CC BY
0Sensors 2010, 10, 6980-7001; doi:10 3390/100706980
O PEN ACCESS
ssnsDrs
ISSN 1424-8220 www m dpicom /jpuincLl/gan^gois
Article
L Band Brightness Tem perature O bservations over a Corn Canopy during the Entire G row th Cycle
Alicia T . Joseph *,Rogier van der Velde 2, PeggyE. O'Neill1, BhaSkar J. Choudhury 1, Roger H . Lang 3, Edward J. Kim 1 and Timothy Gish 4
1 HydtoLbgxal Science Branch/614 3, Hydrosphere and Biosphere Sciences Laboratory, NASA/oddardSpace Flight Center, Greenbelt, M D 20771, USA; E-M ails: PeggyE ONeill@ nasagov PEO .); BhaskarJChoudhury® nasa.gov (B JC .); Edward JKim@ nasagov (E JK .)
International Institute for Geo-Inform aton Science and Earth Observation (ITC), HengeHaostraat 99, PO . Box 6, 7500 AA Enschede, The Nettherands; E-M ail: veLde@ itcnl
3 Deparmentof Electl±alEhgieerhg & Computer Sciences,the George W ashingtonUniversity, W ashington,DC 20052, USA; E-M ail: lang@ gwuedu
4 USDA-ARS Hydrology and Remote Sensing Laboratory, Buildiig 007, BARC-W EST, BellsviDe, MD 20705, USA; E-M ail: TimothyGish@ arsusdagov
* Authorto whom cbrlespondence should be addre^ed; E-M ail: Alicia.T Joseph® nasagov; Tel: +1-301-614-5804; Fax: +1-301-614-5808.
Received: 26 March 2010; in revised form : 11 June 2010 /Accepted: 11 June 2010 / Published: 20 July 2010
Abstract: During a field cam paign covering the 2002 corn grow ing season, a dualpolarized tow er m ounted L-band (14 GHz) radiom eter (LRAD) provided brightness tem perature (Tb ) measurem ents at preset intervals, incidence and azimuth angles. These radiom eter m easurem ents w eue supports by an extensive characterization of land surface variables including soil m oisture, soil tarn perature, vegetation biom ass, and surface roughness. In the period M ay 22 to August 30, ten days of radiom eter and ground m easurem ents are available fora corn canopy with a vegetationwatercontent (W ) range of 00 to 4 3 kg m . Using thus data set, the effects of corn vegetation on surface em issbns are investigated by m eans of a sen i-smpidcal radiative transfer model. Additionally, the impact of roughness on the surface emission is quantified using TB measurements over bare soil conditions. Subsequently, the estim ated roughness param eters, groundm easurem ents andhorizontally
(H)-polarized TB are employed to invert the H -polarized transm issivity (yh) for the m onitored corn grow ing season.
K eyw ords: field cam paign; L-band radiom etry; vegetation effects; surface roughness
1. Introduction
Low frequency passive m ±row ave observations have been intensively studied for their potential of retrieving soil moisture e.g., [1-3]. Studies have demonstrated that when an appropriate characterization of vegetation, soil surface roughness and dielectric properties are applied, soil moisture can be retrieved fairly accurately from the brightness ternperatures TB's) measured by m icrowave radiom eters eg., [4.5] .A s a result, the Soil M oisture and O cean Salinity (SM O S [6]) m issbn is the first of three L-band radiom eters designed for global soil m oisture m onitoring purposes to be launched. I the near future, the A quarius and Soil M oisture Active Passive SM A P [7]) m issions will follow; their expected launch dates are in spring 2010 and in 2015, respectively. W ith this increased availabilityof low frequency spaceborne radiom eter observations, new opportunities arise form onitDring soilm oisture globally.
Am ong the challenges in retrieving soil m oisture from TB m easurem ente is to account for soil surface roughness andvegelationeffecte. M ost retrieval approaches utilize sim ilar radiative transfer equations [8-10], in which the effects of vegetation are param eterized by the vegetation transm issivity fy) and the single scattering albedo (ty). Parde et al. [11] concluded that for retrieving soil moisture globallythe a> canbe assum edcDnslant over tim e. Conversely, the y changes over tim e because its m agnitude is proportional to the biom ass andis also affected by vegeta^tiongeom etry eg., [12,13]. M oreover, the y is known to dependon the sensing configuration (eg., frequency, view angle and polarization) eg., [14-16].
M ethods forestm ating the y use either m ultiple channel m icrow ave data orancilay data.. A direct estm ation of the y from m icrow ave data is preferred because the ancillary data needed at a global scale for soil m oisture retrieval m ay notbe available. How ever, its dependence on the instrum entparam eters complicates the inversion of y from TB's measured at different frequencies, view angles and polarizations. Large scale soil moisture monitoring studies eg., [17-19] have, therefore, frequently adopted the ancilarydata approach to determ ine the y, which has been extensivelydescribediinthe scientific literature eg., [20,21]. For this characterization, the y is related to the vegetation optical depth (r), w hich is estm ated as a fUnclionof the vegetation w ater content (W ) anda crop-specific empirical param eter,b, which depends on the instum entparam eters.
Various implementatons of this approach within soil moisture retrieval algorithm s have been reported. For exam ple, Jackson et al. [8] used a land cover m ap to define for each crop type a specific b value and utilized the Normalized Difference Vegetation Index NDVI) to estimate the W . Similarly, Bindlish et al. [22] adopted the NDVI as a proxy for the W , but inverted the b values from dual-polarized X-band (10.65 GHz) TB by assum ing that the r is the same forboth horizontally (H) and vertically (V) polarized data.. This polarization dependence is taken into account by the SM O S level 2 processor [12] as its effect is expected to be more significant at the lower L-band frequency. In
addition, a m ore sophisticated approach for m odeling the view angle dependence of r is inclludedin SM OS processor because its TB's are collected from different angles [12]. Further, apart from the NDVIalso the LeafArea Index (LAI) has been found to be a good estimator for W [23] and isused for the SM O S soilm oistuue retrieval.
AUthoughiterative procedures for inverting the r have been developedeg., [1122], initial values and uncertainty ranges for the r are stiUl needed as input. The selection of the appropriate param eterizationfbr a specific landcover r^ies, how ever, often on param eter sets derived from TB measurem ents collected during past intensive field campaigns eg., [16.20]. By default, the validity of those param eterizations is restricted to the conditions for which they have been derived. M any of the past field cam paigns covered, for exam ple, a part of the grow th cycle of agricultural crops. Therefore, the tem poral evolution of the y and b param eterthroughout the grow th cycle is not fully understood.
This paper contributes to this understanding by analyzing the L-bandH -polarized TB 's measured throughout the complete 2002 corn (Zea mays L ) growth cycle. The utilized data set has been collected at one of the fields of the Beltville Agricultural Research Center (BARC) by an auttom ated towermounted L-band (1.4 GHz) radiom eter (called LRAD) Parting from May 22nd till the beginning of September. These radiom eter measurements were supported by a detailed land surface characterization, which took place about once every week and included measurements of the vegetation bioma^, soil moisture and soil temperature. Despite mechanical difficulties with the scanning system of LRA D thatproduced gaps in the data record, a total of ten days distributed over the grow ing season of both radiom eter and ground m easurem ents are available covering a W range from 00 to 43 kg m 2.
The objective of this investigation is to evaluate the variations in the y and the em pidcalparam eterb over the monitored corn growth cycle. First the impact of the surface roughness on the surface em ission is quantified using the LRAD TB's over bare soil conditions and an older data set collected at the BARC facility. Subsequently, the y (and b parameter) are inverted from individual TB m easurem ents using the estim ated roughness param eterizaton, andm easured soil m oisture and soil tem perature. In addition, an analysis is presented of the sensitivityof the derived b param eters tor uncertainties in the LRAD TB and the assigned single scattering albedo (y).
2. Theoretical B ackground
The starting point for the com putation of m icrow ave em i^ion from vegetated surfaces is the ^m i-em pirical radiative transfer approach by Mo et al. [24], which is based on the asaim ption that at L-band absorption isdom iiantover scattering,
where,TBp is the polarize brightne^ tem perature, Rsp is the soil arrface reflectivity (= 1 - em issivity),
yp is the transm issivity of vegetation, is the single scattering albedo, Ts and Tv are the soil and canopy tem peratures, respectively, and superasqpt and subscrptp indicates the polarization.
The first term on the right hand side of Equation (1) represents the m icrow ave em i^ion directly by vegetationand the radiationem itted by the vegetation reflected by the soil arrface back tow ards the
sensor. The second term quantifies the em ission contdbutionfrom the soil, corrected for the attenuation by the vegetation layer.
The solution to the radiative transfer equation requires param eterization of the vegetation and soil surface lyer radiative transfer properties. Further, tern peratures of the vegetation and soil surface layer are required. However, when assum ing the vegetation and soil surface are in thermal equilibrium with each other, Ts and Tv can be considered equal; this condition occurs typically near daw n. The required tem perature is then considered representative forthe em itting layer.
21. Em issdon from Soil
The surface em issivity is typically described in term s of the surface reflectivity. This is convenient because the m ±row ave reflectivity under sm ooth surface conditions can be calculated using the Fresnel formulas fbrreflectivity (Rp), which read forthe H and V polarization,
rh (0) =
cos#-(£r - sin2#)2
cos^ + (fr - sin2 (>y
RV(0)=
er oos# - (sr - sin2
£r oos0 + ( - sin2
where, sr is the dielectric constantof soil, d is the incidence angle.
In this study, the approach described by W ang and C houdhury [25] has been adopted to account for the effect of surface roughness on the rflectivity. This approach involves two param eters, where one param eter has an attenuating effect on the surface reflectivity and the other accounts for the depolarizing effect of the surface roughness,
Rsp(0) = [(1-Q)Rp + QRq]exp(-h) with h = h,G (#) (3)
where, ho is roughness param eter given by 4k2c2 w ithk as the wavenumber (2 %IX) and a as the root m ean square (rm s) height of the surface height variations, Q is a polarization m iing factor, G (9) is a function accounting for the angular dependence of surface roughness effect on surface em ission and superscript q represents the polarization orthogonal to polarizationp, which can be eitherH orV .
Originally, W ang and Choudhury [25] took the function G (9) equal to cos^O. How ever, W ang et al. [26] have found that the dependence of cos20 is much too strong and replaced it by G (9) = 10 for best fitting theirdata. The latter is initially adopted here.
2 2. Vegetation Effects on Soil Surface Em issbn
W ithin the radiative transfer approach, vegetation effects are characterized by two param eters: transmissivity (y) and single scattering albedo (y).The a> isa measure for the fraction of attenuated radiation scattered from the canopy,
= p 5 p (4)
where, ^sp and *-ap are the scattering and absorption coefficients, respectively.
These scattering and absorption coefficients can be obtained through application of the discrete medium approach eg, [27,28], inwhich individual com ponents of the vegetationlayer leaves and
stem s) are represented by elliptical and/or cylindrical dielectric scatterers. In som e cases, the w is assum eel to be negligible or a variable dependent on the growth stage, which can be determ ined from controlled experim ente where all other variables (eg, soil m oisture, temperature of em itting layer, surface roughness and transm issiviy) are measured.
The transm issivity describes the am ount of soil em ission passing through the vegetation layer and is an important variable for quantification of the effect of vegetation on microwave emissbn. The one-way transm issivity through the canopy layer is form ulated as,
y = exp
y cos^ J
where, rp is the polarization dependent optical depth or canopy opacity, which can be calculated using,
Tp = kepHv (5)
4n / r \
kep =T ^(V (6)
where, Hv is the canopy height, kp is a polarization dependent extinction coefficient, no is the num ber of phyttoelam ents per unit volum e, X is the w avelength and im ^ fp^ is the im aginary part of the
polarization dependent scattering m atrix of the phyttoelam ents in the forward (direction.
Several studies [15,16,20] have shown thatrp can be related to the vegetation watercontentas,
tp = bp •W 7)
where, W is the vegetaton water content (kg m ) and bp is an empirical param etervarying with crop type, canopy structures, wavelength, and polarization.
Equation 7) requires inform ation about the W and bp param eters for different types of vegetation. This approach has been frequently used for soil m oisture retrieval purposes eg, [18/19] and has been proposed as part of the soil m oisture retrieval algorithm s for current and future m icaow ave radiom eters eg , [29]. The SM O S level 2 soil m oisture retrieval processor adopts a sdm ilar approach relating the rp to the leaf area index (LA I) instead of the W [30].
3. The OPE3 Experim ent
31. Site D escrptin
The present study was conducted atOptm izing Production Inputs forEconom ic and Environm ental Enhancement (OPE3) test site managed by the USDA-ARS (United States Department of Agriculture-AgriculturalResearch Service) [31]. The site consists of four adjcent watershecls with sam ilar surface and sub-surface soil and w ater flow characteristics and covers an area of 25 ha near Bet/ille, Maryland (Figure 1). Each of the fourwatersheds is formed from sandy fluvial deposits and has a varying sOope ranging from 1% to 4% . The soil textural properties are classified as sandy loam with 235% silt, 603% sand, 161% clay, and bulk density of 125 g cm "3. A detailed description of the research activities can be found athttp //hydrolab arEusda.gov/ope3. (Verified Decam ber23, 2009).
Figure 1. Location and sdhem atization of the OPE3 rem ote sensing experim ental setup in 2002.
32. Ground M easurements
The in-siaim easurem ent strategy w as deigned to provide ground inform ataonto supple ent the radar and radiom eter data acquisitions, and took place every W ednesday, rainy days excluded. in this paper, an analysis of the radiom eter observations is presented. A descriptionof the radar data set is given in Joseph etal. [32].
During the field campaign M ay 10 to October 2, 2002) representative soil moisture, soil temperature, vegetation biomass (wet and dry) and surface roughnessmeasurementswere taken around the radiom eter footprints. Soil moisture and soil temperature measurements were collected at twenty-one sites located at the edge of a 671 m x 33 5 m rectangular area depicted in Figure 1. Vegetation biom ass end surface roughness m easurem ents were taken around the study area at representative locations.
Soil m oisture and soil tem perature
Soil m oisture w as m easured using gravim etac, portable im pedance probe—Delta-T theta probe The US G overnm ent does not endorse any specific brand of im pedance probe for m easuring soil moisture or any specific brand of digital thermometers), and buried impedance probe Time Domain Reflectometry TDR)) techniques. Soil samples of the top 6-cm soil layer were collected at the beginning of each day in conjunction w ith the theta probe m easurem ents prim arily for calibration purposes. Theta probe m easurem ents w ere collected typicallyat 8 0)0, 10 0)0, 12:00 and 14:00 hours
(USA Eastern). The buried TDR probes were installed at locations RS, Rll andRlB Figure l) at various depths (S, l0 and 20 cm ) and insertion angles (horizontal, vertical, and 4S degrees).
The soi. dielectric constant (pr) m easuied by the theta piobe w as converted to volum etric so;lL m oisture (M v) values by fitting a linear regression function through the follow ing leLationship Figure 2a),
yfej = a0 + a^M
where, a0 and al are regression param eters.
Figure 2. a) Comparison of the calibrated theta probe against the gravimetric Mv; b) Com parison of the theta probe m easuied sr against the calculations m ade using the D osbon soil m ixing m ode] (c) M v m easuied by the theta probe, TD R and giavim etric sam p]ing technique plotted against tim e.
(0 4-1
и 15 -
Gravimetric Mv [m3 m-3]
5 10 15 20 25
Theta probe diel. const.
Gravimetric Mv О Theta Probe Mv О TDR Mv ^^ Rainfall
5/1/02 6/1/02 7/1/02 8/1/02 9/1/02 10/1/02
Date [mm/dd/yy]
W hile general юй texture-gpecific param eters are available [33], a site specific calibration was performed. To achieve this, юЯ moisture determined giavmettically from the soil samples was converted to M v andused w ithconcuirent piobe observations to fitioreach site a set of a0 and al values. Com parisonof the calibrated theta piobe M v values w iththe giavim etric M v (œe Figure 2a) grives a root m ean squared error (RSM E) of 0 024 m3 m _3, which is com parable to calibiationerrois
obtained with theta probe observations collected in several rem ote sensing cam paigns [34]. In addition, Figure 2b shows the srm easured by the Theta probe plotted against the sr calculated with soilm ixing m odel of D obson et al. [35] using the soil texture and the gravim etric M v. The RM SE of 187 and coefficient of determ ination R ) of 0 77 com puted betw een the m easured and calculated sr indicates that both m ethods for quantifying er are in agreem ent w ith each other. Further, the M v determ ined using the gravim etric:, Theta probe and TD R probe techniques are displayed as time series in Figure 2c forcom parson purposes. As show n by the plot, sim iar tem poral soil m oisture variations are observed by the three m easurem entapproaches, which justify the use of each of aeirprbducts.
Figure 3. M ean and standard deviation of twenty-one soilmoisture a) and, 3-cm and 7-cm
soil tem perature; (b) m easurem ents collected around the radiom eter footprints.
5/1/02 6/1/02 7/1/02 8/1/02 9/1/02 10/1/02
Date [mm/dd/yy]
Soil temperature measurem ents were taken manually at soil depths of 3-and 7-cm at each of the twenty-one sampling locations annotated as R1 to R21 in Figure 1) throughout the experiment using Extech Instrum ents digital stem therm om eters. On intensive sam pling days the œil tem peratures were measure at 8:00, 10 0)0, 12 0)0, 14:00 hours, and the measurements on other days were taken approxim ately every two days at8 00 and 14 00 hours.
Although the study area was selected to m inim ize the effects of land surface heterogeneity, sm all surface height and soil texture variations could potentially influence the representativeness of the m easun soil m oisture and tem perature for the radiom eter footprints. These effects are studied by presenting the tem poral evolution of the m ean and standard deviation (stdev) of the twenty soil m oisture and soil tem perature m easurem ents in Figure 3. Figure 3a show s that the m ean soil m oisture changes in response to antecedent rain events. Also, the soil moisture stdev varies over time
ftom 0.003 m3 m"3 under extern e dry conditions to 0 036 m3 m"3 in the m id soil moisture range. On average, however, the stdev rem ains quite stable around values of about 0 020 to 0 030 m 3 m"3, which is com patible w ith the Theta probe calibration uncertainty of 0 024 m3 m "3. Further, the spatial tem p^ture variability at soil depths of 3 cm Figure 3b) and 7 cm Figure 3c) is quite low w ith averaged stdev values of 0.73 and 0 58 oC , respectively. Given the fairly stable soil moisture stdev and low tem perature stdev observed, the gpatal heterogeneity around the footprint is expected to have only a m inoreffecton the representative m ean of the twenty-one m easurem ents for the radiom eter footprint. The m ean soil m oisture and soil tem perature values are, therefore, used for further analysis.
Vegetation
Corn was planted on April 17, reached peak bioma^ around July 24 and was harvested on October 2. Vegetation biom ass and morphology were quantified through destructive m easurem ents applied to 1 m area approximateily 12 plants) once every week at800 am . The water content, fresh and dry biom asses w ere determ ined separately for the individual plant constituents, arch as leaves stem s and cobs (when present).
Figure 4a show s the developm ent biom assas end w ater content of the total plant over tim e and Figure 4b ULstralEs the tem poral evolution of the w ater content in individual plant com ponents. It follow s from Figure 4b that inthe beginning of the corngrow ing season, the canopy w as prim arily m ade up of leaves and stalks. m the m iddle of the grow ing season the stem contribution becom es m ore dominant and cobs' water content increases to levels exceeding the leaf contribution. Near senescence, water content in the leaves is reduced further, whereas the contribution of the cobs to the total biom ass rem ained constant.
Figure 4. a) Total plant w ater content, fTeshanddrybiom ass plotted against time, (b) W ater content in the leaves, stem s and cobs plotted against time. The m ankers indicate the dates atwhich measurem entswere m ade.
Fresh biomass
6.0 -,
5/1/02 6/1/02 7/1/02 8/1/02 9/1/02 10/1/02
Date [mm/dd/yy]
Total plant Leaves Stems Cobs
5/1/02 6/1/02 7/1/02 8/1/02 9/1/02 10/1/02
Date [mm/dd/yy]
Surface roughness
During the experiment surface roughness was characteried on M ay 25 using the grid board technique. A 2-meter long grid board was placed in the soil and photographs were taken with the aoil aufface in front In total, ten aufface height profiles were recorded. The surface height profile in these pictures w as digitied at a 05-cm interval, from which two roughness param eters w ere derived: the
imsof the surface height andthe correlationlengtn(L). The averaged rm s height and L for the ten observed surface roughness profiles were found to be 162 and 12.66 cm , respectively. Figure 5 show s an example of a photograph taken for this roughness characterization and lists the roughness param eters calculated from the digitized surface height profiles.
Figure 5. The left panel shows an example of a picture taken for surface roughness characterization and the right panel lite the derived surface roughness param eters.
3 3. Radiometer
The deployed radiometer was a dual-polarized L-band passive microwave sensor, called LRAD . The instrument was mounted on a portable 18 m tower and was designed to collect data autom atcally (for this experiment every hour) at five incidence angles 25, 35, 45, 55, and 60 degrees) and three azimuth angles over a range of 40 degrees. LRAD had a 3 dB beam width of approximately 12 degrees, which corresponds to footprints varying from 45 to 155 m eters for the 25 to 60 degrees incidence angle range. M echanical difficulties with the scanning system restricted the LRAD data collection, and produced considerable gaps in the season-long record. Nevertheless, ten days of complete record (ground m easurem ents and radiom eterobservations) were available forthe present analysi.
Each LRAD data run consisted of a pre-calibration, a m easuring sequence, and a post-calibration. During each of the two calibration periods one m icrowave observation was acquired from a m icrowave absorber target of know n tern perature hot target) and one m icrow ave observation was acquired of the sky (cold target), which has at L band an TB of ~5 K (3 K cosm ic background radiation and 2 K atmospheric contribution). These two so-called "hot" and "sky" target observations can be used to calibrate, through linear interpolation, the radiom eterobservations of the land surface using,
T — T
tp -^u + T .
xb u - u Up ^
u hot u sky
T — T
hot sky Uhot - Usky
where TB is the brightness tern perature [K], T indicates the tern perature [K] of the specified target and U represents the LRAD voltage observations [Volt with subscripts hot and sky indicating the hot and sky targetproperties.
For proce&ing the LRAD measurements to TB's the pre-calibration was used, while the post-calibration was only employed to detect anom alous values. The estim attd uncertainty of the calibrated H -polarized TB is about ±10 K . W hile m ^jsur^ ems w ere also collected for vertical polarization, there remain some unresolved i^ies with respect to the calibration of these m EasuIEm ents. Thus, V-polaried m Easu:^ ems are notbeing presented at this tim e.
4. Results
41. Surface Roughness Estimation Using H-Polarized TB
W ithin the bare aoil em issionm odelby W ang and ChbLdhLIy [25], afce roughness effects are characterized by: 1) mcdiEcatiorL of the reflectance (h parameter), and 2) rEdiistriibutbh of the H- and V-polarized emitted radiation (Q parameter). Since only reliably calibrated H-polarized TB m eaaurem ents are available for analysis, the Q param eter is om itted ie, Q = 0), which e^entially reduces the em ission m odel to the one proposed by C houdhLIy et al. [36]. This form ulaton has been adopted previously in several other studies ie, [17,22]. Based on this asaim ption, the h param etercan be estim attd from H-polariiz^ TB 's measured over bare soil using,
{RH (*)] exp(-h)
where, tbh is the H-polarized brightne^ temperature, TS is the soil temperature, RH is the H- polarized Fresnel reflectivity.
Table 1. Surface param eiters obtahedthrough inversiorLof H -polarizEdTB observation's acquired ovErbaIE aoil conditions.
35 degrees View angle 45 degrees 60 degrees
h = ho-cos2 d 0 641 0 867 1663
h = ho-cos d 0 525 0 613 0j832
h = h 0.429 0j434 0416
h = ho-sec d 0 352 0 307 0j208
h = h0'£EC2 6 0 288 0 217 0j104
The LRAD observations during the OPE3 campaign started on M ay 22, when com crops had just
■ —2 ■ ■ ■
em erged and the total fresh biom ass was less than 0 04 kg m . For these low biom ass conditions, the
measurd TB's are used to estimate the h parameter whereby the mean of twenty-one soil tempera-tuues m EasuIEd at a 3 cm soil depth is adopted as Ts. Unfortunately, for this partof the experim ent, the microwave observations were only collected from view angles of 35, 45 and 60 degrees. The h param eters inverted for these view angles are given in Table 1 for G (9) functors equal to cos2 9, cos d, 1 d and sec2 d.
The derived h param eters fall within the range that has been reported previously. W ang et al. [26] reported a 000-053 h parameterrange forauffaces with a xmsheightvarying from 0 21 to 2 55 cm for
a sim ilarsetting. Considering an averaged imsheight of 1.62 cm was measured around the radiometer footprint, the h param etervaluesobtained from the LRAD observations appears reasonable.
A point of discussion coulld, how ever, be the angular dependence of the h param eter. This is absent for the 35 to 60 degrees view angle range, which is in agreem ent with previous reports eg, [26,38]. An angular dependence is som etim es expected. because when a radiom eterobserves the land surface at different angles surface roughness may have a different impact on the surface emission, while recognizing that Equation (10) is also an approximation [30]. However, the angular dependence of the h param etercould also be a result from the assum ptbn Q = 0. The Fresnel reflectivities forthe H- and V-polarization are both a function of the incidence angle; excluding one of the two polarization com ponents, as is done by assum ing Q = 0 in Equation (3), induces an angular dependence of the h param eter.
4 2. Surface Roughness Param eter Estimation Based on D ual-Polarized TB
The surface roughness param eter h from the present data set dem onstrates an angular dependence that is equal to adopting G (9) = 1 (see Table 1).A lim itation of the present data set is that only H-polarized TB observations are available to som e degree of confidence. For retrieving the h param eter from these TB vales Q was assum ed zero, which m ight alter the angular dependency as discussed above (mixing of polarization). To elaborate on these findings, dual polarized L-band (~1.4 GHz) radiom eter data sets collected over bare soils w ithin the general area of the present study [23] are utilized to invert h and Q sim ultaneously.
The m ethodology used to retrieve the Q and h param eters has been adopted from W ang and C houdhury [25], which is based upon the follow ing two relationships,
X(ö) =
tnb (0)-tnhb (0)
1r 1--TV
1 0L N
(0) + tnb (0)]
■= 2
RH (0) + RV(0)
(1- 2Q )
Y (*) = 1- 1t (0) + Tnhb (*)] = 1(0) + RV (0)] exp("hG (0))
where, TNpB is the normalizó brightness tem perature forpolarization p, according to tbp/ts , X (9) isthe
surface roughness coefficient for deriving the Q param eter, Y(0) is the surface roughness coefficient forderiving the h param eter.
Equations (11) and (12) can be rew ritten to give the Q and h explicitly resulting in,
x (*) '
2[ P(*)]
2Y (0)
P (*) =
RH (0)- RV (0)' RH (0) + RV (0)
RH (0)+ RV (0)
The data set described in W ang et al [26] includes ground m easurem ente of soil m oisture and temperature observed at various depths: 0-0 5, 2 5-5.0, 5 0-10.0 cm for soil moisture and 125, 2 5,
7 5 and15 0 cm for soil tem pera-tuue. ]hacldiitbh, dual-polarized TB observations w ere collected at view angles of 10, 20, 30, 40, 50, 60 and 70 degrees. These m easutem enls have been collected over soil surfaces w ith different roughness characteristics. For this investigatori, a sm ooth and a rough arnface are ^1x^X1111^ analysis witiam easuied im s height of 0 73 and2 45 cm , respectively. B ecause the present data set rnclides racLbm eter observatbrls for an incidence angle range belw een 35 and 60 degrees, only the TB m EasutEd overtthe 20 tb 60 degrees incidence angle range are utilized.
Figure 7. h-parameter as a functbn of incidence angle calculated from dual-polaIiizEd L-bardTB's measured over (a) sm oothbare soil surface and (b) rough bare soil amface. (c) Q-param eters as a fjnctbn of the incidence angle for sam e sm ooth and rough airfaces.
£ 0.2
20 40 60
Incidence angle [degrees]
— cos2(theta)
— cos(theta)
— G(theta)=1
20 40 60
Incidence angle [degrees]
Incidence angle [degrees]
The extensiveness of the radiometer and ground measurements permits all unknowns in equations 13) and 14) tb be derived, and allow s the com putatbn of surface toughne^ param eters Q and h. In analogy wilh the previous roughness computatbns, the soil moisture content integrated over 0-5 0 cm has been used tb com pute the relative dielectric constant and the soil tem peratuue at 2 5 cm has been used tb derive the norm alized brightness tern petature. The resulting h param eters are plotted as a fjncton of the incidence angle fortthe rough and m ooth bare soil afce in Figures 7a and 7b respectively, whereas the com puted Q values are show n as a fLlnction of the incidence angle for boththe roughand sm oothsurfacE irFiigulIE 7c. The h-param enters show ninF jcjuue 7a and7b have been computed a^um rng Iuee d^iff^=n^tG (9) rElatbhshiips,which are: G (9) = cos2^, cosd and 10.
Figures 7a and 7b show a different- angular behavior of the em dtebn m easured over the trough and the sm ooth surface. For the rough surface, it is observed that the functbn G (9) = cos d results in angular independent h parameter. However, none G (9) functbns are able tb suppress the angular
dependence of the h parameter from the smooth surface, while G (9) = cos2^ provides the best approxim atton. Further, an angulardependency of Q param eter is noted in Figure 7c forboth the rough and sm ooth surfaces.
The discussion above and previous result egg., [2636,38] indicate that consistencies in the angular dependence of roughness effecton m icrow ave em issbn are difficult- to identify.H ence, forSM O S soil m oisture processor h is approxim attd by,
h = ^cos^ Q (15)
where, Nrp quantifies the angular dependence of ho, which is also assumed to be polarization dependent.
The param eters, h0 and NRH, have been fitted to m atch our m ulti-angular data collected over nearly bare soil conditions. The obtained param eter values, and RM SE com puted betw een the m easurrd and retrievedTB's andM v's (RMSETb and RM SDmv) are presented in Table 2. In addition, the optimized h0 as well as the RM SETb and RM SDmv obtained with the more frequently used NRP values are given in Table 2.
Table 2. param eter inverted using multiangular H -polarize TB's m easurrd over bare soil and assum ing d^ifferremt NRH values, in bold are the h0 and NRH param eters sim ultaneously inverts from the m ulti angular data..
NRH h0 TB K] M v [m 3 m "3]
Nrh = 0.05 (opt.)* 0.411 1.205 0.0053
NRH = "2 0104 11.225 0.0616
NRH = "I 0 277 6.155 0.0074
NRH = 0 0.407 1.208 0.0065
NRH = 1 0.613 7.441 0.0082
Nrh = 2 0.784 13 734 0.0074
* Nrh and ho are calibrated simultaneously.
An analysis of the param eter values show n in Table 2 dem onstrates the advantage of incorporating the Nrp parameter. RMSE's between the measured and simulated TB's increase from about 1.2 K to more tthanio0 K when the NRP is change from 0 to -2or +2. Surprisingly, this reductioninthe ability to simulate TB's only reduces the soil moisture retrieval accuracy significantly when NRP is taken equal to -2. This is explained by the fact that the M v is retrieved by using as cost function the RM SE computed between the TB simulated and measurrd from different view angles at a given time step. For less negative and positive NRP values, the underestim attonof m easured TB at low or high) view angles is com pensated by an overestim atton at high or low) view angles. H en<ce, the increase in the retrieval uncertainty is for various NRH values m uch sm a Her than would be expected based on the model's ability to simulate TB's. It should, however, be noted that in this case only bare soil conditions are considered and, thus, results m ay be different undervegetation conditions.
4 3. Estimation of the H-polarized Transm issivity
W hen soil moisture and surface temperature are known, H -polarized transm issivity (yh) can be estm attd throughthe inverr^nof Equation 1) assum ing that tarn poral variations inthe roughness param eters are small and the a> equal to zero. Estimates of the yh are only presented for retrievals from H-polarized TB 's measured in the early morning (around 8:00 AM ) because at that tim e of the day the soil surface and vegetation are typically found to be in therm al equilibrium eg., [40]. This assumption permits using a single so-called 'effective' temperature as input for Equation (1), for which the temperature measured at a 3 cm soil depth is adopted. Further, for calculation of the Fresnel reflectivity, the eris obtained through application ofDobson's soil mixing model [35] with inputof soil textual properties and the m easured soilm oisture.
Table 3. H -polarized transm issvities and b param eters estim attd over the 2002 com grow th cycle using m ulti anguarbrghtness tem peratures.
Date W Transm issdvity B param eter
kg m ~2 35o 45o 60o 35o 45o 60o
M ay29,2002 0 1 0 945 0.951 0 967 0 .431 0.356 0 .167
June 5, 2002 0 3 0 .857 0.878 0.881 0 .423 0.306 0 .211
June 19, 2002 1 9 0 .784 0.830 0.763 0 .105 0.070 0 .071
June 26. 2002 3 1 0 .678 0 684 0.641 0 .101 0.085 0 .070
July 3,2002 3 .7 0 .695 0 679 0.629 0 .081 0.075 0 .063
July 9,2002 4 2 0 .640 0.552 0.556 0 .088 0.101 0 .070
July 12,2002 4 3 0 639 0.532 0.517 0 .085 0.103 0 .076
August21,2002 2 6 0 783 0.758 0.736 0 .078 0.076 0 .060
August30, 2002 2 0 0 821 0786 0.732 0 .081 0 086 0 .079
The retrieved yh foreach day and view angles of 35, 45 and 60 degrees are given in Table 3 and are
plotted in Figure 8a against the total plant W . In Figure 8a, yh com putatons are also presented for an
2 _ 1 ... ■
assumedb parametervalueof0117m kg , which isthe median of L-band corn b values presented in
Jackson and Schm ugge [20]. Further, b param eters have been derived from the retrieved yh's, which
are given in Table 3 and plotted against the total plant W in Figure 8b. It should be noted that m ost b
■ • ■■■ —2
param eters have previously been derived for dense corn canopies with W in the range 12-6.0 kg m .
A com parison of b param eters derived for M ay 29 and June 5 (W = 01 and 0 3 kg m ) against p^r^io^slyreport^ values is, therefore, not optim al. The fieldcondiliiorLs observedfrom June 19to August30 (W = 19 - 43 kg m~2) are, however, compatible in term s of biomass to corn canopies referred to in these previous investigations.
Figure 8. H -polarized corntransm issivities (a) andb param eters (b) inverts from TB's m easured at incidence angles of 35, 45 and 60 degrees plotted against the totalplant W .
£ 0.4
O 35 degrees O 45 degrees # 60 degrees — — Theory 35 degrees
-----Theory 45 degrees
-Theory 60 degrees
i i i 12 3
W [kg m-2]
* i 8 , is
2 3 4 5
W [kg m-2]
Figures 8a show s that the retrieved yh follow s a different pattern than expected based on the literature.At the beginning of the growth cycle, the yh is smaller than expected, while closer to peak biom ass the yh is larger-. In term s of the b param eter, the obtained values are higher than the literature reports just after em ergence of the corn crops and som ewhat lower at higher W levels (>19 kg m ). The dependence of the b param eter on W can be argued based on previous investigations. Le Vine and Karam [41], am ong others, have show n that the attenuation by canopies com posed of elem ents w ith sam ilar dim ensions as the w avelength is also specific to the vegetation m orphology. A s changes in biom ass (or W ) are typically associated with different grow th stages and also architectural changes in the canopy, the b param eter can be expected to vary throughout the grow th cycle.
Table 4. H-polarized transm issivities and b param eters inverted from TB 's measured under
— 2 —2 sparsely M ay 29th, W = 01kgm 2) and densely CJuly9th, W = 42 kgm 2) vegetated
conditions and perturbed by +10), 0.0 and -10) K, respectively.
Transm issiviy 35o 45o 60o b-parram eter 35o 45o 60o
M ay 29th W = 01 kg m ~2 TB - 10 K TB TB + 10 K 0 958 0 958 0 973 0 949 0 951 0 967 0 940 0 943 0 962 0 355 0 300 0139 0.431 0 356 0167 0 510 0413 0195
July 9th W = 4 2 kg m ~2 TB - 10 K TB TB + 10 K 0.690 0 593 0 577 0.640 0 552 0 556 0 584 0 506 0 535 0 073 0 089 0 066 0 088 0101 0 070 0106 0116 0 075
On the other hand, it shouldbe noted that the presented yh's and b param eters are also subject to various sources of uncertainty em bedded within the inversbnprocedure. For exam ple, corncrops at the beginning of the growing season are very sm all which lad to relatively large uncertainties in the m easured W . M oreover, the contribution of the vegetation em issaon to the m easured TB is also sm all at the earlygrowthstage. Uncertainties inthe measured TB may have, therefore, a large impact onthe
derived b param eters. To dem onstrate the im pact of such TB uncertainties on the derivation of b
param eters from m easurem ents acquired over sparse and dense vegetation, the yh's have been inverted
— 2 —2 afterperturbiing the TB measured on M ay 29th (W = 01 kg m 2) and July 9th (W = 4 2 kg m 2)
by ±1.0 K . The obtained yh's and b param eters for these two dates are given in Table 4. These results
confirm that'under sparsely vegetated conditions TB uncertainties have a larger im pact on the derived b
param eters than under densely vegetation conditions. The b values retrieved for M ay 29th range
2 _ 1 ■ ■ from 0 355 to 0 510 m kg for the 35 degrees view angle, while for the sam e angle the b param eter
2 —1
from July 9th range from 0 073 to 0106 m 2 kg 1.
The somewhat higher yh's (and lower b param eters) obtained over more dense vegetation are explained by the effects of scattering within the canopy, which has notbeen considered as a> = 0 0 has been assumed. W hen the attenuation by vegetationis m all, the scattering withinthe canopy can be a^um ed negligible because the emission by vegetation ism all eg, [14]. This justifies using a> = 00. A s the biom a^ increases, vegetation em ission alto increases and scattering w ithin the canopy w ill have a m ore im portant im pact on the m easured TB . The previously reported a> values tabulated in Van de Griend and W igneron [37] may reach forL-band and corn up to values of 013.
Table 5. Single scattering albedo (y) inverted from LRAD TB m easured on June 9, 2002
_2 2 —1 (W = 4 2 kg m ) asEummg a range b param eters from 010 to 015 m kg .
b-param eter Single scattering
albedo
m2 kg-1 35o 45o 60o
010 0 014 0 017 0.033
011 0 016 0.021 0.037
012 0 018 0.024 0.040
013 0 020 0.027 0.043
014 0 021 0.028 0.044
015 0 022 0 030 0.045
To quantify this effect of mattering under densely vegetated conditions, the co is inverted, instead of yh, from the TB 's m easured on July 9th for assum ed b param eters of 010, 011, 012, 013, 014 and015m2 kg"1 The obtained cy's are given in Table 5, which illustrate the numerical correlation between the parameters b and a> within TB simulations using Equation (1); namely for mall b param eters, a> is alto sm all. Further, it is noted that the inverted cy's are dependent on the view angle. This can be argued for since scattering w ithin the canopy is affected by orientation of batterers eg, stems, leaves and cobs) relative to the view angle eg, [27,28].
The previous discussion on the effects of vegetation on L-band H-polarized TB 's dem onstrates that the strength of scattering and absorption within a corn canopy changes over the grow th cycle. This can be attributed to changes in the canopy's architecture as the corn crops develop. As a result, the rh is foundtobe a nonlinear functbnof the W , while m ost of current soil m oisture retrieval algorithm s adopt a linear relationships. To evaluate how this a^um ptininfluences the reliabilityof retrievals, soil m oisture is inverted by m inim izing RM SE betw een sim ullated and m easured TB's for view angles
of 35, 45, and 60 on [30], respectively.
■ 2 _ 1 and assum mg a constant b and w of 012 m kg and 0 0 based
Figure 9. Soil moisture m easur^ ents andretrievals obtainedby assum ing a constant b
2 _ 1 ■ parameterandco of 012 m kg and 0 0), respectively.
RMSE = 0.054 m3 m-3 R2 = 0.824
6/1/02
7/1/02
8/1/02
9/1/02
Measured Retrieved W
The retrieved andm easured soil m oisture is plotted against tin e along w iththe total plant W in
Figure 9. The plot show s an underestm aton of m easured soil m oisture over sparse vegetation
_2 —2 W < 10 kg m ) andan overestm atonfor densely vegetated conditions W >15kgm ).As the
contribution of vegetation on both TB m easurem ents and sm ulations is m ore significant at a high
biom ass, the im perfect vegetation param eterizationleads to a larger overestm aton for dense
vegetation as com pared to the underestm aton found for sparse vegetation; RM SE = 0.021 m3 m "3 for
— 2 3 —3 —2 .
W < 10 kg m 2 and RM SE = 0 065 m3 m 3 for W > 10 kg m . Based on these results rtm ay be concludedthat consideration of the canopy's architecture for determining the vegetation parameters will assist in further improving the reliability of soil moisture retrievals especially over dense vegetation.
5. Concluding Rem arks
In this investigation, the H-polarized TB's measured by a tower mounted L-band (1.4 GHz) radiom eter were used to analyze the vegetation effects on surface em isson throughout the 2002 corn grow th cycle:. C oncurrent with the radiom eterm easurem ents an extensive land surface characterization took place about once a w eek including soil m oisture, soil tem perature and vegetation biom ass measurem ents. Over the period from M ay 22 to August 30, ten days with a complete record of ground
and radiom eter m easurem ents are available for analysis covering a vegetation watercbntent (W ) range
of 0 0 to 4 3 kg m .
The roughness param eter, h, needed to correct for the effects of surface roughness, w as inverted from H -polarized TB m easured early in the cbrngrbw ing season over essentiallya bare soil surface. Since V -polarized TB m easurem ents w ere not available for this investigation, the surface em ission
model by Choudhurry etal. [34] (i^um ing Q = 0 0) was adopted and differrnt G (9) functions were used to analyze the angular dependence of h. W hile recognizing that both V - and H -polarized reflectivities dqpendon the view angle, the a^um ptionQ = 00 couldaffected the obtainedangular dependency of h. Therefore, a dual-polarized L-band radiom eter data ^t from 1981 [26] was used to investigate the im pact of a^um jng Q equal to 0 0. It w as found that even w ithin this com plete radiom eter data ^ts consistencies in the angular dependence of the h are difficult to identify, which is in line with the parameterization G (9) = cos1^ (9) adopted for the SM OS level 2 soil moisture proce^or. Using this formulation, a good agreement was obtained between the measured and com puted TB .
Based on the derived surface roughness formulation, H -polarized corn transm issivities (yh) have been retrieved using the radiative transfer equation and assum ing the single scattering albedo (cyh) equal to zero. The derived yh's were converted into b param eter values using the m easured total plant W . For gpar^ vegetation, the obtained rh's and b param eters were found to be larger thanthe values reported inthe literature. This is partlyexpllainedby the fact that under low biom ass conditions TB uncertainties result in a particularly large uncertainty in the derived b param eter. For dense vegetation, the inverted b param eters were somewhat smaller thanexpected, which was attributed to scattering
within the canopy that was not accounted for, since a> was initially a^um ed to be zero. By a^um ing
■ 2 _ 1 ■ that the corn b param^t^varri^ from 010 to 015 m kg , the cohwas derived from TB measureents.
For this range of b param eters, the obtained range in wh's is in agreem ent with "literature reports, but
displays a strong angular dependence.
This study show s that the strength of scattering and absorption w ithin a corn canopy changes
throughout the growth cycle, which can be largely attributed to changes in architecture of vegetation
layer. For further im provem ent of the soil m oisture retrieval reliability over dense vegetated conditions
the canopy's architecture should be taken into consideration for determining vegetationparam eters.
Analysis of additional radiom eter data sets and sim ulations by advanced vegetation scattering m odels
is recom m ended to obtain a m ore thorough understanding of the behavior of the b param eters
throughout the grow th cycle.
A cknow ledgem ents
The authors would like to acknowHedge that the field campaign was financially supported through NASA and we would like to thank various students forparticipating in the field campaign.
References and Notes
1. Jackson, T J. M easuring large scale surface soil moisture using passive microwave remote sensing. Hydrol. Process. 1993, 7, 139-152.
2. W igneron, JP.; Kerr, Y .; W aHdteufel, P.L-band microwave emission of the biosphere (L-M EB) m odel: Description and calibration against experim ental data sets over crop fields. Rem ote. Sens. Environ. 2007, 107, 639-655.
3. Owe, M .; De Jeu, RAM .; Holmes, TRH. M uM-sensor historical climatology of satellite-derived global land surface m oisture. J". Geophys. Res. 2008, 113, doi1029/2007JF000769.
4. Saleh, K .; Kerr, Y H .; Richaum e, P.; Escorihuela, M J.; Panciera, R .; D elwart, S .; Boulet G .; M aisongrande, P.; W alker, JP.; W ursteisen, P.; W igneron, JP. SoiL m oisture retrievals at L-band using a two-step inversion approach.. Remote. Sens. Environ. 2009, 113, 1304-1312.
5. Panciera, R .; W aLker, JP.; Kalma, JD .;K im, E J; Saleh, K .; W igneron, JP. EvaLuationof the SM OS L-M EB passive microwavesoilmoisture rtrieval algorithm . Remote. Sens. Environ. 2009, 113,435-444.
6. Kerr, Y H .;W aHteu-fel P.; W igneron, JP.; M artinuzzi, JM .; Font J; Berger, M . Soil moisture retrieval from spacer The soil moisture and Ocean Salinity SM OS) m ission. IEEE Trans. Geosci. Remot. Sen. 2001, 39, 1729-1735.
7. Entekhabi, D .; Njoku, E G .; Houser, P. The Hydrosphere Staate Hydros) mission: An Earth system pathfinder for global m apping of soil m oisture and land freeze/thaw . IEEE Trans. Geosci. Remot. Sen. 2004, 42, 2184-2195.
8. Jackson, T J; Le V ine, DM .; H su, A .Y .; O ldak, A .; Starks, P J.- Swift, CT.; Isham , JD .; Haken, M . Soil moisture m apping at regional scales using m icrowave radiom etry: The Southern GreatPlains hydrology experiment. IEEE Trans. Geosci. Remot. Sen. 1999, GE-37, 2136-2151.
9. Owe, M .; De Jeu, RAM .; W alker, JP. A m ethodology for surface soil m oisture andvegetation opticaldepthretrievalusing the m icrowave polarizationdiifference index. IEEE Trans. Geosci. Remot. Sen. 2001, 39, 1643-1654.
10. W en, J.; Su, Z .; Ma, Y . Determ ination of land surface tern perature and soil m oisture from tropical rai^fallmeasurhg m issbn/n icrowave imagerremote sensing data. J. Geophys. Res. 2003, 108, 4038-4050.
11. Parde, M .; W igneron, JP.; W aldteufel P.; Kerr, Y H .; Chanzy, A .; SObjaerg, S S.; Skou, N . N-param eter retrieval from L-band m icrow ave observations acquired over a variety of crop fields. IEEE Trans. Geosci. Remot. Sen. 2004, 42, 1168-1178.
12. W igneron, JP.; Kerr, Y .; W aldteuiel P.; Saleh, K .; Escorihuela, M J.; Richaum e, P.; Ferrazzoli, P.; De Rosnay, P.; Gurney, R .; Calvet JC .; Grant JP.; Guglielmetti, M .; Hombuckle, B .; M ätzler, C .; Pellarin, T .; Schwanke, M . L-band m icrowave em issdonof the biosphere (L-M EB) model: Descriptbh and calLibraticrL against experim ental data setsover crop fields. Remote. Sens. Environ. 2007, 107, 639-655.
13. W igneron, JP.; Chanzy, A .; Calvet, JC .; Bruguier, N . A simple algorithm to retrieve soil moisture and vegetation biom ass using passive microwave m easurem ents over crop fields. Remote". Sens. Environ. 1995, 51, 331-341.
14. Jackson, T.J.; O'Neill, P.E. Attenuation of soil m icrowave em isson by com and soybeans at1.4 and 5 GHz. IEEE Trans. Geosci. Remot. Sen. 1990, 28, 978-980.
15. W igneron, JP.; Calvet, J.C .; De Rosnay, P.; Kerr, Y .; W aUteudel P.; Saleh., K .; Escorihuela, M J.; Kruszew ski, A . Soil m oisture retrievals from biangular L-band passive m icrow ave observations. IEEE Geosci. Remote. Sens. Lett. 2004, 1, 277-281.
16. Van de Griend, A A .; W igneron, JP. The b-factor as a function of frequency and canopy type at H-polarization. IEEE Geosci. Remote. Sens. Lett. 2004, 42, 1-10.
17. Drusch, M .; W ood, EF.;G ao, H .; Thiele, A . Soilm oisture retrieval during the SouthernG reat Plains Hydrology Experiment 1999: A comparison between experiment remote sensing data and operationalproducts. W aterResour". Res. 2004, 40, W 2504.
18. Cashon, J; Lakshm i, V ; Bosch, D ; Jackson, T J. M icrbwave rem ote sensing of soilm dure: Evaluation of the TRM M m lcrbw ave im ager (TM I) satellite for the Little RiverW atershed Tifton, Georgia.. J". H ydrol. 2005, 307, 242-253.
19. Bindlih, R ; Jackson, T J; Gasiewski, A ; Stankov, B ; Klein, M ; Co^, M H ; M ladenova, I; W atts, C ; Vi/oni E ; Lakshm i, V ; Bolten, J; KeeHer, T. Aircraft based soil moisture retrievals underm ixed vegetation and topographic aonditionLs. Remote. Sens. Environ. 2008, 112, 375-390.
20. Jackson, T J; Schmugge, T J. Vegetation effects on the m icsowave em ission of axils. Remote. Sens. Enviion. 1991, 36, 203-212.
21. Schm ugge, T J; Jackson, T J. A dielectric m odel of the vegetation effects on the m icrow ave em ission from soils. IEEE Trans. Geosci. Remot. Sen. 1992, 30, 757-760.
22. Bindlin, R ; Jackson, T J; W ood, E ; Gao, H ; Starks, P ; Bosch, D ; Lakshm i, V . Soil m oisture estim atts from TRM M M icrow ave im agerobservationLs overthe Southern United States. Remote. Sens. Enviion. 2003, 85,507-515.
23 . Saleh, K ; W igneron, JP ; De Rosnay, P ; Calvet; J.C ; Esao:lihuela., M J; Kerr, Y ; W aldteufel P. im pact of rain interception by vegetation and m ulch on the L -band em i^oon of natural gra^. Remote. Sens. Environ. 2006, 101, 127-139.
24. Mo, T ; Choudhury, B J; Schm ugge, T J; W ang, JR. ; Jackson, T J. A m odel for m icrow ave em ission from vegetation-covered fields. J. Geophys. Res. 1982, 87, 11229-11237.
25. W ang, JR ; Choudhury, B J. Remote ^nsing of soil moisture content over bare field at 1.4 GHz frequency. J". Geophys. Res. 1981, 86, 5277-5282.
26. Wang, J.R.; O'Neill, P.E.; Jackson, T.J.; Engman, E.T. Multifrequency measurements of the effects of soil moisture, axil texture, and surface roughness. IEEE Trans. Geosci. Remot. Sen. 1983, GE-21, 44-51.
27. Lang, R "H ; Sidhu, JS. Electrom agnetic backscatteang from a layer of vegetation: A discrete approach. IEEE Trans. Geosci. Remot. Sen. 1983, 21, 62-71.
28. Chauhan, N S. Soil moisture estimation under a vegetation cover: Combined active passive m icrow ave remote sensing approach. Int. J. Remote. Sens. 1997, 18, 1079-1097.
29. Njoku, E G. AM SR land surface param eters: Surface soil moisture, land surface temperature, vegetation water content; Algorithm Theoretical Basis Documen, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA, 1999.
30. Kerr, Y H ; W aldteufel P ; Richaum e, P ; Davenport, I; Ferrazzoli P ; W igneron, JP. SM OS level 2 prooe^or axil moisture Algorithm Theoretical Basis Document ATBD). SM -ESL (CBSA),CESBD, ToulouseSO-TN-ESL-SM -GS-0001, V5a, 15/03/2006, 2006.
31. Gish, T J"; W alhall C L ; D aughtry, C S T ; D ulaney, W P ; M ocarty, G W . W atershed--Scale Sensing of Subsurface Flow Pathways atOpe3 Site. In Proceedings of the First Interagency Conference on Research in the W aterdneds, Benson, AZ, USA. October 27-30, 2003; pp" 192-197"
32. Joseph, A.T.; Van der Velde, R.; O'Neill, P.E.; Lang, R.H.; Gish, T. Soil moisture retrieval during a corn growth cycle using L-band 1.6 GHz) radar observationLs. IEEE Trans. Geosci. Remot. Sen. 2008,46,2365-2374.
33. M iller, JD ; Gaskin, G D . Theta probe M L2x—Prinalple of operation and applications, 2nd ed; M acaulay Land Use Research Institute (M LURI). Crairipbi^аkler;Aberdeen, UK , 1996.
34. Cosh .M H .; Jackson, T J. B indii, R .; Fam iic^li^tlti, JS .; Ryu, D . C aDbratenof an im p^danœ piobe forestm atbn of surface so:Dwatsrcьntsntover large regions. J. H ydrol. 2005, 311, 49-5B.
35. D obsen, M C .; UDaby, F T .; HaDDkaiien, M T .; El-Rayes, MA. M iciewave dieDecto behavior of wet ^il^Part II: DieDectcic mixing medeDs. IEEE Tians. Geesci. Remot. Sen.. 19B5, GE-23, 35-46.
36. Choudhuiy, B J.; Schmugge, T J.; Newton., R W .; Chang, A T C . Effect of surface ioughne^ on m irewave em i^sion of кэ! J. Geephys. Res. 1979, B4, 5699-5706.
37. Van de Griend, A A .,-W igneren, JP. On the measuismentof m irewave vegetatenpiopeities: Seme guidelines fera protocol IEEE Trans. Geesci. Remot. Sen. 2004, 42, 2277-22B9.
3B. W igneron, j.; Lagueire, L.; Kerr, Y . A simple param €1э;тз2^1|;Ьп of the L-band microwave em issbn from rough agriculturals:âte. IEEE Trans. Geesci. Remot. Sen. 2001, 39, 1697-1707.
39. EsoorhueDa, M j.; Kerr, Y H .,• De Rosnay, P.; W igneren, JP.; Calvet, JC .; Lemaire, F. A smpDD medeD of the bare юй m iciewave em iœien atL-band. IEEE Trans. Geesci. Remot. Sen.. 2007, 45, 197B-19B7.
40. Choudhuiy, B J. Reflectivities of selected Dand-airface types at 19 and 37 GHz from SSM /I observations. Rem ote. Sens. Environ. 1993, 46, 1-17.
41. Le Vine, DM .; Kaiam , MA. D ependence of attenuation in a vegetation canopy on frequency and pln^ater-contEnt. IEEE Trans. Geesci. Remot. Sen. 1996. 34, 1090-1096.
© 2010 by the authors; Licensee M DPI, Basel, Switzerland. This articDe is an Open Aoœ^ article
distrbuted under the terms and conditions of the Creative Commons Attrbutbn License
(http //crœtveocm m ensc^B^/Do^T^^tiy/î 0/.
Copyright of Sensors (14248220) is the property of MDPI Publishing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
