In situ measurements
There are many methods for measuring SM (Robock et al., 2000; Walker et al., 2004). The main traditional methods are detailed below.
The gravimetric method is the only one that can measure SM directly (Robock et al., 2000). It consists of taking a soil samples and you putting them in a cylinder of a known weight and volume, weighing them and then drying them by placing them in an oven at a temperature of 105 °C until the weight is constant, usually after 48 hours, and re-weigh a second time in order to deduce the mass moisture of each sample. The difference in mass between the two weighings corresponds to the volume of water evaporated. The knowledge of the soil apparent, which corresponds to its mass per unit volume of the dry soil in place, allows us to determine the volume water content of the system.
Despite the indispensability of this method to calibrate other measurement methods, it nevertheless has drawbacks: long to implement, does not integrate large areas, destructive of the soil, especially if we plan to realize a profile of moisture over several meters of soil.
According to (Robinson et al., 2003), at the moment, the majority of SM measurements are made using electromagnetic techniques from capacitive probes that use time or frequency domain reflectometry (TDR, FDR). A capacitive probe set up in ground, behaves like a resistance-inductance-capacitance (RLC) circuit whose resonance frequency can be determined, which permits to calculate the capacitance (C) and to deduce the soil relative permittivity (εr). This method allows to have SM measurements over a long time period with high temporal resolution. In addition, the acquisition of these data can be automated thanks to a data-logger. The electromagnetic techniques are non-destructive, non-radioactive and non-expensive (depends on the measurement number).
The knowledge of the soil relative permittivity variation makes it possible to know the soil volume moisture using a relation SM f(εr). This relation varies from one soil type to another.
It is therefore necessary to calibrate, using the gravimetric measurements, the measurements for each soil type in order to match the measurement (in mV) with actual SM. The accuracy is about 2% of SM volume. Those moisture probes provide only punctual spatially measurements. The in situ SM data used throughout this manuscript comes from the TDR and Theta-Prob (measures the dielectric constant of soil) sensors calibrated using the gravimetric technique for each field.
The TDR method is based on the fact that water has a dielectric constant (= 80) much higher than that of air and minerals (air = 1 and 2 <mineral <7). TDR measurements are based on the propagation of an electromagnetic wave. It involves measuring the propagation time of an electromagnetic wave before reflection at the end of a waveguide installed in the ground, generally consisting of a set of 2 to 4 parallel metallic needles. The propagation time permits to calculate its velocity (v) which is related to the real and imaginary parts of the relative permittivity of the ground.
Thetaprobe measures the volumetric water content of soils. A high frequency wave (100 MHz) is applied along the electrodes. The difference between the emitted and the reflected wave by the ground is a function of the dielectric properties of the surface, and therefore of the water content.
It is one of the easiest method to use and gives accurate measurements in real time despite the need of precautions of handling (Hillel, 1998). The operating principle is based on measuring the amount of slow neutrons reflected in a volume of soil surrounding the radioactive source. The fast neutrons emitted by the probe are progressively slowed down by the ground. The slow neutron flux is proportional to the density of hydrogen atoms at the ground, with most of the hydrogen atoms belong to the water molecules. The relationship between the amount of neutrons detected and the water content of the soil needs to be calibrated in each experimental situation. The volume of soil prospected by radiation (sphere of influence) is about 40 cm in radius (Daudet and Vachaud, 1977).
Conveniently, the neutron probe is placed on an aluminum tube placed in the ground. This tube, is installed once for all on a given site. A detector placed near the source counts the number of slow neutrons returning to the source. The calibration of these probes should be done for each soil type and for a period of time with different SM values. The neutronic method requires a skilled labor and precautions of handling must be taken, due to radioactive materials used. In addition, the cost of these probes is high. Note that these probes cannot be used for frequent and automatic measurements.
Remotely sensed approaches
At the local scale and at a given time, the SM estimation is relatively easy with the methods mentioned above. However, to have representative measurements of a large area, the procedure is already complex because it involves a dedicated sampling strategy. Among these methods is remote sensing.
Soil moisture indices
Nowadays, remote sensing provides a prevailing method and approach for monitoring the spatio-temporal variations and quantitative estimations of SM. SM is difficult to be accurately evaluated, due to its strong spatial and temporal variability, resulting primarily from the variations in soil type, local topography, and land use. Many studies have attempted to establish different methods for retrieving SM based on the relationships between the SM and satellite derived land surface parameters. When such relationships are indirect, a SM index is often derived instead of the SM in absolute value. Therefore, several factors including LST and vegetation status can be indirectly used for SM estimation. Numerous indices are detailed in the literature based on remote sensing data derived from optical, thermal and microwave bands.
Other studies have been carried out, establishing a link between SM in the soil profile and drought. To this aim different indices have been developed to quantify droughts. The most commonly used drought indices are based on precipitation measurements: the Palmer Drought Severity Index (PDSI, Palmer, 1965), the Rainfall Anomaly Index (RAI, Van Rooy, 1965), the National Rainfall Index (NRI, Gommes and Petrassi, 1994), the Standardized Precipitation Index (SPI, Guttman, 1999; Mckee et al., 1993; McKee et al., 1995). Concerning drought indices based on SM estimations, only a limited number of studies have been made, e.g. the drought index called Soil Moisture Drought Index (SMDI, Hollinger et al., 1993).
These indices are mainly based on the reflectances to estimate SM. The reflectance images are available at different resolutions which makes the reflectance domain more operational. In the near-infrared band, SM can be derived from the vegetation traits that occur under water stress. The vegetation indices are sensitive to water stress, which makes them widely used to detect the drought condition.
Among those indices, (Kogan, 1995, 1990) established the vegetation condition index (VCI) which is based on the statistical Normalized Difference Vegetation Index (NDVI) time series normalized by its maximum (NDVImax) and minimum (NDVImin) values. The NDVImin and NDVImax are estimated are estimated on NDVI large time series’s. The VCI index was used as a drought index. This index has the potential to remove the influences of weather and geographic location unlikely to NDVI. VCI is estimated as:
Some research evidenced that the shortwave infrared (SWIR) region could provide better results in order to detect soil water content. Gao, (1996) proposed a Normalized Difference Water Index (NDWI, Eq. I.2) using the reflectance at 1.24 μm and 0.86 μm. The NDWI is derived from the near infrared (NIR) water absorption bands. A low NDWI value indicates canopy water stress, dry vegetation and/or bare soil (both dry and wet). On the contrary, a high NDWI index value indicates green and healthy vegetation.
Based on the NDWI, normalized multi-band drought index (NMDI) uses the NIR centered approximately 0.86 μm channel as the reference; instead of using a single liquid water absorption channel, however, it uses the difference between two liquid SWIR water absorption channels centered at 1.64 μm and 2.13 μm as the soil and vegetation moisture sensitive band. Strong differences between two water absorption bands in response to soil and leaf water content give this combination a potential to estimate SM for both bare soil and vegetated areas. Different applications have demonstrated that NMDI has the potential to provide a quick response to SM changes (Wang et al., 2010, 2008). The NDMI is calculated as:
For thermal infrared wavebands, SM is derived from the parameters related to the soil thermal properties. The thermal inertia and temperature index methods are two main methods for estimating SM. Both categories are based on the surface temperature variations, which are strongly correlated with the SM. In this subsection we will present the different indices that are based on the LST variations. The LST variable is an indicator of the water stress. Over bare soil LST refers to the soil temperature while over a vegetated area LST indicates the water status of the plant: an increase in the LST (for all other parameters and forcing data remaining unchanged) is a sign that the vegetation is undergoing water stress.
McVicar et al. (1992) developed an index which they called the normalized difference temperature index (NDTI) to reflect the SM status. The index is a normalization of the LST by its endmembers:
The LST∞ and LST0 are the LST endmembers corresponding to a surface impedance (resistance) equal to infinite and zero, respectively. The extremes of LST are obtained from a surface energy balance model at specific atmospheric forcing and surface impedance.
The NDTI index reflects accurately the spatio-temporal variation of SM by eliminating the effect of the seasonal variations of LST. The limitation of this index basically comes from the availability of the atmospheric (solar radiation, air humidity, wind speed…) and vegetation (eg. Leaf area index, LAI) data.
Alternatively, other approaches are based on the relationship between SM and ET and the available energy. High ET indicates the presence of sufficient SM which leads to a lower vegetation temperature, while a weak ET indicates a water deficit in the surface. Therefore, the ratio of actual to potential ET can be used as a proxy for crop water stress:
For a specific atmospheric condition, Moran et al., (1994) proposed the crop water stress (CWSI) based on the surface energy balance model. CWSI is developed as an indicator of the SM by using the vegetation canopy temperature Tc and the air temperature Tair:
where (Tc − Tair)max and (Tc − Tair)min are the differences between the canopy and air temperature without transpiration and at potential ET, respectively. The CWSI was established based on the single canopy energy balance model, which is less effective for early crop growth. In addition, this approach requires meteorological data and the calculation process is complex. Furthermore, the extrapolation methods used for meteorological data, which are mainly obtained from ground weather stations, have important impacts on the accuracy of the CWSI determinations.
The stress index can also be computed based on a combination of thermal and reflectance data. Combining remotely sensed visible and thermal infrared data can provide more information for estimating SM than the single one. It is important to determine how to combine these methods reasonably to obtain highly accurate SM. Among such indexes, the Surface Energy Balance Index (SEBI, Menenti and Choudhury, 1993), Water Deficit Index (WDI, (Moran, 2004; Moran et al., 1994) are different expressions of the stress Factor, and have been derived from a surface energy balance model, based on the same theory as the CWSI. The WDI can be used in different surface conditions, like covered area and sparsely covered one. Relying on the surface energy balance principle, the soil adjusted vegetation index (SAVI, Huete, 1988) and the temperature difference form a trapezoidal space, and the index can be directly calculated from remotely sensed data without any on leaf and air temperature measurements.
Some index like Temperature Vegetation Index (TVI, Prihodko and Goward, 1997), Temperature Vegetation Dryness Index (TVDI, Sandholt et al., 2002) and Vegetation Temperature Condition Index (VTCI, Wan et al., 2004; Wang et al., 2004), do not rely on any parameterization of the energy balance and can thus be computed directly from remote-sensing data. The above stress index can be estimated depending on the triangle/trapezoid method. These methods are usually based on the trapezoidal or triangle shape formed between LST and other vegetation variables, the most commonly variables representing the vegetation cover fraction are NDVI or SAVI.
VTCI is defined as a ratio of the dry to actual LST difference to the dry to wet LST temperature difference, with wet/dry LST being estimated as the minimum/maximum LST that the surface can reach for a given meteorological forcing.
Where LSTNDVI.max and LSTNDVI.min are the maximum and the minimum LST of pixels which have same NDVI value in the studied area, respectively. The both extremes temperature are estimated from the dry and the wet edge in the LST-NDVI space. The most important in the image-based method is that, the number of pixels should be sufficient to cover all conditions. In the LST-VI (vegetation index) space the conservation of energy is a key elements, the surface energy maintains balance. In the LST-VI space the wet edges represents an adequate SM and higher ET and the dry edge represents that the vegetation is subjected to water stress in which evapotranspiration reaches the minimum and the SM is minimal.
The above thermal- and shortwave-derived SM index has a wide coverage and a fine spatial resolutions due to the existence of space-borne satellites. The thermal-based index uses the responses of the soil energy balance to soil moisture to determine SM. By contrast, the shortwave-based indexes use characteristic changes in the soil reflectance or vegetation physiology to estimate SM. One of the drawback of these methods is the temporal resolution. The thermal/optical remote sensing SM indices can be inferred only on clear sky days. In addition, using Landsat’s TIR data we will have the data at best every 16 days.
Alternatively, a microwave-based SM index could be used to estimate SM. Microwave data are not affected by clouds, which allows SM to be monitored at high frequency. However, microwave-based SM indexes are easily perturbed by vegetation and surface roughness. Considerable efforts have been made for the characterization of the spatial and temporal variability of SM over vegetated areas. Passive and active microwave-based indices are developed for monitoring SM from space.
Among the active microwave-based indices, SM can be retrieved using multi-date SAR imagery. These indexes are based on change detection techniques for multi-temporal SAR data. Shoshany et al. (2000) presented the normalized radar backscatter soil moisture index (NBMI), which is obtained from the backscatter measurements at two different times (t1 and t2) over the same area.
The advantage of these techniques is that, the difference in backscatter between two dates can be related solely to a change in the dielectric properties of the surface i.e., the NSSM, in cases where surface roughness and vegetation remain unchanged in time.
Another SM index (∆ ) has been developed by Thoma et al, (2004), based on the same assumption as NBMI, that the roughness and the vegetation remain time-invariant over the selected area. The ∆ index is calculated as the difference of the wet and dry backscatter images divided by the reference dry backscatter.
Where the 0 0 are the average backscatter from wet and dry soil, respectively. The extremes of backscatter in wet and dry conditions must be acquired with the same view angle and the same wavelength in order to predict SM accurately and the resulting backscatter changes between repeat passes can therefore be attributed to changes in SM.
An images-difference-based index similar to the delta index, based also on the amplitude of SAR signal was developed to reflect SM variation. Wagner et al., (1999b) proposed a SM index (ms(t)) by normalizing the sigma nought values by the highest and lowest σ0, 0 and 0 . The method compared time series data of σ0 with standard reference incident angle 40° of ERS Scatterometer data. 0 (40, ) was founded to be affected by vegetation, and increases from winter to summer due to vegetation growth, while 0 (40, ) more and less depends on vegetation status. The radar observations lie between these two extreme values and the observations can be converted meaningfully to SM values from 0 to 1 relative SM by using the saturation and wilting point SM.
Esch et al. (2018) used the SM index (SMI) which is calculated similar to ms(t) developed by Wagner et al., (1999b). The SMI is used to estimate SM for each land use group (cereals, sugar beet, and grassland). This simple linear indexing is used, because the relationship between surface SM and radar signals is well described by linear relationships.
Passive microwave data have been also used to predict surface SM. Most passive microwave-derived indexes are based on the brightness temperature TB. Paloscia et al. (2001) used the Polarization Ratio (PR) to retrieve SM using SSM/I passive microwave data (Temimi et al., 2007). The PR index is based on the brightness temperature in both polarizations V and H.
The PR dynamics is mainly linked to SM but it is influenced also by vegetation water content, which affects the SM accuracy because the vegetation water content and the SM have opposite effects on the PR index. This index has been used by Paloscia and Pampaloni. (1984) to detect plant water stress from Ka band, and it has been found that the PR index is correlated to CWSI index with 90% of accuracy.
Gruhier et al. (2008) evaluated the AMSR-E volumetric SM products based on ground measurements. The TB at horizontal and vertical polarizations were used at 6.9 and 10.7 GHz (C and X bands). They found a very low RMSE equal to 0.038 % m3.m-3.
Liu and Shi (2012) established a new passive Microwave Soil Moisture Index (MSMI) based on the TB difference between daytime (ascending cross) and night (descending cross). The MSMI index is similar to PR index, and is calculated as:
Where the TBPAand TBPD are the brightness temperature in ascending and descending pass across the equator, respectively.
The results showed that the MSMI can show the spatial and temporal changes of SM at global scale. When SM is higher, the MSMI is lower, and when SM content is lower, the MSMI is higher.
Taking the advantage of both active and passive microwave techniques that are less disturbed by weather and optical/thermal data that have the potential to remove vegetation disturbance and roughness that affect easily the radar signal, the combination of optical and thermal and microwave remote sensing may have broad application prospects. Additionally, microwave remote sensing can obtain data all-time and all-weather, which can provide great help for SM products over long time series.
Remote sensing has demonstrated a strong potential for estimating the NSSM in the first cm of soil (Bruckler et al., 1988; Du et al., 2000; Engman, 2000) while SM can be estimated using optical/thermal data (Gillies and Carlson, 1995; Sandholt et al., 2002). Several studies have generally acknowledged that microwave techniques have a higher potential for retrieving SM on a regular basis, either from active (Dubois and Engman, 1995a; Ulaby et al., 1979, 1978; Zribi et al., 2005; Zribi and Dechambre, 2003; Akbar et al., 2016; Gorrab et al., 2015; Balenzano et al., 2011; Bousbih et al., 2018, 2017; Gherboudj et al., 2011a) or passive (Akbar et al., 2016; Sabaghy et al., 2018; van der Velde et al., 2014; Kerr et al., 2010; Entekhabi et al., 2010) sensors.
Microwave methods are based on the large difference existing in the dielectric constant between a dry soil (around 4) and that of water, which is around of 80 at microwave frequencies (Ulaby et al., 1986). Radiative transfer models have hence been developed based on the SM dielectric constant relationship (Dobson et al., 1985).
There are two types of microwave sensors, depending on their source of electromagnetic energy: active and passive sensors.
The active microwave technique has its own source of electromagnetic radiation to measure the energy that is reflected and backscattered from its origin. The sensor emits a signal, an electromagnetic wave, with known frequency and polarization towards a target. The receiver records the amount of energy reflected by the target’s surface as well as its polarization and the time traveled by the wave. The amount of energy perceived by the sensor is determined by the amount of energy absorbed by the surface and how the wave is reflected. The fraction of the absorbed signal will be mainly determined by the dielectric constant which is varying according to the surface conditions and especially the soil water content.
Passive sensors measure indirectly SM, by measuring the brightness temperature (TB), which is the temperature that would have a black body if it radiated at the same energy as the gray body. TB depends on the soil dielectric properties, the vegetation water content and the soil temperature. The basis for the passive microwave remote sensing of SM is that the emissivity (ɛ) in the microwave domain is a function of the dielectric constant of the soil-water mixture and therefore of the SM content.
Active systems are more sensitive to the structural characteristics of the surface, such as roughness or canopy structure. The surface roughness conditions the reflection of the wave, which will be specular if the surface is perfectly smooth, or affected by a large dispersion in case of high surface roughness. Among these radars, we quote the so-called SAR radar (Synthetic Aperture Radar) which measures the backscatter coefficients 0 (dB) that expresses the amount of energy received by the sensor after that the electromagnetic wave has come into contact with the target. On the other hand passive systems have a stronger dynamic depending on the SM.
Microwave emission is influenced by different surface conditions (SM, surface roughness, vegetation water content) depending on the frequencies ranging from 1 to 40 GHz (see Table I.3) (Kerr, 1996). The three frequency bands (L, C and X) are monstly used for SM remote sensing. These bands do not react with the same sensitivity to changes in NSSM (Schmugge et al., 1988). The L-band (1-2 GHz) has the most interesting characteristics for detecting surface SM while minimizing disruptive effects (roughness, temperature, etc.). C-band is appropriate in areas of low arid and semi-arid vegetation cover. The Soil Moisture and Ocean Salinity (SMOS) satellite (Kerr et al., 2001) is the first satellite dedicated to the study of low frequency SM. It is a radiometer that operates at L-band (1.4 GHz). It worth to mention that, the first satellite with a special sensor microwave/imager on board was launched in 1987. They operate in Ku and Ka band with a 25 km spatial resolution.
The main objection to remote sensing of SM is that direct measurement is not just about the surface layer. For example, in the X band, the first few mm are probed. For the L-band, 4-5 cm are probed on average and it can reach up to 15 cm depending on the soil characteristics and condition (soil moisture mainly). However, since it is necessary to know the total available water in the unsaturated zone, a direct approach was considered with even lower frequencies (wavelengths of several meters) to reach deeper layers. This poses major problems in terms of spatial resolution (a few hundred km) and ionospheric effects. Thus, this option is not retained at present. To address these issues, several studies have recently demonstrated the feasibility of the disaggregation approach of SMOS SM data (Malbéteau et al., 2016; Merlin et al., 2006b; Pellenq et al., 2003).
Table of contents :
Chapter I Bibliographic synthesis
I.2 Soil moisture
I.2.1 In situ measurements
I.2.2 Remotely sensed approaches
I.2.2.1 Soil moisture indices
I.184.108.40.206 Shortwave-based index
I.220.127.116.11 Thermal-based Index
I.18.104.22.168 Thermal/shortwave-based index
I.22.214.171.124 Microwave-based index
I.2.2.2 Soil moisture retrieval
I.2.2.3 Soil moisture missions
I.3 Land surface temperature
I.3.1 Remote sensing approaches
I.3.2 Spatio-temporal representativeness
I.4.1 Direct measurements of ET
I.4.2 Factors conditioning ET
I.4.3 Remote sensing-based modelling approaches
I.4.4 Surface evaporative efficiency
Chapter II Data & study sites description
II.2 Sites and in situ data description
II.2.1 Study areas
II.2.2 Meteorological data
II.2.3 In situ soil moisture data
II.2.4 In situ LST data
II.2.5 Flux data
II.3 Preprocessing satellite data
II.3.1 Satellite data characteristics
II.3.2 Data preprocessing
II.3.2.1 Thermal infrared (TIR) data
II.3.2.2 Radar imagery
II.126.96.36.199 Thermal noise removal
II.188.8.131.52 Radiometric calibration
II.184.108.40.206 Terrain correction
II.220.127.116.11 Filtering speckle effects
II.3.2.3 High resolution reflectances
Chapter III Models & methods
III.2 Soil moisture indices (SMP)
III.3 Endmembers temperatures estimation
III.3.1 Modelling extreme temperatures: physically based energy balance model
III.3.2 Image based extreme temperature: contextual method
III.4 Integrating the SM indices to improve the water need estimates
III.4.1 Enhance Penman-Monteith method to estimate ET: thermal-based SMP
III.4.2 Calibration of the radar data to retrieve SM: radar/thermal based SMP
III.4.2.1 Benchmark approach: based only on radar data
III.4.2.2 New approach: combined radar/thermal data
III.4.3 Improve the spatio-temporal resolution of MODIS LST data: radar-based SMP
III.4.3.1 MLR technique
III.4.3.2 RTM technique
III.18.104.22.168 Model description
III.22.214.171.124 LST endmembers
III.126.96.36.199 Backscatter endmembers
III.5 Models evaluation
Chapter IV Results and discussions
IV.2 Consistency between image- and EBsoil-based extreme soil temperatures
IV.3 Wheat evapotranspiration using thermal/optical-based approach
IV.3.1 Relationship between surface resistance and stress index
IV.3.2 Evapotranspiration estimation at parcel scale
IV.3.3 Evapotranspiration mapping at perimeter scale
IV.3.3.1 Wheat stress index mapping at 100 m resolution
IV.3.3.2 Wheat evapotranspiration mapping at 100 m resolution
IV.3.3.3 Validation over flood and drip irrigation parcels
IV.4 Improving the LST spatio-temporal resolution
IV.4.1 Application to aggregated Landsat-7/8 data: R3 and Sidi Rahal sites
IV.4.2 Application to MODIS data: R3 area
IV.5 Surface soil moisture at parcel scale
IV.5.1 Sensitivity of VV- and VH-polarized data to soil moisture
IV.5.2 Relationship between thermal-derived SMPTs and radar signal
IV.5.3 SM estimation at high spatio-temporal resolution
IV.5.3.1 SM retrieval
IV.5.3.2 Sensitivity to temperature endmembers
IV.5.3.3 SM validation: Improvement of soil evaporation estimation
IV.6 Summary and conclusion
Conclusions and perspectives