey et al., 2009; Hossain et al., 2020). The 511 behaviour of soil electrical properties and texture features has been used to characterise soil 512 salinity, but for non-saline soil conditions they have also been useful (Domsch and Giebel, 2004; 513 McCutcheon et al., 2006). Variations in the clay and organic matter content of the soil have 514 permitted the development of pedotransfer functions predictive of apparent electrical conductivity 515 (Saey et al., 2009). In several studies, correlations between soil clay content and ECe have been 516 found, whereby increasing finer particle sizes increased the ECe value (Sultanov et al., 2018; 517 Hasab et al., 2020; Hossain et al., 2020). Salts can degrade soil structure by causing the expansion 518 and dispersion of clays, blocking soil pores through which water and oxygen circulate (Sultan, 519 2006). On the other hand, coarser textures associated to a better soil drainage led to a lower salt 520 content (Hossain et al., 2020). These findings are in agreement with the results of our study, where 521 20 clay had a significant positive relationship with ECe, while sands had a negative relationship with 522 ECe (table 4). 523 4.3 The role of environmental factors on soil salinity in a semi-arid environment 524 Based on the positive results of other studies (Sultanov et al., 2018; Peng et al., 2019; Xu et al., 525 2020; Wang et al., 2020; Wang et al., 2021), several spatial information layers related to 526 topography and geomorphological features were elaborated in our research (Table 2). Landform 527 indices, altitude and slope, which can explain the general distribution of ephemeral channels that 528 flow into the Mar Menor coastal lagoon as non-permanent courses of water activated after intense 529 rainfall events, were not significant in the statistical correlation analyses with ECe. 'Solar', which 530 provides average values of solar radiation as a parameter associated with the topographically 531 determined balance between evapotranspiration and soil moisture (Akramkhanov and Vlek, 532 2012), was also poorly correlated. Only the altitude values, expressed from 'DTM', were 533 significant in the PCA results, suggesting that altitude intervenes as an indirect factor. That is, its 534 contribution would be associated with the geographical position conditioned by other 535 environmental aspects, rather than the single fact of its altitudinal position (Wang et al., 2020). 536 The lack of effect of these topographical variables on soil salinity raises interesting questions for 537 discussion. Clearly, one of the possible reasons may be due to the structure of the analysed plots, 538 coupled with the geomorphological homogeneity of the study area. But a major driver in spatial 539 analysis is the scale factor, understanding the mechanisms occurring in a system can be affected 540 by dependence on the level of representation of the environment being examined (Ruiz-Navarro 541 et al., 2012). Therefore, studying the linkages between soil attributes and their environment at 542 different spatial resolutions will allow a better comprehension of which landscape processes 543 control soil salinity and whether this control is affected by the resolution of the landscape 544 representation (Welle and Mauter, 2017). As already described in this work, many studies have 545 been reported on precision agriculture using remote sensing techniques, in which different 546 platforms (satellite, aircraft or UAV) and sensors are used to detect and monitor the physical 547 characteristics of crops and soil. Hyperspectral or multispectral information, medium, high or 548 very high resolutions have demonstrated to be efficient at detecting soil salinity processes. 549 21 However, substantial issues remain that need to be resolved to better understand the interaction 550 of large-scale processes, dominated by natural factors, and processes at finer scales, where 551 anthropogenic factors are involved more intensely. 552 Especially in arid and semi-arid environments, the movement and deposition of salts on the soil 553 surface is controlled by geomorphic, climatic, hydrological and edaphic factors on a large scale, 554 affecting the soil-water balance (Masoud et al., 2019; Wang et al., 2020); factors that have been 555 identified in the results achieved in the PCA (Figure 3). Soil clay content, followed by land surface 556 temperature (TIRS1), have a strong weight in component 1. Both factors exert a significant 557 influence at catchment scale, while the effect of salinity indices (SI and NDSI) exhibit a lower 558 representativeness at this scale, in contrast to the relevance shown in their correlation with ECe 559 (Table 6). These indices based on wavelengths in the red and near-infrared range provide a 560 relationship to vegetation cover. The increased heterogeneity of plant species on a larger scale, as 561 opposed to the uniformity of citrus crops analysed at plot scale, may interfere with their detection 562 capability. In addition, differences reported in other works on relative salinity tolerance among 563 various plants classes, as well as varying types of salinity, often complicate salinity monitoring 564 (Allbed and Kumar 2013; Zhang et al., 2015; Yu et al., 2018). This first component contributed 565 40.6% of the variance in our data, while component 2 determined 26.2% of the variance, 566 associated with soil-related factors. Bedrock' and 'SMI' were mainly involved, although 'Bedrock' 567 seems to be influencing in an overestimated form outside the crop areas. In both cases, a linkage 568 to the potential water content of the soil is apparent. This factors directly affects ECe, as well as 569 physical properties of soil and vegetation, also related to ECe (Corwin and Scudiero, 2019); hence 570 in future works it would be important to estimate soil moisture content, which can be varied for 571 different land uses, as well as in depth (Yu et al., 2018). In total, combining the two eigenvectors 572 in the PCA can only explain 66.8% of the variance of the dataset, because the rest of the 573 components provide redundant or irrelevant information as they do not exceed the value of 1 in 574 'SS loadings'. 575 This may suggest that others drivers have not been taken into account, such as factors related to 576 anthropogenic activities (Zhang et al., 2011b). Campo de Cartagena, as is the case along the 577 22 Mediterranean coast, has become an area of economic development, affected by the impact of 578 human activities (Newton et al., 2020; Fernández-García et al., 2021). Anthropogenic factors 579 affecting soil salinity occur on a smaller scale and often in a localised manner, in contrast to 580 natural factors involving large-scale processes. Both factors should be analysed in combination 581 to provide more realistic results (Welle and Mauter, 2017). In agriculture, impacts from socio582 economic and policy factors such as population density and location of settlements, large-scale 583 land use changes, overgrazing, deforestation, intensive or inappropriate use of land and water 584 resources were the main drivers of salinisation differences at the basin level (Seydehmet et al., 585 2018a; 2018b; Chi et al., 2019; Wanget al., 2020; Fernández-García et al., 2021). More 586 concretely, the management of the farming system is an essential element in the processes of 587 secondary salinisation. In different studies, irrigation and drainage infrastructure distances, 588 fertiliser use, cultivated area and cropping pattern, and livestock numbers have been reported as 589 variables influencing salinity (Akramkhanov and Vlek, 2012; de Lima et al., 2017; Ivushkin et 590 al., 2018; Wang et al., 2020); issues to consider, especially if research is conducted on local scale. 591 592 5 Conclusion 593 The sustainability of agriculture irrigated with degraded water will require a comprehensive 594 approach to soil, water, and crop management consisting of site-specific preventive measures and 595 management strategies. 596 In this study, applied spectral indices (SI and NDSI) have been found to be useful at characterizing 597 soil salinity in citrus orchards in the semi-arid environment of south eastern Spain. Other data that 598 have characterized different environmental variables have also been important, such as soil 599 moisture (SMI), the content of clay chemicals in the soil (CMI), the temperature values of the 600 citrus leaf cover (TIRS1), or the textural composition of the soil (Soil clay and Soil sand). The set 601 of this environmental characterization has provided useful information to determine the variables 602 that primarily influence salinity in crops in these semi-arid conditions. Although the data extracted 603 from the ACP, indicate that to apply soil salinity predictive models, other variables related to 604 23 agricultural management practices and a better characterization of the irrigation water supplied, 605 must be incorporated. 606 Therefore, the long-term feasibility of using degraded water to irrigate citrus and also their 607 capacity to adapt to the semi-arid conditions must be performed cautiously as it will only be 608 successful with irrigation management measures and intensive monitoring. 609 610 Fundings 611 This work was supported by the research project ‘Use of Advanced information technologies for 612 Site-Specific management of Irrigation and SaliniTy with degraded water’ (ASSIST) funded by 613 SENECA Foundation on the Regional Program "SAAVEDRA FAJARDO", and the Project SHui 614 which is co-funded by the European Union Project GA 773903 and the Chinese MOST. 615 616 617 618 619 References 620 Akramkhanov A, Vlek PL. (2012). The assessment of spatial distribution of soil salinity risk using 621 neural network. Environ Monit Assess. 184(4):2475-85. doi: 10.1007/s10661-011-2132-5. 622 Alcon, F.; Tapsuwan, S.; Brouwer, R.; de Miguel, M. D. (2014). Adoption of irrigation water 623 policies to guarantee water supply: A choice experiment. Environmental Science & Policy, 624 44: 226-236. 625 Alexakis, D.D.; Daliakopoulos, I.N.; Panagea, I.S.; Tsanis, I.K. (2016). Assessing soil salinity 626 using WorldView-2 multispectral images in Timpaki, Crete, Greece. Geocarto Int. 33, 321– 627 338. http://dx.doi.org/10.1080/10106049.2016.1250826 628 Allbed, A.; Kumar, L. (2013) Soil salinity mapping; monitoring in arid; semi-arid regions using 629 remote sensing technology: A review. Adv. Remote Sens. 5: 43–52. 630 24 Asfaw, E.; Suryabhagavan, K.; Argaw, M. (2018). Soil salinity modeling and Mapping Using 631 Remote sensing and GIS: The case of Wonji sugar cane irrigation farm, Ethiopia. 17: 250632 258. 10.1016/j.jssas.2016.05.003. 633 Ayers, R.S., Westcot, D.W., (1985). Water quality for agriculture. Irrig. Drain. FAO, Rome, 174 634 pp. 635 Bastida, F.; Abadía, J.; García, C.; Torres, I.; Ruiz Navarro, A.; José Alarcón, J.; Nicolás, E. 636 (2017). The impacts of irrigation with transferred and saline reclaimed water in the soil 637 biological quality of two citrus species: Adaptations to low water availability. In EGU 638 General Assembly Conference Abstracts (p. 12437). 639 Boman, B.J., Stover, E.W., (2008). Outline for managing irrigation of Florida citrus with high 640 salinity water. ABE 332. University of Florida IFAS Extension. 641 Burrough, P.A. and McDonnell, R.A., (1998). Principles of geographical information systems. 642 Oxford University Press, Oxford, 327 pp. 643 Chi, Y., Sun, J., Liu, W., Wang, J., & Zhao, M. (2019). Mapping coastal wetland soil salinity in 644 different seasons using an improved comprehensive land surface factor system. Ecological 645 Indicators, 107, 105517. doi:10.1016/j.ecolind.2019.105517 646 CHS (2021). Segura Hydrographic Confederation. Retrieved from 647 https://www.chsegura.es/es/cuenca/redes-de-control/piezometria/series-piezometria/ 648 CNIG (2021). National Centre for Geographic Information. Retrieved from 649 https://centrodedescargas.cnig.es 650 Congedo, L. (2020). Semi-Automatic Classification Plugin Documentation. DOI: 651 http://dx.doi.org/10.13140/RG.2.2.25480.65286/1 652 Contreras S, Pérez-Cutillas P, Santoni CS, Romero-Trigueros C, Pedrero F, Alarcón JJ. (2014) 653 Effects of reclaimed waters on spectral properties and leaf traits of citrus orchards. Water 654 Environ Res. 86(11):2242-50. doi: 10.2175/106143014x14062131178637. 655 Contreras, S., J.E. Hunink, and A. Baille (2014), Building a watershed information system for the 656 Campo de 848 Cartagena basin (Spain) integrating hydrological modeling and remote 657 sensing. Report FutureWater: 125. 849 Cartagena, Spain, 59pp. 658 25 Cooper, W.C., Peynado, A., (1959) Chloride and boron tolerance of young-line citrus trees on 659 various rootstocks. J. Rio Grande Valley Hortic. Soc. 13, 89–96. 660 Corwin, D.L., (2020), Climate change impacts on soil salinity in agricultural areas. Eur. J. Soil 661 Sci. 72. 662 De Lima, Y.B., da Silva Sá, F.V., Neto, M.F., de Paiva, E.P., Gheyi, H.R. (2017). Accumulation 663 of salts in the soil and growth of cowpea under salinity and phosphorus fertilization. Revista 664 Ciencia Agronomica, 48 (5), pp. 765-773. 665 Dehni, A.; Lounis, M. (2012). Remote Sensing Techniques for Salt Affected Soil Mapping: 666 Application to the Oran Region of Algeria, Procedia Engineering, 33: 188-198. 667 https://doi.org/10.1016/j.proeng.2012.01.1193. 668 Dilts, T.E. (2015) Topography Tools for ArcGIS 10.1. University of Nevada Reno. Available at: 669 http://www.arcgis.com/home/item.html?id=b13b3b40fa3c43d4a23a1a09c5fe96b9 670 Domsch, H., Giebel, A. (2004). Estimation of soil textural features from soil electrical 671 conductivity recorded using the EM38. Precision Agriculture, 5 (4), pp. 389-409. 672 Du, J.-L., Jin, M.-G., Ouyang, Z.-P., Liu, Y.-F. (2008). Characteristics of soil salinity profiles and 673 relationship between salinity and soil particle composition in Yanqi basin of Xinjiang, 674 China. Diqiu Kexue - Zhongguo Dizhi Daxue Xuebao/Earth Science - Journal of China 675 University of Geosciences, 33 (1), pp. 131-136. 676 ESRI (2007). ArcHydro Tools for ArcGIS 9.2Tutorial published by ESRI Press, Redlands CA 677 May 2007. 678 Fernández-Ballester, G., García-Sánchez, F., Cerdá, A., Martínez, V., (2003). Tolerance of citrus 679 rootstock seedlings to saline stress based on their ability to regulate ion uptake and 680 transport. Tree Physiol. 23, 265–271. 681 Fernández-García, M.F., Custodio, E., Llamas, M.R. (2021). Ethical Aspects of Water Use in the 682 Campo de Cartagena and the Associated Impacts on the Mar Menor. Advances in Science, 683 Technology and Innovation, pp. 125-127. 684 Fitzpatrick R. W., Merry R., and Cox J. (2000). What are saline soils and what happens when 685 they are drained? J. Austral.Assoc. Nat. Res. Manage. (AANRM) 6, 26 – 30. 686 32 Romero-Trigueros, C., Nortes, P.A., Pedrero, F., Mounzer, O., Alarcón, J.J., Bayona, J.M., 849 Nicolás, E., (2014). Assessment of the viability of using saline reclaimed water in 850 grapefruit in medium to long term. Span. J. Agric. Res. 12 (4), 1137–1148. https:// 851 doi.org/10.5424/sjar/2014124-5495. 852 Romero-Trigueros, C., Parra, M., Bayona, J.M., Nortes, P.A., Alarcón, J.J., Nicolás, E., (2017b). 853 Effect of deficit irrigation and reclaimed water on yield and quality of grapefruits at harvest 854 and postharvest. LWTFood Sci Technol. 85, 405–411. 855 Ruiz-Navarro, A., Barberá, G.G., García-Haro, J., Albaladejo, J. (2012). Effect of the spatial 856 resolution on landscape control of soil fertility in a semiarid area. Journal of Soils and 857 Sediments, 12 (4), pp. 471-485. 858 Saey, T., Van Meirvenne, M., Vermeersch, H., Ameloot, N., Cockx, L. (2009). A pedotransfer 859 function to evaluate the soil profile textural heterogeneity using proximally sensed apparent 860 electrical conductivity. Geoderma, 150 (3-4), pp. 389-395. 861 Scudiero, E.; Skaggs, T.; Corwin. D. (2015). Regional-scale soil salinity assessment using 862 Landsat ETM+ canopy reflectance. Remote Sensing of Environment, Volume 169, Pages 863 335-343. https://doi.org/10.1016/j.rse.2015.08.026. 864 Seydehmet, J., Lv, G., Nurmemet, I., Aishan, T., Abliz, A., Sawut, M., … Eziz, M. (2018a). 865 Model Prediction of Secondary Soil Salinization in the Keriya Oasis, Northwest China. 866 Sustainability, 10(3), 656. doi:10.3390/su10030656 867 Seydehmet, J., Lv, G.-H., Abliz, A., Shi, Q.-D., Abliz, A., & Turup, A. (2018b). Irrigation Salinity 868 Risk Assessment and Mapping in Arid Oasis, Northwest China. Water, 10(7), 966. 869 doi:10.3390/w10070966 870 Soto-García, M., Martínez-Alvarez, V., García-Bastida, P. A., Alcon, F., & Martin-Gorriz, B. 871 (2013). Effect of water scarcity and modernisation on the performance of irrigation districts 872 in south-eastern Spain. Agricultural Water Management, 124: 11-19. 873 Sultanov, M., Ibrakhimov, M., Akramkhanov, A., Bauer, C. & Conrad, C. (2018). Modelling End874 of-Season Soil Salinity in Irrigated Agriculture Through Multi-temporal Optical Remote 875 33 Sensing, Environmental Parameters, and In Situ Information. PFG 86, 221–233. 876 https://doi.org/10.1007/s41064-019-00062-3 877 Tack, J., Singh, R.K., Nalley, L.L., Viraktamath, B.C., Krishnamurthy, S.L., Lyman, N. and 878 Jagadish, K.S., 2015. High vapor pressure deficit drives salt‐stress‐induced rice yield losses 879 in India. Global change biology, 21(4), pp.1668-1678. 880 Tian, F., Hou, M., Qiu, Y., Zhang, T., Yuan, Y. (2020). Salinity stress effects on transpiration and 881 plant growth under different salinity soil levels based on thermal infrared remote (TIR) 882 technique. Geoderma, 357, art. no. 113961 883 Touchette BW, Adams EC, Laimbeer P (2012) Agespecific responses to elevated salinity in the 884 coastal marsh plant black needlerush (Juncus roemerianus Scheele) as determined through 885 polyphasic chlorophylla fl uorescence transients (OJIP). Mar Biol 159:2137–2147 886 Tounekti T, Abreu ME, Khemira H, Munné-Bosch S (2012) Canopy position determines the 887 photoprotective demand and antioxidant protection of leaves in salt-stressed Salvia offi 888 cinalis L. plants. Environ Exp Bot 78:146–156 889 USGS (2021). Landsat-8 image courtesy of the U.S. Geological Survey. Retrieved from 890 https://earthexplorer.usgs.gov/ 891 UN-Water, 2020: UN-Water Analytical Brief on Unconventional Water Resources. Geneva, 892 Switzerland. Available at: https://www.unwater.org/app/uploads/2020/08/UN-Water893 Analytical-Brief-on-Unconventional-Water-Resources.pdf (Accessed on 10 July 2020) 894 Vigo, C.; Therios, I. N.; Bosabalidis, A. M. (2005). Plant Growth, Nutrient Concentration, and 895 Leaf Anatomy of Olive Plants Irrigated with Diluted Seawater, Journal of Plant Nutrition, 896 28:6, 1001-1021, DOI: 10.1081/PLN-200058897 897 Wang, F.; Yang, S.; Wei, Y.; Shi, Q.; Ding, J. (2021). Characterizing soil salinity at multiple 898 depth using electromagnetic induction and remote sensing data with random forests: A case 899 study in Tarim River Basin of southern Xinjiang, China. Science of The Total Environment, 900 Volume 754, 142030. https://doi.org/10.1016/j.scitotenv.2020.142030. 901 Wang, N.; Xue, J.; Peng, J.; Biswas, A.; He, Y.; Shi, Z. (2020). Integrating Remote Sensing and 902 Landscape Characteristics to Estimate Soil Salinity Using Machine Learning Methods: A 903 34 Case Study from Southern Xinjiang, China. Remote Sensing. 12(24):4118. 904 https://doi.org/10.3390/rs12244118 905 Weiss, A. (2001). Topographic position and landforms analysis. In Poster presentation, ESRI 906 user conference, San Diego, CA (Vol. 200). 907 Welle, P.D., Mauter, M.S. (2017). High-resolution model for estimating the economic and policy 908 implications of agricultural soil salinization in California. Environmental Research Letters, 909 12 (9), art. no. 094010. 910 World Bank. (2017). Beyond Scarcity: Water Security in the Middle East and North Africa. 911 Overview booklet. World Bank, Washington, DC. License: Creative Commons Attribution 912 CC BY 3.0 IGO 913 World Economic Forum (WEF). (2020). The Global Risks Report 2020. 14th Edition. Geneva: 914 World Economic Forum. Available at: https://www.weforum.org/reports/the-global-risks915 report-2020 (Accessed on 10 July 2020) 916 Xu, H.; Chen, C.; Zheng, H.; Luo, G.; Yang, L.; Wang, W.; Wu, S. & Ding, J. (2020). AGA917 SVR-based selection of feature subsets and optimization of parameter in regional soil 918 salinization monitoring. International Journal of Remote Sensing, 41:12, 4470-4495, DOI: 919 10.1080/01431161.2020.1718239 920 Xu, Y., Smith, S.E., Grunwald, S., Abd-Elrahman, A., Wani, S.P. (2017). Evaluating the effect 921 of remote sensing image spatial resolution on soil exchangeable potassium prediction 922 models in smallholder farm settings. Journal of Environmental Management, 200, pp. 423923 433. 924 Yang, J.-S., Yao, R.-J., Zou, P., Liu, G.-M. (2008). Characteristics of soil salinity profiles and 925 their electromagnetic response under various vegetation types in coastal saline area. 926 Chinese Journal of Applied Ecology, 19 (10), pp. 2117-2124. 927 Yu, H.; Liu, M.; Du, B.; Wang, Z.; Hu, L.; Zhang, B. (2019). Mapping Soil Salinity/Sodicity by 928 using Landsat OLI Imagery and PLSR Algorithm over SemiaridWest Jilin Province, China. 929 Sensors, 107, 105517. http://dx.doi.org/10.3390/s18041048 930 35 Yu, P.; Liu, S.; Yang, H.; Fan, G.; Zhou, D. (2018). Short-term land use conversions influence 931 the profile distribution of soil salinity and sodicity in northeastern China, Ecological 932 Indicators, 88:79-87. https://doi.org/10.1016/j.ecolind.2018.01.017. 933 Zhang, J, Zhang, Z, Chen, J, et al. (2021). Estimating soil salinity with different fractional 934 vegetation cover using remote sensing. Land Degrad Dev. 32: 597– 612. 935 https://doi.org/10.1002/ldr.3737 936 Zhang, J., Han, W., Huang, L., Zhang, Z., Ma, Y., Hu, Y. (2016). Leaf chlorophyll content 937 estimation of winter wheat based on visible and near-infrared sensors. Sensors 938 (Switzerland), 16 (4), art. no. 437, 11 p. 939 Zhang, T.; Qi, J.; Gao, Y.; Ouyang, Z.; Zeng, S.; Zhao. B. (2015). Detecting soil salinity with 940 MODIS time series VI data. Ecological Indicators 52, 480-489 941 Zhang, T.; Zeng, S.; Gao, Y.; Ouyang, Z.; Li, B.; Fang, C.; Zhao, B. (2011a). Using hyperspectral 942 vegetation indices as a proxy to monitor soil salinity. Ecol. Indic. 11, 1552–1562. 943 Zhang, T.; Zeng, S.; Gao, Y.; Ouyang, Z.; Li, B.; Fang, C.; Zhao, B. (2011b). Assessing impact 944 of land uses on land salinization in the Yellow River Delta, China using an integrated and 945 spatial statistical model. Land use policy 28 (4), 857-866 946 Zhao-Liang Li, Hua Wu, Ning Wang, Shi Qiu, José A. Sobrino, Zhengming Wan, Bo-Hui Tang 947 & Guangjian Yan. (2013). Land surface emissivity retrieval from satellite data, 948 International Journal of Remote Sensing, 34:9-10, 3084-3127, DOI: 949 10.1080/01431161.2012.716540 950 Zhou, D.; Lin, Z.; Liu, L.; Zimmermann, D. (2013). Assessing secondary s1oil salinization risk 951 based on the PSR sustainability framework. J. Environ. Manag. 128, 642–654. 952 Zhu, K., Sun, Z., Zhao, F., Yang, T., Tian, Z., Lai, J., Zhu, W., Long, B. (2021). Relating 953 hyperspectral vegetation indices with soil salinity at different depths for the diagnosis 954 ofwinter wheat salt stress. Remote Sensing, 13 (2), art. no. 250, pp. 1-21. 955 956 957 958 959 960 36 961 962 963 964 965 966 967 968 969 970 971 972 973 Table 1. Citrus farm characteristics description 974 975 Farm Irrigation water applied (mm year -1) Crop Variety Rootstock Age (years) Size (has) 1 573 lemon verna macrophyla 30 100 123 mandarin murco carrizo 2 10 2 354 lemon eureka macrophyla 7 11 3 380 lemon eureka macrophyla 7 20 4 310 lemon eureka macrophyla 6 10 5 626 lemon fino macrophyla 30 24 6 594 orange furel carrizo 16 16 7 570 mandarin fortuna orri carrizo 25 19 8 645 orange parfil creopatra 12 14 9 580 mandarin murco macrophyla 10 12 976 977 978 979 980 981 982 983 984 Table 2. Salinity indices, spectral indices and environmental data description used on the 985 experiment. 986 987 Data Description Units Pixel size (m) Source Tabla con formato 37 ECe A Electrical Conductivity of the saturated soil Extract in dry soil mS·cm-1 - Own elaboration ECe B Electrical Conductivity of the saturated soil Extract in wet soil mS·cm-1 - Own elaboration SI Salinity Index from Landsat8 (OLI) Dimensionless 30 Own elaboration NDSI Normalized Difference Salinity Index from Landsat8 (OLI) Dimensionless 30 Own elaboration VSSI Vegetation Soil Salinity Index from Landsat8 (OLI) Dimensionless 30 Own elaboration TIRS1 B1 from Landsat 8 Thermal Infrared Sensor (TIRS) °C 100 Own elaboration SMI Soil Moisture Index from Landsat8 (OLI) Dimensionless 30 Own elaboration CCI Clay Chemical Index from Landsat8 (OLI) Dimensionless 30 Own elaboration Soil clay Soil clay content % 210 soilgrids.org Soil silt Soil silt content % 210 soilgrids.org Soil sand Soil sand content % 210 soilgrids.org OC Organic Carbon (Dry Combustion) g/kg 210 soilgrids.org Bedrock Depth to bedrock cm 210 soilgrids.org Water table Piezometric levels in 2020 m 50 Own elaboration DTM Digital terrain model m 5 centrodedescargas.cnig.es Slope Slope made from DTM Degrees 5 Own elaboration SII Solar Illumination Index made from DTM Dimensionless 5 Own elaboration LC Landform Classifcation made from DTM Ordinal classes 5 Own elaboration SPC Slope Position Classification made from DTM Ordinal classes 5 Own elaboration TWI Topographic Wetness Index made from DTM Dimensionless 5 Own elaboration TPI Topographic Position Index made from DTM Dimensionless 5 Own elaboration 988 989 990 991 992 993 994 995 996 997 998 Table 3. Descriptive statistics of environmental data information and ECe samples collected in 999 the ground. 1000 1001 Data N Mean Median Std. Dev. Minimum Maximum ECe A 80 3.56 2.87 ±2.10 1.23 9.60 ECe B 80 5.63 4.55 ±3.55 1.49 16.7 38 SI 80 0.169 0.169 ±0.025 0.128 0.228 NDSI 80 -0.526 -0.513 ±0.122 -0.725 -0.241 VSSI 80 -1.88 -1.85 ±0.154 -2.45 -1.55 TIRS1 80 20.8 20.5 ±0.954 19.1 23.0 SMI 80 0.981 0.981 ±0.004 0.966 0.987 CCI 80 1.64 1.51 ±0.272 1.18 2.22 Soil clay 80 26.2 26.0 ±2.00 22 30 Soil silt 80 32.3 32.0 ±1.85 29 36 Soil sand 80 40.4 40.0 ±2.95 36 47 Organic Carbon 80 2.41 2.24 ±0.495 1.72 4.13 Bedrock 80 1031 1246 ±377 140 1428 Water table 80 25.5 23.5 ±10.8 9.95 55.4 Solar 80 140 140 ±0.318 139 140 DTM 80 62.3 52.2 ±32.5 17.3 148 Slope 80 1.83 1.29 ±1.6 0.078 7.33 TWI 80 163 25.6 ±394 0 2512 TPI 80 -0.143 0.006 ±0.754 -3.99 2.17 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 39 1028 1029 1030 1031 1032 1033 1034 Table 4. Correlation coefficients of the environmental variables and the ECe values. 1035 Data Correlation coef. A_ECe B_ECe Pearson's r p-value Pearson's r p-value SI Pearson's r 0.449*** < .001 0.489*** < .001 NDSI Pearson's r 0.568*** < .001 0.639*** < .001 TIRS1 Pearson's r 0.498*** < .001 0.550*** < .001 SMI Pearson's r -0.330** 0.003 -0.338** 0.002 CCI Pearson's r -0.409*** < .001 -0.454*** < .001 Soil clay Pearson's r 0.456*** < .001 0.462*** < .001 Soil silt Pearson's r 0.031 0.787 -0.256* 0.022 Soil sand Pearson's r -0.343** 0.002 -0.151 0.181 Bedrock Pearson's r -0.251* 0.024 -0.425*** < .001 SPC Kendall's Tau B -0.209* 0.019 -0.140 0.116 Note. * p < .05, ** p < .01, *** p < .001 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 40 1058 1059 1060 1061 1062 1063 1064 Table 5. One-Way ANOVA (Welch's) of the environmental variables and the ECe values. 1065 Data F df1 df2 p ECe A 13.6 8 28.5 < .001 ECe B 15.4 8 29.3 < .001 SI 46.5 8 27.0 < .001 NDSI 52.8 8 27.2 < .001 TIRS1 31.4 8 28.2 < .001 SMI 32.2 8 26.7 < .001 CCI 25.9 8 25.6 < .001 Soil clay 44.0 8 27.9 < .001 Soil silt 27.9 8 27.9 < .001 Soil sand 349.3 8 27.2 < .001 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 41 1091 1092 1093 1094 1095 1096 Table 6. Component loadings on the factors obtained in the PCA. 1097 Component 1 2 Uniqueness ECe A 0.657 0.538 ECe B 0.747 0.423 SI 0.601 0.202 NDSI 0.838 0.134 TIRS1 0.860 0.256 Soil clay 0.790 0.371 Bedrock -0.420 0.722 SMI -0.827 0.192 CCI -0.702 0.220 DTM 0.825 0.265 ACP 'varimax' rotation was used 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 48 Table 9. Soil salinity (EC. dS m-1) and chemical compositions ([Ca], [K], [Mg], [Na], meq l-1) between furrows (dry area) (A) and on the ridges (wet area) (B) 1213 from the citrus farm studied 1214 1215 Farm Soil samples Ca (meq/L) K (meq/L) Mg (meq/L) Na (meq/L) CEe 1 A 20,61 1,51 5,61 7,31 1,81 B 21,13 3,15 9,50 15,56 4,43 2 A 11,21 2,25 3,10 3,54 0,95 B 20,14 2,14 5,45 8,20 3,62 3 A 32,21 4,25 21,10 28,52 4,51 B 8,74 2,55 10,41 9,21 10,21 4 A 40,92 4,68 16,34 37,75 6,03 B 41,48 1,93 11,13 8,67 12,35 5 A 17,43 9,26 8,54 21,95 5,32 B 9,95 18,95 10,57 35,35 3,78 6 A 25,20 10,00 2,45 2,58 3,25 B 12,21 1,98 3,65 6,56 3,62 7 A 12,25 1,25 2,15 2,36 0,85 B 18,25 1,89 3,65 4,58 2,26 8 A 23,32 3,89 17,23 30,85 5,89 B 14,25 1,45 9,85 8,11 9,54 9 A 36,25 5,45 22,41 40,23 8,20 B 54,25 2,32 12,25 8,52 11,2 1216