SEARCH

SEARCH BY CITATION

Keywords:

  • FY-3;
  • microwave humidity sounder;
  • neural network;
  • water vapor density

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Principle on MWHS Sounding Atmosphere Profiles on FY-3 Satellite
  5. 3. Sounding Theory
  6. 4. Retrieval Method Development
  7. 5. Experimental Data and Results
  8. 6. Conclusion
  9. Acknowledgments
  10. References
  11. Supporting Information

[1] The atmospheric humidity profiles of clear sky in the Arctic regions were retrieved from the microwave humidity sounder on China's FY-3A satellite using the back-propagation neural network algorithm. The algorithm was developed using the reliable measurements of surface temperature, humidity, and pressure as well as atmospheric temperature and humidity profiles from radiosonde observations. Considering the influence of sounding geometry, different surface types, and atmospheric conditions, we improved the commonly used back-propagation artificial neural network by treating the Mexican hat wavelet function as a transfer function and transforming the input data space. The retrieved root-mean-square (RMS) error is about 0.12 g/m3 in absolute humidity (water vapor density) profiles and 12.7% in relative humidity profiles. Water vapor density retrievals in winter are in acceptable agreement with profiles from radiosonde, but the agreement of the summer data was not as good. Furthermore, the retrieval model has been used in another Arctic station with a mean water vapor density RMS error of 0.185 g/m3 and 18.3% for a relative humidity profile for all seasons in 2008 at 12:00 UT.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Principle on MWHS Sounding Atmosphere Profiles on FY-3 Satellite
  5. 3. Sounding Theory
  6. 4. Retrieval Method Development
  7. 5. Experimental Data and Results
  8. 6. Conclusion
  9. Acknowledgments
  10. References
  11. Supporting Information

[2] In the Central Arctic, there is little rainfall, but it is as wet as above the ocean. Temperatures are very low, and the water vapor can be as low as 2 g/m3. However, the mean relative humidity may be up to 85% higher than most of the midlatitude regions. In summer, the relative humidity can be as high as 95%. Furthermore, there is no meteorological station in the central Arctic area. Therefore, it is difficult and complex to forecast the temperature and water vapor density in the Arctic area.

[3] The second satellite FY-3B was launched in Taiyuan, China, on 5 November 2010. AsZhang et al. [2008]described, the second microwave humidity sounder (MWHS) on board the satellite, similar to MWHS on FY-3A, which was deployed in China in May 2008, was developed and built by the Center for Space Science and Applied Research, Chinese Academy of Sciences (CSSAR) [Zhang et al., 2006]. These two MWHS instruments are now working jointly for better meteorological services. MWHS measures brightness temperatures with three double-sideband channels centered at ±1, ±3, and ±7 GHz from the 183 GHz water vapor line and two additional channels centered at 150 GHz, with vertical and horizontal polarization. In the absence of clouds the atmospheric emission at these frequencies is primarily due to water vapor.

[4] Water vapor density profiles as well as integrated cloud liquid water play a key role in the study of global atmospheric circulation and evolution of clouds, especially in the Arctic area. The amount of water vapor in the air varies considerably, from practically none at all up to about 4% by volume, which includes water vapor, ozone, nitrogen, carbon dioxide, argon, and others. Certainly, the fact is that water vapor is the source of all clouds and precipitation, which would be enough to explain its importance.

[5] Therefore, it is very important to retrieve water vapor density profiles in the Arctic area. This paper mainly discusses water vapor density profiles in clear-sky conditions and will be a favorable foundation to research cloudy situations in the future.

[6] The paper is arranged as follows. Section 2 describes the details of the characteristics and operation mode of the MWHS instrument. Section 3 presents the sounding theory of the satellite microwave humidity sounder, including radiative transfer theory, atmospheric microwave absorption theory, and principle of humidity sounding on MWHS. Section 4 presents the neural network algorithm, its development, and application. Section 5 describes the experimental results, including the brightness temperature measurements of MWHS and retrievals, comparison, and analysis of water vapor density profiles in different altitudes and in different seasons. Finally, the paper is concluded in section 6.

2. Principle on MWHS Sounding Atmosphere Profiles on FY-3 Satellite

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Principle on MWHS Sounding Atmosphere Profiles on FY-3 Satellite
  5. 3. Sounding Theory
  6. 4. Retrieval Method Development
  7. 5. Experimental Data and Results
  8. 6. Conclusion
  9. Acknowledgments
  10. References
  11. Supporting Information

[7] The new microwave humidity sounder is part of a suite of instruments on board the FY-3A satellite, which is the second generation cross-track meteorological satellite and was launched in China in May 2008. FY-3A MWHS consists of five channels centered at 150 GHz (vertical and horizontal) and 183.3 GHz ±1, ±3, and ±7 GHz, known as channels 1–5, respectively [Zhang et al., 2006, 2008]. The first two channels fall in an atmospheric window which is affected mainly by the water vapor continuum and is strongly affected by the presence of liquid water as well. The other three high-frequency channels are centered in the wing of the 183.31 GHz water vapor absorption line as described inTable 1, which lists the characteristic of FY-3A MWHS for each channel.Figure 1shows the photo of MWHS on board the FY-3A satellite.Figures 2 and 3show the scanning geometric and pixels of MWHS. The instrument's antenna scans in cross-track type, sampling 98 Earth views within ±53.35° of nadir, each separated by 1.1° (16 km × 16 km on surface at nadir and 41 km × 27 km for the outer pixel), referred to as pixels 1–98. It also views an internal calibration target and cosmic space, which are used for radiometric real-time calibration. The scanning period is 2.667 s. Combined with radiosonde and other microwave sensors, these channels will allow regional sounding and global sounding of atmospheric humidity. Compared to similar instruments such as Advanced Microwave Sounding Unit B (AMSU-B) [Saunders et al., 1994; Goodrum et al., 2000], Microwave Humidity Sounder (MHS) on the POES satellite [Lambrigtsen and Calheiros, 2003], and Humidity Sounder for Brazil (HSB) [Aumann et al., 2003], MWHS has a wider single-track coverage of about 2700 km, while the swath of AMSU-B and MHS are 2250 km and 2180 km, respectively. Also, the space resolution is 16.3 km for AMSU-B and less than 16 km for FY-3A MWHS. Therefore, it has smaller scanning blindness and space resolution improvement of water vapor density retrievals, especially in the low latitude areas.

Table 1. Designed Channel Characteristics of FY-3A MWHS
ChannelCenter Frequency (GHz)Polarization V/HBand-Width (MHz)NEΔT (K)LO Precision (MHz)Antenna Beam Efficiency3 dB Beam WidthDynamic Range (K)
1150V10000.950≥93%1.13–340
2150H10000.750≥93%1.13–340
3183.31 ± 1H500130≥95%0.93–340
4183.31 ± 3H1000130≥95%0.93–340
5183.31 ± 7H20001.230≥95%0.93–340
image

Figure 1. The picture of FY-3A microwave humidity sounder.

Download figure to PowerPoint

image

Figure 2. Diagram showing scanning geometry of microwave humidity sounder.

Download figure to PowerPoint

image

Figure 3. The schematic of scanning pixels.

Download figure to PowerPoint

3. Sounding Theory

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Principle on MWHS Sounding Atmosphere Profiles on FY-3 Satellite
  5. 3. Sounding Theory
  6. 4. Retrieval Method Development
  7. 5. Experimental Data and Results
  8. 6. Conclusion
  9. Acknowledgments
  10. References
  11. Supporting Information

[8] Figure 4shows the high opacity of atmospheric microwave attenuation coefficients (oxygen and water vapor) using U.S. standard atmospheric profile (1976). Between 0 and 200 GHz, there are two oxygen lines which relate to the atmospheric temperature profiles and two water vapor lines which relate to the atmospheric humidity profiles. The water vapor lines are at 22.235 GHz and 183.31 GHz. The former line has low atmosphere attenuation which is too low to permit profiling, and its partial transparency is used to obtain the total columnar content, while the 183.31 GHz line and its wings are suitable for measuring vertical distribution of high-altitude atmospheric water vapor density profiles [Rosenkranz, 1993; Liebe, 1989]. Between these two lines, water vapor continuum slowly increases with frequency. Therefore, in clear sky conditions, the effect of resonance line absorption and emission from cloud liquid water and scattering may be neglected [Eymard, 2002].

image

Figure 4. Atmospheric absorption coefficients (dB/km) using U.S. standard atmosphere, 1976.

Download figure to PowerPoint

[9] According to the radiative transfer equation [Janssen, 1993] and atmospheric absorption model shown in equations (1)(4), the brightness temperature values of different channels based on radiosonde data sets can be calculated. The radiative schematic is shown in Figure 5, from which one can see that the brightness temperature observed by MWHS mainly includes three parts, i.e., the downwelling radiation and cosmic microwave background reflected from the surface back toward the satellite, which is denoted as Tr; the upwelling radiation from the surface at z = 0 to the satellite altitude H, which is denoted as Tu; and the radiation emitted from the surface, which is denoted as Te. All three are measured in Kelvins.

  • display math
  • display math
  • display math
  • display math
image

Figure 5. The schematic of observation brightness temperatures from cross-track satellite.

Download figure to PowerPoint

[10] According to Ulaby et al. [1986] and Rosenkranz et al. [1982], the water vapor weighting function can be expressed as follows:

  • display math

[11] Here kv(v) is the water vapor absorption coefficient per mass:

  • display math

and hv(z) is the water vapor burden:

  • display math

where α(z) is the absorption coefficient at height z in dB/km (Neper/km = dB/km × 0.2303); z is the height between the surface and the satellite altitude in kilometers; H is the satellite altitude in kilometers; θ is the incidence angle of satellite in degrees; εs is surface emissivity, which is different at different frequency; Ts is surface temperature in Kelvins; and Tb0 is the radiation from cosmic microwave background in Kelvins.

[12] In the Arctic, atmospheric humidity values are not much different from midlatitude conditions in summer. In winter the total water vapor is low (a few kg/m3), and as a consequence, all weighing function of MWHS operated at the wings of the 183 GHz absorption line are expected to peak near the surface and to have high surface contribution.

[13] Figure 6 shows the typical weighting function distributions for all MWHS channels calculated for the Arctic at nadir where surface temperature is 271.1 K, surface pressure is 100.2 kPa, and total water vapor content is 20 kg/m2 using a microwave propagation model. The weighting functions indicate the radiative contribution of each atmospheric layer to the measured radiance. For given atmospheric profiles and frequency, the peak altitude of the weighting function increases with the zenith angle increasing. This is due to increasing optical path length between the satellite and the surface when the instruments scan from nadir to larger angles. In window channels the weighting function peaks have their maximum closer to the surface. Most of the radiance measured by these window channels comes from the surface and the boundary layer, and these channels can be used to derive total precipitable water [Grody et al., 2001; Zhao and Weng, 2002].

image

Figure 6. The MWHS water vapor weighting functions of five channels for a typical Arctic atmosphere at nadir.

Download figure to PowerPoint

4. Retrieval Method Development

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Principle on MWHS Sounding Atmosphere Profiles on FY-3 Satellite
  5. 3. Sounding Theory
  6. 4. Retrieval Method Development
  7. 5. Experimental Data and Results
  8. 6. Conclusion
  9. Acknowledgments
  10. References
  11. Supporting Information

[14] The back-propagation (BP) artificial neural network (ANN) has been used widely and has retrieved temperature and humidity profiles with high accuracy [Shi, 2001]. It has many variants, so it is of very different behaviors to derive good results for a wide variety of problems when little is known about the research space like in many scientific disciplines. Churnside et al. [1994] successfully retrieved the temperature profiles using ANN with three layers in 1994. Solheim et al. [1998] used ANN and other retrieval methods to retrieve temperature profiles and humidity profiles successfully in 1998. Yao et al. [2005]demonstrated the BP ANN can significantly improve the temperature retrievals in all the weather conditions comparing to the IAPP model, especially at the lower levels in 2005. RPG-HATPRO and MP3000A series also constructed retrieval models using neural network algorithm to derive temperature and humidity profiles with high accuracy [Rose and Czekala, 2005, 2006, 2008; Radiometrics Corporation, 2008].

[15] ANN is essentially a nonlinear statistical regression between a set of predictors (in this case the observation vectors X) and a set of predictands (in this case profiles of atmospheric temperature Z). The structure of the ANN is shown in Figure 7. In this paper, we construct a three-layer ANN model. The layers 1, 2, and 3 represent the input layer, the hidden layer, and the output layer, respectively. The neurons of the input layer are represented by vectorXi (X1, X2X3, …, XL), where L is the number of the input neurons. The neurons of the middle layer are represented by vector Yi (Y1Y2Y3, … YM), where M is the number of the hidden neurons. The neurons of the output layer are represented by vector Zi (Z1Z2Z3, … ZN), where N is the number of the output neurons.

image

Figure 7. The schematic of artificial neural network with three layers.

Download figure to PowerPoint

[16] For the jth node in the hidden layer, it can be expressed as

  • display math

where S denotes the sigmoid function

  • display math

wij is the weighting of the connection between the jth hidden neuron and the ith input neuron, and bj denotes the bias in the jth neuron of the hidden layer.

[17] In order to overcome the disadvantages of reaching convergence slowly and easily to get stuck in local minimum, this paper presents BP neural network based on Mexican hat wavelet function as a transfer equation between the input layer and the hidden layer:

  • display math

[18] The Purelin linear function is applied between the output layer and the hidden layer, which can be referred from the Matlab neural network toolbox. As a result, the output values can be arbitrary within the range [0, 1]. The neuron of the output layer can be expressed as

  • display math

where wij is the weighting of the connection between the jth hidden neuron and the ith input neuron and bj denotes the bias in the jth neuron of the hidden layer.

[19] In the experiments, we first scale real continuous input data to fit in the range from 0 to 1; this is necessary to improve the retrieving accuracy. It is advisable to consider any other transformation of an input channel as a new input channel because transforming the input data space equivalently transforms the search spaces, which are different. The input channels chosen for the neural network can be considered to be taken from the set of possible channel transformations and combinations. For the problems presented in this paper, both input and output data are described in equations (8)(11).

[20] BP is the most popular method used to select values for ANN free parameters. It is done iteratively, calculating the error gradients of the data with respect to the free parameters and then updates them appropriately. The error gradients are calculated starting from the error on the outputs and moving backward. Epoch is named as iteration for all the training data.

5. Experimental Data and Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Principle on MWHS Sounding Atmosphere Profiles on FY-3 Satellite
  5. 3. Sounding Theory
  6. 4. Retrieval Method Development
  7. 5. Experimental Data and Results
  8. 6. Conclusion
  9. Acknowledgments
  10. References
  11. Supporting Information

5.1. Measurements for MWHS on Board FY-3A Meteorological Satellite

[21] This paper uses brightness temperatures of FY-3A satellite-based MWHS observations from 1 June 2008 to 31 May 2009 distributed at latitude from 78.5° to 82.5° and longitude from 10° to 60° where station 20046 and 01004 matched and the surface types are land. The brightness temperature observations are shown inFigure 8, located at 78.92°N latitude and 11.93°E longitude. The five colors indicate brightness temperatures in different channels. From Figure 8, one can easily see that the brightness temperatures in the channel of 182.31 GHz are most stable in the whole year. Brightness temperatures in each channel reflect the cumulative water vapor contribution at different altitudes. In order to simulate brightness temperature values at the same time every day, we use radiosonde information including temperature profiles, humidity profiles and pressure profiles, wind speed, and other information. Figure 9 shows the surface water vapor density values of the whole year from June 2008 to May the next year.

image

Figure 8. Brightness temperatures in a whole year in the Arctic region at 12:00 UT.

Download figure to PowerPoint

image

Figure 9. Surface water vapor density values in a whole year in Arctic region at 12:00 UT.

Download figure to PowerPoint

[22] The brightness temperature input variables in five channels are critical in the ANN retrieval model. The retrievals benefit their sensitivity and accuracy, therefore, the observation values in water vapor channels and window channels from FY-3A MWHS are shown inFigure 10 with a range of 0°–20°E latitude and 60°–90°N longitude.

image

Figure 10. The brightness temperature observations in five channels from FY-3A MWHS (time 20090901_0742). The range is within 0–20°E and 60–90°N.

Download figure to PowerPoint

[23] It is demonstrated that the brightness temperature values in the Arctic region are generally much more stable (approximately located at 250–280K) than midlatitude and tropical regions above the same surface type. The reason is due to the lower temperature and relatively larger water vapor density with less variance compared to midlatitude and tropical regions.

5.2. Data Processing and Simulation

[24] The paper uses three data sets, including training, test, and validation data sets from two stations. One is 01004, with latitude 78.92°N and longitude 11.93°E, and the other is 20046, with latitude 82.61°N and longitude 58.05°E. Then we selected radiosonde profiles from these stations at 12:00 UT during 1 year (from 1 June 2008 to 31 May 2009). For the radiosonde data sets, they provide the profiles of temperature, mixer ratio of water vapor, pressure, height and relative humidity. Exclude the radiosonde data which the height less than 15 km, and use cloud-judge function (if relative humidity is larger than 95%, then we assume it is cloudy) to exclude the data sets in cloudy sky. Here, we received a total of 361 data, and exclude 3% rainy data and incomplete data and 23% cloudy data. Then these remaining profiles are processed at discrete levels every 200 m up to 15 km. Although the number of independent measurements is only 50 levels output, this sampling ensures the retrieval profiles can accurately represented on the fixed levels.

[25] The brightness temperature values from five channels are read form MWHS level 1 data. Here, the surface temperature, surface pressure and relative humidity are also directly connected. The radiosonde data sets include various climate facets of the variability in clear-sky Arctic region and are numerous enough to be split into training, test and validation sets. Notice that training data sets must be representative for all the test and validation sets. In the training process, surface information (surface temperature, pressure and emissivity) and brightness temperatures in five channels are linearly normalized as input values with the range of [0, 1]. Similarly, water vapor density profiles are normalized as output values. The Gaussian noises are added in the surface temperature, pressure and observed brightness temperatures, which are 0.5 K, 0.3 kPa and 0.5 K, respectively. This extends the training data sets slightly and reduces the sensitivity of the network to noise in the data and can represent all the errors affecting the observations.

5.3. Retrievals and Analysis

[26] Typical results for all seasons are shown in Figures 11 and 12. Figure 11 shows the comparison between water vapor density retrievals from ANN model and from radiosonde data sets. Each single dot represents water vapor density in one layer. Here “23” and “215” mean the number of test data and training data. There is a good linear relationship which indicates that the water vapor density retrievals are well agreement with the profiles from radiosonde observations. Figure 12 shows bias and RMS in percentage of the water vapor density between water vapor density retrievals from ANN model and from radiosonde data sets. It has the same specification with single point and test box. The largest bias error appeared at the altitude of ∼1.5–2 km. It has higher accuracy at higher altitude. But in Figure 12 (right), the relative water vapor RMS is higher in high latitude. This is because its value is relative to the signal, and in higher layers, the water vapor content is much lower than the layers near the surface.

image

Figure 11. Comparison between relative humidity retrievals using ANN and relative humidity from radiosonde data sets.

Download figure to PowerPoint

image

Figure 12. (left and right) Relative humidity RMS error between relative humidity retrievals using ANN and relative humidity from radiosonde data sets.

Download figure to PowerPoint

[27] To explain the accuracy of water vapor density retrievals explicitly, Figure 13 shows one profile comparison and bias of water vapor density profile between retrievals using ANN method and from radiosonde data sets in randomly 1 day at a certain time. Here the date is at 12:00 UT on 29 November 2008; the surface temperature is 267.05 K, the total water vapor is 6.103 kg/m2, and the surface pressure is 98.16 kPa.

image

Figure 13. One example of comparison between radiosonde and ANN retrievals from FY-3A MWHS.

Download figure to PowerPoint

[28] The RMS error is considered to be the criteria to judge the retrievals deviated from parameters from radiosonde data. It can be expressed as

  • display math

where zrad and zretr are the radiosonde observation and water vapor density retrievals, respectively, and N is the total number of comparisons.

[29] Relative RMS can be expressed as

  • display math

[30] Here, V is integrated water vapor content and can be calculated as

  • display math

where q is specific humidity, p is pressure, ρ is liquid water vapor density, g is acceleration of gravity, and p0 is surface pressure.

[31] Figures 14 and 15 are mean RMS error for discrete altitudes and for different data. In Figure 14, it apparently can be seen that in higher altitudes, FY-3A MWHS can retrieve water vapor density values with smaller bias and root-mean-square error; this means that in higher altitudes, MWHS retrievals are in good agreement with the values from radiosonde observations and vice versa in the lower altitudes. This is the system characteristic of satellite microwave humidity sounder. Also in different seasons, the RMS error values are varied as the season changes.Figure 15 shows water vapor density RMS of the period 1 June 2008 to 31 May 2009 in station 01004; several days were selected in each month. In the later days of spring and earlier days of summer, the RMS is larger than other days in the whole year.

image

Figure 14. Vertical mean RMS of water vapor density profiles from 1 June 2008 to 31 May 2009.

Download figure to PowerPoint

image

Figure 15. The RMS of different data sets according to the time from June 1, 2008 to May 31, 2009.

Download figure to PowerPoint

[32] As far as we know, the wavelet function-based BP neural network has not been used in the water vapor density retrievals. This paper presents a Mexican hat function-based BP neural network which plays an important role in retrieval research.Figure 16shows the comparison and BIAS and RMS of water vapor density retrievals with respect to radiosonde data sets. It shows that all the above methods can retrieve humidity profiles with high accuracy. Based on the character of satellite-borne MWHS, these methods have better retrievals in higher altitude than in lower altitude of troposphere, while the BP neural network with the Mexican hat method has relatively better retrievals than the BPANN method, especially in the layers from 3 km up to 10 km.

image

Figure 16. Comparison of water vapor density profiles using the three methods.

Download figure to PowerPoint

[33] Besides analyzing the RMS error in different atmospheric altitude, it also analyzes the variation among the whole year from 1 June 2008 to 31 May 2009. A sample was collected every 10 days throughout the year, a total of 36 data sets were chosen, and the root-mean-square errors in different seasons are shown inTable 2.

Table 2. The Root-Mean-Square Error in Different Seasons in Testing Area
Station: ArcticSpringSummerAutumnWinter
Mean Temperature (K)256.0281257.3486254.3049249.9409
Mean Pressure (kPa)99.150298.204999.086698.5673
Mean Water Vapor Density (g/m3)1.18021.22931.00390.5991
Mean RMS (g/m3)0.13410.10110.10430.1202
Mean RMS (%)13.707912.023913.427311.7503

[34] From Table 2, we can easily see that at station 20046, in summer, the RMS error is the largest one; it means that in summer, the water vapor density profile retrievals are the least accurate. In winter, the RMS error is smallest, and the water vapor profiles in winter can achieve the best agreement with them from radiosonde observations. In spring, the RMS error of retrievals is larger than in autumn. In the last row, the retrievals are expressed as a percentage, not an absolute number. It demonstrates that, compared to integrated water vapor content, in winter it has the smallest relative RMS (11.75%) and in spring it has the largest relative RMS (13.70%). Compared to surface mean water vapor density, in winter the mean water vapor density is 0.5991 g/m3 with a RMS of ∼0.1202 g/m3 that is about 20%. In the other seasons the mean percentage RMS error is about 10%.

[35] To demonstrate the performance of this algorithm outside of the region, the algorithm used in this paper has been applied to locations that do not have radiosonde soundings. For a reliable comparison, we selected another site in the Arctic around the radiosonde station (01004) where the climate situations are mostly within training data sets to validate the algorithm (without retraining the neural network). Through this experiment, we can assess how well it will perform on other regions and judge the impact of these retrievals. The validation results are shown in Table 3. Because there are only a few radiosonde stations in the Arctic region, the authors used brightness temperature values for five channels in different locations, like latitude 78°N–82°N and longitude 12°E–50°E, and retrieved the water vapor density profiles. Compared with profiles from other instruments, the RMS is less than 0.3 g/m3 and relative RMS (compared to integrated water vapor content) is about 21%, which are acceptable.

Table 3. The Root-Mean-Square Error in Different Seasons in Validating Area
Station: 78°N to 82°N 12°E to 50°ESpringSummerAutumnWinter
Mean Temperature (K)264.2842277.7597269.3013261.9129
Mean Pressure (kPa)101.1343101.2170100.4361100.3406
Mean Water Vapor Density (g/m3)2.01945.69382.64041.7771
Mean RMS (g/m3)0.14350.26030.19720.1373
Mean RMS (%)18.9720.9817.4315.92

6. Conclusion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Principle on MWHS Sounding Atmosphere Profiles on FY-3 Satellite
  5. 3. Sounding Theory
  6. 4. Retrieval Method Development
  7. 5. Experimental Data and Results
  8. 6. Conclusion
  9. Acknowledgments
  10. References
  11. Supporting Information

[36] MHWS plays an important role in studying global climate and is the main remote sensing instrument for meteorology and disaster. It works in all weather and all day providing the observation of brightness temperature which is a cumulative contribution of vertical water vapor, humidity, temperature, cloud liquid water content, and so on. The MHWS water vapor channels have unique advantages, such as high sensitivity to water vapor, integrated water vapor content, and cloud liquid water content. Water vapor density profile information can be obtained from microwave radiance measurements from five channels of MWHS. It has coarse vertical resolution but can provide useful information for humidity analysis, especially in the upper troposphere. The radiances are to be used to generate a water vapor analysis for numerical weather prediction, especially in Arctic regions where there is no meteorological observing station.

[37] The ANN method handles the high variability of the water vapor density problem very well, particularly taking into account surface information. The use of the back-propagation ANN can best update the weights for the particular problem. The retrieved RMS (root-mean-square error) is within 0.12 g/m3in water vapor density profiles and 12.7% in relative humidity profiles. For the whole year in this experiment, the water vapor density retrievals are in best agreement with the profiles from radiosonde observations in summer and are worst in spring, and there is a relatively small difference between them. Because of the central Arctic area, cloud liquid water content appeared in all seasons and nearly 30% of the data sets are cloudy contamination. While this paper mainly talks about clear-sky situations, part of the cloudy areas with smaller liquid water content have been considered as clear situations.

[38] To validate the algorithm (without retraining the neural network), to assess how well it will perform on other regions, and to judge the impact of these retrievals, another station is selected. From Table 3, we demonstrated that the algorithm ANN with a Mexican hat function can retrieve water vapor density profiles with a RMS of 0.1435 g/m3, 0.2603 g/m3, 0.1972 g/m3, and 0.1373 g/m3 for each season in 2008 at 12:00 UT. Also, Table 3 shows in all seasons that the RMS error is higher than that of the testing area. While in spring and winter the increase is almost negligible, in summer it increases by a factor of almost 2.5 to 0.26 g/m3, because the validation region has a large difference in longitude compared to the test region, and they have a large difference in water vapor content and distribution in summer and spring. The experiments in this paper demonstrate that MWHS of FY-3A is helpful to improve numerical weather prediction in Arctic regions. In the future, we will combine the second MWHS on board the FY-3B satellite to retrieve more accurate clear and cloudy water vapor profiles in Arctic regions.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Principle on MWHS Sounding Atmosphere Profiles on FY-3 Satellite
  5. 3. Sounding Theory
  6. 4. Retrieval Method Development
  7. 5. Experimental Data and Results
  8. 6. Conclusion
  9. Acknowledgments
  10. References
  11. Supporting Information

[39] The MWHS data used in this article were provided by National Satellite Meteorological Center, China Meteorological Administration. Part of the radiosonde data sets were provided by Meteorological Observation Center, China Meteorological Administration. The authors are grateful to the Department of Space Sciences, Wyoming University, and the Comprehensive Large Array-data Stewardship System (CLASS) of the U.S. National Oceanic and Atmospheric Administration for providing the data sets (http://weather.uwyo.edu/upperair and http://raob.fsl.noaa,gov/).

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Principle on MWHS Sounding Atmosphere Profiles on FY-3 Satellite
  5. 3. Sounding Theory
  6. 4. Retrieval Method Development
  7. 5. Experimental Data and Results
  8. 6. Conclusion
  9. Acknowledgments
  10. References
  11. Supporting Information
  • Aumann, H. H., et al. (2003), AIRS/AMSU/HSB on the Aqua mission: Design, science objectives, data products, and processing systems, IEEE Trans. Geosci. Remote Sens., 41, 253264, doi:10.1109/TGRS.2002.808356.
  • Churnside, J. H., T. A. Stermitz, and J. A. Schroeder (1994), Temperature profiling with neural network inversion of microwave radiometer date, J. Atmos. Oceanic Technol., 11(1), 105109, doi:10.1175/1520-0426(1994)011<0105:TPWNNI>2.0.CO;2.
  • Eymard, L. (2002), Retrieval of integrated water vapour and cloud liquid water contents, in Remote Sensing of Atmosphere and Ocean From Space: Models, Instruments and Techniques, edited by F. S. Marzano and G. Visconti, pp. 89106, Kluwer Acad., Dordrecht, Netherlands.
  • Goodrum, G., K. B. Kidwell, and W. Winston (Eds.) (2000), NOAA KLM user's guide, report, NOAA, Silver Spring, Md. [Available at http://www2.ncdc.noaa.gov/docs/klm/index.htm.]
  • Grody, N., J. Zhao, R. Ferraro, F. Weng, and R. Boers (2001), Determination of precipitable water and cloud liquid water over oceans from the NOAA 15 advanced microwave sounding unit, J. Geophys. Res., 106, 29432953, doi:10.1029/2000JD900616.
  • Janssen, M. A. (1993), Atmospheric Remote Sensing by Microwave Radiometry, Wiley-Intersci., New York.
  • Lambrigtsen, B. H., and R. V. Calheiros (2003), The humidity sounder for Brazil - An international partnership, IEEE Trans. Geosci. Remote Sens., 41, 352361, doi:10.1109/TGRS.2002.808304.
  • Liebe, J. H. (1989), MPM–An atmospheric millimeter-wave propagation model, Int. J. Infrared Millimeter Waves, 10(6), 631650, doi:10.1007/BF01009565.
  • Radiometrics Corporation (2008), MP-series microwave profilers, operating manual, Boulder, Colo. [Available at http://www.radiometrics.com/MP_Specifications_7-3-07.pdf.].
  • Rose, T., and H. Czekala (2005), Accurate atmospheric profiling with the RPG-HATPRO humidity and temperature profiler, operating manual, Radiometer Phys. GmbH, Meckenheim, Germany.
  • Rose, T., and H. Czekala (2006), RPG-HATPRO, RPG-TEMPRO, RPG-HUMPRO: Humidity and temperature profilers, operating manual, Radiometer Phys. GmbH, Meckenheim, Germany.
  • Rose, T., and H. Czekala (2008), RPG-150-90/RPG-DP150-90 high sensitivity LWP radiometers: RPG's atmospheric remote sensing radiometers, operating manual, Radiometer Phys. GmbH, Meckenheim, Germany.
  • Rosenkranz, P. W. (1993), Absorption of microwaves by atmospheric gases, in Atmospheric Remote Sensing by Microwave Radiometry, edited by M. A. Janssen, pp. 3779, Wiley-Intersci., New York.
  • Rosenkranz, P. W., M. J. Komichak, and D. H. Staelin (1982), A method for estimation of atmospheric water vapor profiles by microwave radiometry, J. Appl. Meteorol., 21, 13641370, doi:10.1175/1520-0450(1982)021<1364:AMFEOA>2.0.CO;2.
  • Saunders, R. W., S. J. English, and D. C. Jones (1994), AMSU-B: A new tool for atmospheric research, Proc. SPIE, 2313, 98, doi:10.1117/12.197338.
  • Shi, L. (2001), Retrieval of atmospheric temperature profiles from AMSU-A measurement using neural network approach, J. Atmos. Oceanic Technol., 18(3), 340347, doi:10.1175/1520-0426(2001)018<0340:ROATPF>2.0.CO;2.
  • Solheim, F., J. R. Godwin, E. R. Westwater, Y. Han, S. J. Keihm, K. Marsh, and R. Ware (1998), Radiometric profiling of temperature, water vapor and cloud liquid water using various inversion methods, Radio Sci., 33, 393404, doi:10.1029/97RS03656.
  • Ulaby, F. T., R. K. Moore, and A. K. Fung (1986), Passive microwave sensing of the atmosphere, in Microwave Remote Sensing: Active and Passive, vol. 3, From Theory to Applications, edited by F. T. Ulaby, pp. 12821378, Artech House, Norwood, Mass.
  • Yao, Z., H. Chen, and L. Lin (2005), Retrieving atmospheric temperature profiles from AMSU-A data with neural networks, Adv. Atmos. Sci., 22(4), 606616, doi:10.1007/BF02918492.
  • Zhang, S., J. Li, J. Jiang, Y. Zhang, Z. Wang, and X. Dong (2006), Microwave Humidity Sounder (MWHS) of the “FY-3” Meteorological Satellite of China, in Proceedings of the Microwave Technology and Techniques Workshop—Enabling Future Space Systems, 15–16 May 2006, ESTEC, Noordwijk, The Netherlands [CD-ROM], ESA SP-632, edited by L. Conroy, Eur. Space Agency, Paris.
  • Zhang, S., J. Li, J. Jiang, M. Sun, and Z. Wang (2008), Design and development of microwave humidity sounder for FY-3 meteorological satellite, J. Remote Sens., 12(3), 199207.
  • Zhao, L., and F. Weng (2002), WENG, F. Retrieval of ice cloud parameters using the advanced microwave sounding unit, J. Appl. Meteorol., 41, 384395, doi:10.1175/1520-0450(2002)041<0384:ROICPU>2.0.CO;2.

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Principle on MWHS Sounding Atmosphere Profiles on FY-3 Satellite
  5. 3. Sounding Theory
  6. 4. Retrieval Method Development
  7. 5. Experimental Data and Results
  8. 6. Conclusion
  9. Acknowledgments
  10. References
  11. Supporting Information
FilenameFormatSizeDescription
rds5847-sup-0001-t01.txtplain text document0KTab-delimited Table 1.
rds5847-sup-0002-t02.txtplain text document0KTab-delimited Table 2.
rds5847-sup-0003-t03.txtplain text document0KTab-delimited Table 3.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.