CN101936921A - Method for inversing soil moisture content from AMSR-E (Advanced Microwave Scanning Radiometer-EOS) data - Google Patents

Abstract

The invention relates to a method for inversing soil moisture content from AMSR-E (Advanced Microwave Scanning Radiometer-EOS) data, which can be used for the remote sensing application departments of monitoring weather and environment, managing land, monitoring a drought and a crop growing situation, monitoring disasters, and the like. The method comprises the following four steps of: firstly, simulating the most part of situations on the earth surface and establishing a database by using a theoretical model; secondly, establishing an inverse algorithm by using a simulation database; thirdly, carrying out the inverse calculation of passive microwave data AMSR-E; and fourthly, carrying out proper evaluation, analysis and modification by using observation data of a meteorological department, thereby increasing the precision and the applicability of the algorithm. The method for inversing soil moisture content from the AMSR-E data can be used for the departments of forecasting weather, monitoring environment, agricultural situations and disaster situations, and the like. The microwave radiation frequency arrangement carried by a Fengyun 3 satellite of China is basically similar to the AMSR-E, and the invention can be applied to the Fengyun 3 satellite of China and a microwave radiometer transmitted by China in the future after being modified.

Description

From AMSR-E data inversion soil moisture method

Technical field

The present invention relates to a kind of method of utilizing the ground level heat radiation information inverting soil moisture that passive microwave sensors A MSR-E on the earth observation satellite obtains, this method has overcome in the past the asynchronous and complicated shortcoming that is difficult to practicality of parameter acquiring in the inversion method.Can be applied in remote sensing departments such as meteorology, agricultural, environmental monitoring and damage caused by a drought monitoring.

Background technology

Soil moisture is an important parameter in the researchs such as the hydrology, meteorology, agricultural and environmental hazard.Along with the development of microwave remote sensor technology, perfect to the deep understanding of face of land microwave radiation mechanism and inverse model and algorithm, passive microwave remote sensing monitoring soil moisture will have more and more broader application prospect.The soil moisture of large scale changes for the water cycle model of setting up the whole world very important, and then can Prediction of Climate Change and flood monitoring.Traditional ground survey station network can not satisfy the time of large scale soil moisture, the needs of spatial variations research.And microwave has special advantages aspect Soil Moisture Inversion.We can say,, will improve and improve the forecast precision of the hydrology and meteorologic model greatly, and provide accurate data for agricultural production and disaster monitoring by passive microwave remote sensing technical monitoring surface temperature and soil water space-time Changing Pattern.At present, the passive microwave remote sensing Soil Moisture Inversion remains a current research focus and difficult point.Though L (1.4GHz) wave band has been turned out to be highly suited in inverting soil moisture, vegetation remains a difficult problem maximum in the Soil Moisture Inversion.There are three main problems at present: the first, the data resolution of low frequency is very low; Second: passive microwave data in the past do not have the data of L-band, and we need understand the situation of change of soil moisture in the past; The 3rd: obtain more face of land information though L-band can penetrate vegetation, it still is subjected to the influence of vegetation.The ground of many small scales and aviation experiment has been carried out and many models and algorithm put forward [

C., Seasonal evolution for microwave radiation from an oat field, Remote Sensing ofEnvironment, 1990,31,161-173; Griend A.A.van de, Owe M., Ruiter J.d., Gouweleeuw B.T., Measurement and behavior of dual-polarization vegetation opticaldepth and single scattering albedo at 1.4 and 5 GHz microwave frequencies, IEEETrans.Geosci.Remote Sensing, 1996,34,957-965; Wigneron J.P., Parde M., Waldteufel P., Chanzy A., Kerr Y., Schmidl S., Skou N., Characterizing thedependence of vegeation model parameters on crop structure, incidence angle, andpolarization at L-band, IEEE Trans.Geosci.Remote Sensing, 2004,42,416-425; Mike Schwank, Christian

Massimo Guglielmetti, Hannes Fl ü hler, L-bandradiometer measurements of soil water under growing clover grass, IEEETrans.Geosci.Remote Sensing, 2005,43,2225-2237.], be proved to be in large scale (on the star) practicality very but go back neither one.

AMSR is modified multi-frequency, dual-polarized passive microwave radiometer.Calendar year 2001 AMSR carries and goes up to the air on the earth observation satellite ADEOS-II of Japan.The AMSR-E microwave radiometer is to improve design on the basis of AMSR sensor, and it carries and launches in 2002 at U.S. NASA earth observation satellite Aqua.The instrument parameter basically identical of these two sensors of AMSR and AMSR-E.Maximum difference is that AMSR passes the equator in the morning about 10:30, and AMSR-E then is in the afternoon about 1:30.The parameter of these two sensors is basic identical, and 6 frequencies of AMSR-E radiometer in the 6.9-89GHz scope are with the microwave radiometer of 12 passages of dual polarization mode.Key instrument parameter [Mao Kebiao, Qin Zhihao, Li Manchun, Xu Bin, data introduction of AMSR passive microwave and main applied research domain analysis, sensor information .2005,3,63-66.] as shown in table 1.

AMSR-E is by measuring the water cycle variation of studying global range from the microwave radiation of earth surface.In hydrology applied research, in order to obtain two with the variation of the soil water content before and after the rain incident, the data that obtain study area continually are very important.The temporal resolution of satellite depends primarily on cuts width, satellite altitude and inclination angle.For AMSR, except the arctic regions, in less than two days time, at rail lift with fall rail and the whole world can be covered once.Fig. 1 is brightness temperature composite diagram [AMSR-ETeam Leader Science Computing Facility Science Software, the 2004.8.AMSR-EBrowse﹠amp that rail falls in AMSR-E; Quick look image (NASA): 1-69].As can be seen from Figure 1, at high latitude and low latitudes, data cover relatively district.At mid latitudes, owing to be subjected to the influence of the figure of the earth, low latitude may will be grown relatively with the cycle that high latitude area covers relatively.Specifically, when falling rail, AMSR-E covered once in two days, and some places are one day or three days.But be to cover once in one day in the area of latitude more than 55 °.

The key instrument parameter attribute of table 1 AMSR-E

The AMSR-E data are mainly used in the research of aspects such as soil moisture, surface temperature, vegetation at present.In the four-dimensional data assimilation of numerical weather prediction model system the inside, the soil water content parameter of large scale is very important.Previous mainly is as the soil moisture index by API (Antecedent Precipitation Index) index to obtaining of this parameter.Owe and Van de Gried[Owe, M., A.A.van de Griend, Dailysurface soil moisture model for large area semi-arid land application with limitedclimate data.Journal of Hydrology, 1990,121,119-132.] by the Soil Moisture Inversion model of large scale of having set up a top modelling, the microwave data of use is SMMR-6.6GHz.The state consistency of whether estimation of this model soil moisture depends on sparse meteorological site data when passing by with satellite upper soll layer.But, studies show that Space Microwave DATA REASONING and soil moisture have presented good correlativity.But its precision and checking remain further to be improved.

Mainly utilize the difference index of the brightness temperature of the level of frequency 37GHz and vertical polarization to come vegetation to study [Mao Kebiao for AMSR-E in the research aspect the vegetation, Tang Huajun, Zhou Qingbo, Chen Zhongxin, Chen Youqi, Zhao Dengzhong, AMSR-E microwave polarization index and the research of MODIS vegetation index relation, land resources remote sensing, 2007,1:27-31.].Though this index on calculating and the spatial resolution on this frequency have advantage, this index can not with the physical quantity direct correlation of vegetation.In addition, because the transmitance of vegetation is an important parameter in the radiation transfer equation, and 6-10GHz and vegetation water cut are approximate linear.Therefore it is very promising studying biomass with low frequency.Many researchists [Njoku Eni G.AMSR Land SurfaceParameters:ALGORITHM THEORETICAL BASIS DOCUMENT Version 2.0, November 10,1997; Jackson T.J., Schmugge T.J., Vegetation effects on themicrowave emission from soils, Remote Sensing of Environment, 1991,36,203-219; Jackson T.J.and O ' Neill P.E., Attenuation ot soil microwave emissision by corn andsoybeans at 1.4 and 5GHz, IEEE Trans.Geosci.Remote Sensing, 1990,28,978-980; Kerr Y.H.and Njoku E.G., On the use of passive microwaves at 37GHz in remotesensing of vegetation, International Journal of Remote Sensing, 1993,14,1931-1943.] eliminate the influence of vegetation with ω-τ model.This method has significant limitation, because present most land passive microwave pixel (tens kilometers * tens kilometers) all are mixed pixels.Give one example, suppose that it all is 0.3 that there is the NDVI of 1000 pixels passive microwave image the inside, but the vegetation pattern of each pixel is all different, and distribution (from stochastic distribution to concentrating) is also different, and we handle the influence of vegetation with a kind of mode, obviously are inappropriate.We think in the process of Soil Moisture Inversion, need utilize low frequency as much as possible and high-frequency data simultaneously.Because low frequency can obtain the soil moisture information below the vegetation, high frequency then can obtain the information of vegetation, in addition, the combination of a plurality of frequency informations can reflect the vegetation distribution and the vegetation structure feature of the pixel the inside of large scale, makes each pixel to be determined by unique.Owing at present the low-frequency band of data lacking very on the star,, be subjected to the influence of mobile phone signal bigger, so will obtain the information difficulty very below the vegetation as AMSRE 6.9GHz signal and unstable.Fortunately, soil moisture can reflect [Jackson T.J. by vegetation, Schmugge T., and Wang J., Passive microwave remote sensing of soil moisture under vegetation canopies, waterresources Resource, 1982,18,1137-1142; Cmaillo P.T., Schmugge T.S., Estimatingsoil moisture storage in the root zone from surface measurements, Soil Science, 1983,135,245.].

Many studies have shown that [Paloscia Simonetta, Paolo Pampaloni, Short communicationsmicrowave remote sensing of plant water stress, Remote Sensing of Environment, 1984,16,249-255; Pampaloni P.and S.Paloscia, Experimental relationship betweenmicrowave emission and vegetation feafures, International Journal of RemoteSensing, 1985,6,315-323; Choudhury B.J., C.J.Tucker, Monitoring global vegeationusing Nimbus-737GHz data some empirical relation, International Journal RemoteSensing, 1987,9,1085-1090.] different frequency or the bright temperature difference (Δ T) of the different polarization of same frequency and the variation of soil moisture be positively related.Say that to a certain extent utilizing this information to come inverting large scale soil moisture at present is a kind of reasonable selection.Of the present invention influence how to avoid the key parameter surface temperature and eliminate roughness as much as possible to be proposed exactly, and utilize emissivity and soil moisture concern the inverting soil water content.

Summary of the invention

The object of the present invention is to provide a kind of method from remotely-sensed data AMSR-E inverting soil moisture, to overcome existing monitoring soil moisture inversion method complexity and to be difficult to satisfy damage caused by a drought monitoring needs, enrich domestic researchist at passive microwave remote sensing device development product algorithm, go up the passive microwave sensor for plan 100 satellites of emission (particularly No. three satellites of wind and cloud) before China 2020 ground Soil Moisture Inversion method reference is provided, and can also further improve Soil Moisture Inversion precision in the present Ministry of Agriculture service operation, improve damage caused by a drought monitoring and crops the yield by estimation precision [the present invention has drafted as one of Soil Moisture Inversion alternative approach in Ministry of Agriculture's farming feelings monitoring].

For achieving the above object, theoretical foundation and the concrete steps from remotely-sensed data AMSR-E inverting soil moisture method provided by the invention are:

Theoretical foundation:

Soil moisture can come direct inverting to be because the variation of soil moisture directly influences the variation of soil dielectric constant with emissivity, and specific inductive capacity is the main factor that the decision emissivity changes.For microwave remote sensor, have only and set up radiation transfer equation by the energy that obtains and come inverting face of land information.Therefore, the Soil Moisture Inversion of passive microwave is to be based upon on the radiation diffusivity equation basis, and promptly the face of land energy that obtains of sensor and the surface radiation relation that can set up energy-balance equation is come inverting soil moisture via satellite.Radiation transfer equation has been described the radiation total intensity that microwave radiometer observed of satellite, and the radiation from the face of land is not only arranged, and also has the path radiation up and down from atmosphere.These radiation components arrive in the process of remote sensor passing atmospheric envelope, also are subjected to the influence of atmospheric absorption and cut down.Therefore, the energy equilibrium of microwave radiation is actually the problem of finding the solution of a complexity.Approximate according to the Ralleigh-Jeans in the microwave region interval, the heat radiation transmission equation can be reduced to shown in the formula 1:

$T_{\mathrm{BT}}={\mathrm{\τ}}_p{\mathrm{\ϵ}}_p{T}_s+(1-{\mathrm{\τ}}_p){\mathrm{\τ}}_p(1-{\mathrm{\ϵ}}_p)T_a^{\↓}+(1-{\mathrm{\τ}}_p)T_a^{\↑}$ (formula 1)

τ in the formula _pThe expression transmitance, ε _pThe expression emissivity, T _BTBrightness temperature on the expression star, T _sThe expression surface temperature, T _a ^↓Be the downward mean effort temperature of atmosphere, T _a ^↑It is on average operative temperature upwards of atmosphere.P represent to polarize vertical polarization (V) or horizontal polarization (H).In formula 1, T _a ^↓With T _a ^↑Approximately equal [Kebiao Mao, Qin Z., Shi J., Gong P., A Practical Split-Window Algorithm for Retrieving Land SurfaceTemperature from MODIS Data, International Journal of Remote Sensing, 2005,26,3181-3204.], so emissivity can be represented an accepted way of doing sth 2.

${\mathrm{\ϵ}}_p=\frac{T_{\mathrm{BT}}-(1-{\mathrm{\τ}}_p)T_a-(1-{\mathrm{\τ}}_p){\mathrm{\τ}}_f{T}_a}{{\mathrm{\τ}}_p{T}_s-(1-{\mathrm{\τ}}_p){\mathrm{\τ}}_p{T}_a}$ (formula 2)

At low-frequency band, very little of the influence that microwave is subjected to atmosphere, when even atmosphere vapour content reaches 5g/cm2, its transmitance still can be approximately equal to 1[Ulaby F T, Moore R K, Fung A K.Microwave RemoteSensing:Active and Passive Dedham.Washington:Artech House, 1986,3,51-80.].Therefore, usually in microwave region, its emissivity can be calculated with formula 3.

${\mathrm{\ϵ}}_p=\frac{T_{\mathrm{BT}}}{T_s}$ (formula 3)

The variation of soil emissivity mainly is subjected to influence [the Wang J R of soil water content and roughness, SchmuggeT J.An empirical model for the complex dielectric permittivity of soil as a function ofwater content.IEEE Trans Geosci.Remote Sensing, 1980, GE-18 (4), 288-295; NjokuEni G, Thomas J.Jackson, Venkataraman Lakshmi, et al.Soil moisture retrieval fromAMSR-E.IEEE Trans Geosci.Remote Sensing, 2003,41 (2), 215-229], therefore can come direct inverting soil water content by the variation of emissivity.The soil moisture is a very crucial parameter in the formula (3), but because the unsettled characteristics of emissivity make the inverting complexity very of surface temperature.Microwave Surface Temperature Retrieval [McFarland M J, Miller R L, Neale C M U, Land surface temperaturederived from the SSM/I passive microwave brightness temperatures.IEEE TransGeosci.Remote Sensing, 1990,28 (5), 839-845; Kebiao Mao, Jiancheng Shi, Zhaoliang Li, Zhihao Qin, Manchun Li, Bin Xu, A physics-based statistical algorithmfor retrieving land surface temperature from AMSR-E passive microwave data, Science in China (Series D), 2007,7,1115-1120.] relatively more difficult, main cause is to be subjected to soil moisture, the uncertain influence of roughness and vegetation.Therefore how to avoid just seem unusual key of this parameter.Here, we use the standardization hpm index to avoid this parameter of surface temperature:

${\mathrm{NDE}}_{i-j}=\frac{{\mathrm{\ϵ}}_i-{\mathrm{\ϵ}}_j}{{\mathrm{\ϵ}}_i+{\mathrm{\ϵ}}_j}=\frac{\frac{T_i}{T_s}-\frac{T_j}{T_s}}{\frac{T_i}{T_s}+\frac{T_j}{T_s}}$ (formula 4)

$=\frac{T_i-T_j}{T_i+T_j}$

Improve the practicality of algorithm at last with partitioned method.Concrete steps divided for three steps:

The first step is to utilize the situation of exhausted big number on the theoretical model simulation face of land and set up database;

The improved IEM model AIEM of newly-developed (Advanced IEM) [Wu T D, Chen K S, Shi J, et al.A transition model for the reflection coefficient in surface scattering.IEEE Trans Geosci.Remote Sensing, 2001,39 (9), 2040-2050; Chen K S, Wu T D, Tsang L, et al.Emission of rough surfaces calculated by the integral equation methodwith comparison to three-dimensional moment method simulation, IEEE TransGeosci.Remote Sensing, 2003,41 (1), 90-101] IEM (Integrated Equation Model) model has been carried out further improvement, the calculating of roughness spectrum and Fresnel reflection coefficient calculating form are improved, and the modeling that makes is more near truth.The IEM model is to propose [FungA K in 1992 by people such as Fung, Li Z, Chen K S, et al.Backscattering from a randomly rough dieletric surface.IEEE Trans Geosci.Remote Sensing, 1992,30 (2), 356-369], this model is based on the face of land scattering model of electromagenetic wave radiation transmission equation, can in a very wide roughness of ground surface scope, reproduce real surface back scattering situation, be widely used in the simulation and the analysis of the scattering of the microwave face of land, radiation, and passed through a large amount of checkings.IEM model (IEM) is to be proved to be one of best model [Chen K S in face of land scattering and the radiomimesis, Wu T D, Tsang L, et al.Emission of rough surfaces calculatedby the integral equation method with comparison to three-dimensional momentmethod simulation, IEEE Trans Geosci.Remote Sensing, 2003,41 (1), 90-101].In recent years, IEM model process is updated and is perfect, and modeling result and precision are improved constantly.IEM model since the scope of its simulation more approach real natural terrain and used widely.The present invention goes up AMSR-E passive microwave sensor 10.7GHz with AIEM simulation earth observation satellite AQUA and 18.7GHz simulates.The physical simulation calculation process as shown in Figure 2.The main input parameter of AIEM is as shown in table 2.Radiation under exhausted big number roughness of ground surface scope and the soil moisture situation is simulated, and set up simulated database.This need of work spends some times, but only need do once.Fig. 3 and Fig. 4 are the relations of soil moisture and emissivity, and as can be seen from the figure, the relation of soil moisture and emissivity is not very good, are column and distribute.

Table 2AIEM model input parameter

Parameter	Minimum value	Maximal value	Step-length	Unit
					Root-mean-square height sig	0.5	3.2	0.3	cm
Persistence length cl	5	30	4	cm
					Soil moisture	0.02	0.45	0.04	％

Surface temperature	0	50	8	℃
					Frequency	10.7	18.7		GHz

The second, utilize simulated database to make up inversion algorithm;

Utilize the database of simulating in the first step, further analyze the relation of microwave polarization index and soil moisture, 10.7GHz and 18.7GHz V polarization polarization index computing formula are suc as formula 5.

${\mathrm{NDE}}_{18.7\mathrm{GHz}-10.7\mathrm{GHz}}=\frac{T_{18.7V}-T_{10.7V}}{T_{18.7V}+T_{10.7V}}$ (formula 5)

Soil moisture and NDE _{18.7GHz-10.7GHz}Relation as shown in Figure 5, comparison diagram 3 and Fig. 4, polarization index is more much better than the relation of emissivity and soil moisture with the relation of soil moisture.The relation of polarization index and soil moisture is suc as formula 6, its related coefficient square reached 0.97.Formula 5 has been avoided surface temperature, can be directly from the star brightness temperature calculate.Therefore formula 6 can finely be used for calculating soil moisture.

$\mathrm{SM}=0.33+10.99947*{\mathrm{NDE}}_{18.7\mathrm{GHz}-10.7\mathrm{GHz}}+563.80628ND{E}_{18.7\mathrm{GHz}-10.7\mathrm{GHz}}^2$ (formula 6)

The 3rd, read in passive microwave data AMSR-E, carry out bright temperature conversion, and calculate soil moisture with formula 5 and formula 6;

The 4th, utilize the meteorological site data that the Soil Moisture Inversion result is done some evaluations, thereby be department's services such as meteorological and agricultural better, also can do some and revise in different local measured datas, thus the precision and the applicability of raising algorithm.

The invention has the beneficial effects as follows, consider the feature of 25 kilometers resolution of passive microwave AMSR-E large scale pixel, utilize 10.7GHz and 18.7GHz V polarization polarization index to avoid the surface temperature parameter, and eliminated the some effects of roughness.Improved the soil moisture estimation precision, and proofreaied and correct, with the practicality of further raising algorithm by face of land measured data.Save computing time, overcome NASA's iterative algorithm complexity in the past, be absorbed in local minimum easily, and the general scientific research personnel shortcoming that is difficult to realize.Be climate change research, weather forecast, evapotranspiration, agricultural feelings monitoring and disaster monitoring etc. provide effective means and technical support, this method to be worked out as one of Ministry of Agriculture's farming feelings monitoring system alternative approach.

Description of drawings

The present invention is further described below in conjunction with drawings and Examples.

Description of drawings

Bright temperature synoptic diagram [AMSR-E Team Leader Science ComputingFacility Science Software, the 2004.8.AMSR-E Browse﹠amp of rail falls in Figure 1A MSR-E; Quick look image (NASA): 1-69].

Surface radiation process under Fig. 2 AIEM simulation different condition

The relation of Figure 31 0.7GHz V polarization soil moisture and emissivity

The relation of Figure 41 8.7GHz V polarization soil moisture and emissivity

Fig. 5 soil moisture and NDE _{18.7GHz-10.7GHz}Relation

Fig. 6 Soil Moisture Inversion main flow synoptic diagram

Fig. 7 inverting obtains Chinese terrestrial soil moisture Butut

Embodiment

Here do concrete inverting with the AMSR-E image data in the Chinese land of scape 1 day 5 February in 2009 district, this method mainly comprises two steps, as Fig. 6.

The first step is read in the data of each frequency of AMSR-E, gets parms from header file, carries out the conversion of DN value to bright temperature;

In second step, utilize formula 5 to calculate bright temperature index;

In the 3rd step, on the basis in second step, utilize formula 6 to calculate soil moisture;

In the 4th step, after estimating in the 3rd step that the soil moisture that obtains is carried out geometry correction and merging, the soil moisture result that estimation obtains is shown in Figure 7;

In the 5th step, the check analysis of estimation distribution of results is from situation, inversion result and the national damage caused by a drought distribution situation basically identical that in February, 2009, whole nation meteorological department and local damage caused by a drought were reported.As we can see from the figure, the distribution trend of soil moisture is more rational.Coastal, the Yangtze river basin, the Huanghe valley, and the soil moisture around some big lakes is high especially, this and actual conditions are consistent.Higher at the southern area inverting value, main cause may be that southern cloud is many, though microwave is subjected to the influence of cloud smaller, in fact still is subjected to the influence of cloud, particularly more than the 18.7GHz.Different face of land types are influential to the algorithm inversion accuracy, the mixed pixel of large scale particularly, need to prove that in addition this algorithm mainly is applicable to area, the exposed face of land and the situation that does not have rainfall, many or rainfall and the thicker area of cloud layer are arranged for vegetation, inversion result need further be revised according to the real site data.

Claims (1)

1. from AMSR-E data inversion soil moisture method, the steps include:

The first step is to utilize the situation of exhausted big number on the theoretical model simulation face of land and set up database, utilizes the database of simulation, further analyzes the relation of microwave polarization index and soil moisture, and 10.7GHz and 18.7GHz V polarization polarization index computing formula are suc as formula 1.

${\mathrm{NDE}}_{18.7\mathrm{GHz}-10.7\mathrm{GHz}}=\frac{T_{18.7V}-T_{10.7V}}{T_{18.7V}+T_{10.7V}}$ (formula 1)

In second step, utilize the data of simulating in the first step to set up soil moisture and NDE _{18.7GHz-10.7GHz}Relation, obtain formula 2;

$\mathrm{SM}=0.33+10.99947*{\mathrm{NDE}}_{18.7\mathrm{GHz}-10.7\mathrm{GHZ}}+563.80628{\mathrm{NDE}}_{18.7\mathrm{GHz}-10.7\mathrm{GHz}}^2$ (formula 2)

The 3rd, calculate soil moisture with formula 1 and formula 2 from passive microwave data AMSR-E, and analyze on the spot, suitably correction has improved precision.

Images