Next Article in Journal
Comparison of River Basin Water Level Forecasting Methods: Sequential Neural Networks and Multiple-Input Functional Neural Networks
Previous Article in Journal
New Textural Indicators for Assessing Above-Ground Cotton Biomass Extracted from Optical Imagery Obtained via Unmanned Aerial Vehicle
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Development of CO2 Band-Based Cloud Emission and Scattering Indices and Their Applications to FY-3D Hyperspectral Infrared Atmospheric Sounder

Center of Data Assimilation for Research and Application, Nanjing University of Information Science & Technology, Nanjing 210044, China
*
Author to whom correspondence should be addressed.
Remote Sens. 2020, 12(24), 4171; https://0-doi-org.brum.beds.ac.uk/10.3390/rs12244171
Submission received: 27 November 2020 / Revised: 17 December 2020 / Accepted: 17 December 2020 / Published: 19 December 2020

Abstract

:
The Hyperspectral Infrared Atmospheric Sounder (HIRAS) onboard the Feng Yun-3D (FY-3D) satellite is the first Chinese hyperspectral infrared instrument. In this study, an improved cloud detection scheme using brightness temperature observations from paired HIRAS long-wave infrared (LWIR) and short-wave infrared (SWIR) channels at CO2 absorption bands (15-μm and 4.3-μm) is developed. The weighting function broadness and a set of height-dependent thresholds of cloud-sensitive-level differences are incorporated into pairing LWIR and SWIR channels. HIRAS brightness temperature observations made under clear-sky conditions during a training period are used to develop a set of linear regression equations between paired LWIR and SWIR channels. Moderate-resolution Imaging Spectroradiometer (MODIS) cloud mask data are used for selecting HIRAS clear-sky observations. Cloud Emission and Scattering Indices (CESIs) are defined as the differences in SWIR channels between HIRAS observations and regression simulations from LWIR observations. The cloud retrieval products of ice cloud optical depth and cloud-top pressure from the Atmospheric Infrared Sounder (AIRS) are used to illustrate the effectiveness of the proposed cloud detection scheme for FY-3D HIRAS observations. Results show that the distributions of modified CESIs at different altitudes can capture features in the distributions of AIRS-retrieved ice cloud optical depth and cloud-top pressure better than the CESIs obtained by the original method.

Graphical Abstract

1. Introduction

Hyperspectral infrared instruments onboard satellites have been a fundamental part of the earth observation system (EOS) since the twenty-first century. Owing to high spectral resolutions, high vertical resolution atmospheric profiles of temperature, humidity and cloud amount, water content, and cloud-top pressure can be retrieved [1,2,3]. The first hyperspectral infrared sounder is the 2378-channel Atmospheric Infrared Sounder (AIRS) onboard the National Aeronautics and Space Administration (NASA) Aqua satellite launched in May 2002 [4]. The second hyperspectral infrared sounder is the 8461-channel Infrared Atmospheric Sounding Interferometer (IASI) onboard the EUMETSAT Polar System’s three polar-orbiting meteorological satellites MetOp-A/B/C [5,6]. The Cross-track Infrared Sounder (CrIS) is the third hyperspectral sounder onboard the Suomi National Polar-Orbiting Partnership (S-NPP) satellite launched in October 2011. CrIS is a hyperspectral Fourier transform spectrometer with 1305 channels at normal spectrum resolution (NSR) initially [7] and was been updated to 2211 channels at full spectrum resolution (FSR) on National Oceanic and Atmospheric Administration (NOAA) -20, which was launched in November 2017. The first Chinese hyperspectral infrared sounder is the Hyperspectral Infrared Atmospheric Sounder (HIRAS), which is a spaceborne Fourier transform spectrometer with 2275 spectral channels. HIRAS has been onboard the FengYun-3D (FY-3D) satellite launched in November 2017 [8,9,10]. Observations from these hyperspectral infrared instruments have been used worldwide for numerical weather prediction (NWP) through data assimilation and climate research using retrieval products [1,11,12].
Atmospheric data assimilation is an important and complex process in NWP, in which the detection of cloud-contaminated satellite data is a key. Cloud detection methods can be broadly categorized into two categories: machine learning methods and classical algorithm-based methods. The machine learning techniques train forecast models by fusing model output with input samples to provide additional decision support [13,14] used an artificial neural net with many variations such as support vector machine, fusing multi-scale convolution features, deep learning, decision tree, Bayesian classification, etc. Machine learning methods are flexible because they simulate decisions on training data, but not consistent because model training depends on input data. Most classical algorithm-based methods used for cloud detection are based on thresholds, which are effective for different types of cloud but with poor universality [15]. Some of these generality issues have been overcome by statistical methods such as Principal Component Analysis [16]. Many researchers implement cloud detection techniques using physical parameters of clouds, such as shape attributes, the fusion of multi-scale convolutional features of the cloud net, color transformation, cloud density, cloud shadows, and clear-sky background differences [15]. Threshold methods identifying cloud-contaminated data directly based on brightness temperature differences have also been widely used in cloud detection and was applied to infrared imagers such as the Advanced Very High-Resolution Radiometer [17], the Moderate-resolution Imaging Spectroradiometer [18], and the Visible Infrared Imager Radiometer Suite [19]. By ranking the channels according to the channel sensitivity to clouds [20] developed a cloud detection method for AIRS observations selecting those channels whose weighting functions peak above the cloud top pressure for assimilation and eliminating lower-level channels in NWP models.
The CO2 absorption at both LWIR (15-μm) and SWIR (4.3-μm) bands from hyperspectral infrared sounders provide a unique opportunity to detect clouds in different layers of the atmosphere. Because CO2 is a uniformly mixed gas in the atmosphere, differences between LWIR and SWIR channels at similar altitudes provide information of cloud in addition to atmospheric temperatures at high vertical resolutions [21,22] developed a CO2 slicing technique to determine cloud-top pressure according to the ratio of observations from two channels at CO2 absorption bands (4.3-μm, 15-μm) that is suitable for cirrus cloud identification [23,24] used the CO2 slicing technique to determine the cirrus cloud occurrence, height and effective emissivity with based on HIRS (High-resolution Infrared Radiation Sounder) observations. A new cloud detection scheme was recently proposed by [25] to define a set of the Cloud Emission and Scattering Indices (CESIs) for cloud detection by pairing CrIS LWIR and SWIR CO2 channels. The algorithm is simple and can detect clouds in different vertical layers of the atmosphere. This study makes further improvements to this method and applies it to HIRAS observations from the FY-3D satellite. The concept of weighting function (WF) broadness is newly added, the threshold of the cloud-sensitive-level difference is made height-dependent when matching the LWIR and SWIR channels of HIRAS, the regression coefficients of CESIs are determined by HIRAS clear-sky observations. Cloud retrieval products from AIRS are used to demonstrate the effectiveness of the improved cloud detection scheme for HIRAS observations.
This paper is organized as follows. Section 2 describes the characteristics of HIRAS observations and channel characteristics. Section 3 gives details about the improved cloud detection scheme. In Section 4, results are provided to compare the actual effects of the improved and original cloud detection schemes when applied to HIRAS observations. Section 5 presents the summary and conclusions.

2. HIRAS Data Characteristics

The FY-3D satellite is one of the Chinese second-generation polar-orbiting meteorological satellite series. It was launched in November 2017 with an altitude of 830.5 km above the earth’s surface, an orbital inclination angle of 98.6°, and a local equator-crossing time of 13:40. The HIRAS has been onboard the FY-3D satellite since its launch. It is the first generation of Chinese hyperspectral infrared sounders, with 2275 spectral channels located in three spectral bands: the LWIR band from 650 to 1135 cm−1, the mid-wave infrared (MWIR) band from 1210 to 1750 cm−1, and the SWIR band from 2155 to 2550 cm−1. The spectral resolution of FSR HIRAS data is 0.625 cm−1 for all channels. Along each scan line of HIRAS, there are 29 fields of regard (FORs) with two adjacent observation positions separated by 3.6°, and each FOR comprises an array of 2 × 2 fields of view (FOVs). The view angle of each FOV is 1.1°, corresponding to a nadir spatial resolution of about 16 km. It takes 10 s to finish a single scan cycle. The swath width is 2250 km, corresponding to a range of scan angles from the nadir of ±50.4°.
Due to the coefficient limitation provided by the Radiative Transfer for TOVS (RTTOV) [26,27], this study uses the NSR (0.625 cm−1 for LWIR and 2.5 cm−1 for SWIR channels) HIRAS data. Figure 1 shows the brightness temperature and peak WF altitudes for HIRAS LWIR (670–755 cm−1, 138 channels) and SWIR (2200–2400 cm−1, 81 channels) channels at CO2 bands simulated by RTTOV (version 12.3) model with the U.S. standard profile as input. The RTTOV model was developed by the European Centre for Medium-Range Weather Forecasts (ECMWF) [17,27]. Although the designed spectral distribution of the simulated HIRAS brightness temperature and peak WFs shown in Figure 1 are similar to CrIS channels [28], we find that brightness temperature observations of HIRAS SWIR channels with peak WFs above 200 hPa are noisy. Figure 2a provides an example for HIRAS SWIR channel 2007 with center frequency located at 2382.5 cm−1, and peak WF located at 133 hPa. Noise dominates the spatial distribution of HIRAS brightness temperature observations at this channel (Figure 2a). The spatial distribution of the CrIS SWIR channel with the same frequency (2382.5 cm−1, Figure 2b) is more reasonable and meaningful. For lower-level channels, noise does not dominate HIRAS observations (it still affects observations in high latitudes). Figure 2c,d show that brightness temperature observations of the HIRAS (Figure 2c) and CrIS (Figure 2d) SWIR channels with the same frequency at 2227.5 cm−1 and peak WF at 535 hPa are similar, showing higher brightness temperatures in the tropics, higher brightness temperatures near nadir (scan angle is zero at nadir) than larger scan angles, and some cloud features with extremely low brightness temperature due to ice scattering. This problem of the HIRAS high-level channels has been reported to a relevant scientist of the HIRAS instrument development team at the China Meteorological Administration. In this study, SWIR channels with WF peaks above 200 hPa are not considered, limiting CESIs to below the 200-hPa altitude.

3. Description of the Modified Cloud Detection Scheme

Here, HIRAS LWIR (670–755 cm−1, 138 channels) and SWIR (2200–2400 cm−1, 81 channels) channels at CO2 bands are paired using the cloud detection method of [25] with modifications. The new method consists of the following three sequential steps: Step 1 pairs LWIR channels with SWIR channels based on WF profiles; step 2 selects the LWIR and SWIR channels paired in step 1, based on similar cloud-sensitive levels and minimum root mean square error (RMSE) between them; and step 3 finally calculates the CESIs. Details are given next.

3.1. Pairing LWIR and SWIR Channels Based on WFs

For each LWIR channel and its paired SWIR channel, the difference between their peak WFs ( Δ p W F p e a k ) should be the same or less than 50 hPa, i.e.,
Δ p W F p e a k = | p L W W F p e a k p S W W F p e a k | 50   hPa ,
where Δ p L W W F p e a k and Δ p S W W F p e a k are determined based on the WF profiles simulated by the RTTOV model with the U.S. standard profile as input. This was the only criterion used by original method in step 1. In fact, a single LWIR channel could be paired with several SWIR channels under the criterion expressed by Equation (1).
Figure 3 shows an example where LWIR channel 143 (13.5 μm) is paired with three SWIR channels, i.e., channels 2023 (4.18 μm), 1715 (4.54 μm), and 1751 (4.5 μm). The WF profiles of SWIR channels 1715, and 1751 are much different from SWIR channel 2023 and LWIR channel 143. So, we propose adding another criterion based on the broadness (B) of the WFs (Figure 4):
Δ B = | B L W B S W | 200   hPa
where B L W and B S W are the broadness of the WFs of the LWIR and SWIR channels paired by Equation (1). Their values are determined by the following constraints:
B L W = p 1 , L W p 2 , L W ,        B S W = p 1 , S W p 2 , S W ,
W ( p 2 , L W ) = W ( p 2 , S W ) ,        W ( p 2 , S W ) = W ( p 1 , S W ) ,
p 1 , L W p 2 , L W W L W d ( ln p ) p s 0 W L W d ( ln p ) = 3 4 ,        p 1 , S W p 2 , S W W S W d ( ln p ) p s 0 W S W d ( ln p ) = 3 4 ,
where W L W and W S W are the WFs of the LWIR and SWIR channels, respectively, p 1 , L W ( p 1 , S W ) and p 2 , L W ( p 2 , S W ) are the lower and upper boundaries of the area satisfying Equation (5) for the LWIR (SWIR) channel, and p s is the surface pressure. Equation (5) is based on the work by [29]. Figure 4 illustrates the areas satisfying Equation (3) of LWIR channel 143 and SWIR channel 1751, accounting for three-quarters of the total area surrounded by the y-axis and the corresponding WFs. For the example shown in Figure 3, only SWIR channel 2023 meets the criteria set by Equations (2)–(5), so that the other two SWIR channels, i.e., channels 1715 and 1751 are eliminated by the newly added criterion on WF broadness in step 1.

3.2. Pairing LWIR and SWIR Channels with Similar Cloud-Sensitive Levels and Minimum RMSE

The cloud-sensitive levels are determined by the following equations [20]:
| R L W c l o u d R L W c l e a r R L W c l e a r | p C S , L W 0.01 , | R S W c l o u d R S W c l e a r R S W c l e a r | p C S , S W 0.1
where R L W c l e a r ( R S W c l e a r ) represents the RTTOV-simulated radiances of the LWIR (SWIR) channel under clear-sky conditions, and R L W c l o u d ( R S W c l o u d ) represents the RTTOV-simulated radiances of the LWIR (SWIR) channel with an opaque cloud (as a blackbody which fills the radiometer field of view) located at pressure level p L . As p L varies from 150 hPa to the surface, the fractional differences in radiance between cloudy and clear-sky simulations, i.e., the left-hand sides of the two equations in (6), are calculated until they meet the criteria set by Equation (6). The values of p L that satisfy Equation (6) are defined as the cloud-sensitive levels, denoted as p C S , L W and p C S , S W for LWIR and SWIR channels, respectively. The thresholds for LWIR (0.01) and SWIR (0.1) in Equation (6) are determined empirically based on their noise level and cloud sensitivity. The impacts of clouds below the cloud-sensitive levels are very small and can be neglected. The difference in cloud-sensitive levels between LWIR and SWIR channels, Δ p C S = p C S , L W p C S , S W , is used for further screen. The original method set a fixed threshold for | Δ p C S | as 50 hPa. We find that this threshold works well for channels whose peak WFs are located in the lower troposphere. However, for channels whose peak WFs are located in the upper levels of the atmosphere, the 50-hPa threshold is not strict enough. In this study, a set of height-dependent thresholds for | Δ p C S | are proposed and used for pairing LWIR and SWIR channels in this step, and in the meantime the brightness temperature RMSE between the chosen LWIR and SWIR channels should be minimum.
Figure 5 shows a flow chart summarizing the major procedures involved in pairing the HIRAS LWIR and SWIR channels described above.

3.3. Calculating the CESI

Having obtained the paired LWIR and SWIR channels, a linear regression model of brightness temperature observations between LWIR and SWIR channels of each pair is established:
T b , r e g r e s s i o n S W I R = α T b L W I R + β
where T b , r e g r e s s i o n S W I R is the modeled SWIR brightness temperature, and T b L W I R represents the observed LWIR brightness temperature. In the original method, simulated brightness temperature was used as T b to calculate the regression coefficients with 50 ECMWF profiles as input for model. In this study, HIRAS brightness temperature observations under clear-sky condition are used as T b for regression calculation. This approach avoids the simulation errors caused by profiles and ensures a sufficient amount of data for regression. The regression coefficients α and β are derived by minimizing the sum of the squares of the deviations between HIRAS LWIR and SWIR observations:
min α , β ( i = 1 n ( T b , S W , i o b s , c l e a r α T b , L W , i o b s , c l e a r β ) 2 )
where the subscript “i” represents an index for summation over a data sample. HIRAS brightness temperature observations under clear-sky condition are selected based on the collocation with MYD035 cloud mask products (retrieved from MODIS). MODIS is a 36-band spectroradiometer measuring visible and infrared radiation, whose retrievals have been widely applied to cloud detection [30]. The MODIS instruments were launched onboard the Terra and Aqua satellites in December 1999 and May 2002, respectively. Aqua is part of the A-train constellation of EOS satellites, which flies in a polar orbit at 705 km. The local equator-crossing time of the Aqua satellite is 13:30, which is close to that of the FY-3D satellite (13:40). The MYD035 cloud mask product has a 5-km resolution, which is higher than the 16-km resolution (at nadir) of HIRAS observations. Thus, HIRAS clear-sky observation is determined as: all collocated MYD035 cloud mask products within the HIRAS observation FOV are clear. The collocation criterion on the observation time difference between HIRAS and MODIS is less than 16 min (Section 4.2).
When the linear relationships of each paired LWIR/SWIR channels are obtained, the CESI can be finally defined as
C E S I = T b , S W r e g ( T b , L W o b s , α , β ) T b , S W o b s = ( α T b , L W o b s + β ) T b , S W o b s .
In the presence of ice particles, the cloud-top emission and scattering intensities between LWIR channel (15-μm) and SWIR channel (4.3-μm) at CO2 bands are different. Based on brightness temperature observations, we find that the cooling effect on SWIR channel is more significant if ice clouds exist. Therefore, CESIs reflect the intensity of cooling for SWIR channels which can be used to determine whether there are ice clouds at different altitudes. Similar CESI cloud detection algorithms have been applied to the microwave temperature sounder and the humidity sounder [31].

4. Results

4.1. Pairing HIRAS LWIR and SWIR Channels

Table 1 lists the total number of LWIR and SWIR pairs selected by original (second column) and modified (third column) method described in Section 3.1 after step 1, respectively. The peak WFs of the paired channels are distributed from about 250 hPa to 775 hPa, and the number of paired channels decreases significantly when adding the criterion based on the WF broadness in this study.
In step 2, a set of sensitivity experiments is carried out by changing | Δ p C S | on different pressure levels and we obtain a set of height-dependent thresholds for the cloud-sensitive-level differences between the LWIR and SWIR channels to be paired. In the meantime, 80 ECMWF profiles [32] are used as input for RTTOV model simulations of brightness temperature to calculate the RMSEs between paired LWIR and SWIR channels. Both the cloud-sensitive-level difference and brightness temperature RMSE between the final chosen LWIR/SWIR channels should be minimum. Table 2 shows the differences of WF peak, WF broadness, cloud-sensitive level and the RMSE between the finally paired LWIR and SWIR HIRAS channels during these two pairing steps discussed above.
A total of 26 paired HIRAS LWIR/SWIR channels are obtained by the modified method and their characteristics can be seen in Table 3. And we also pair HIRAS LWIR/SWIR channels by the original method, obtaining 24 pairs with their characteristics in Table 4. Among the results obtained by these two methods, 12 pairs are exactly the same. The modified method provides channels with peak WFs located at the middle troposphere (pair 8 and 9) which original method cannot.
Figure 6 shows the distributions of their peak WFs with pressure respectively, which can provide information of the atmosphere about around 240 hPa to 760 hPa. The WF broadness comparison between the paired LWIR and SWIR channels derived from these two methods are shown in Figure 7. For pairs on the same pressure level, the broadness difference between LWIR and SWIR channels obtained by modified method is generally smaller. In other words, LWIR/SWIR channels obtained by modified method can provide more detailed atmosphere information.

4.2. HIRAS CESIs

Before calculating the regression coefficients for CESIs, we need to select clear-sky HIRAS observations by collocating with MYD035 cloud mask products. The HIRAS clear-sky observation is determined as: all collocated MYD035 cloud mask products (5-km resolution) within the HIRAS observation FOV (16-km resolution at nadir) are clear. Figure 8a,b show the local time distributions of MODIS (tracks indicated by dashed curves) and HIRAS (tracks indicated by solid curves) observations from ascending nodes during 0000–2400 UTC 1 January 2019. Their local equator-crossing times are very close and the distribution of time difference between them from ascending and descending nodes on 1 January 2019 are shown in Figure 8c,d, respectively.
While ensuring enough data volume, the clear-sky selection will be more accurate if the observation time difference is smaller. Based on the discussion above, the collocation criterion on the observation time difference between HIRAS and MODIS data used in this study is less than 16 min. In this study, MYD035 cloud mask products from 1–20 January 2019 are used for collocating HIRAS clear-sky observations. Figure 9a takes an example of a regional distribution of the MYD035 cloud mask products from descending nodes on 1 January 2019, with bright green dots showing matched HIRAS clear-sky observations. Take pair 6 in Table 3 (LWIR 98, SWIR 1783) as an example, these HIRAS clear-sky observations are picked out and plotted as a scatter plot in Figure 9b. It can be seen that there is a good linear relationship between the clear-sky brightness temperature of LWIR and SWIR channels. Scatter shown in Figure 9c are HIRAS cloudy observations picked out from this area. As can be seen from this figure, in the presence of clouds, there is a more pronounced drop in brightness temperature of SWIR channel than LWIR channel which is consistent with the statement in Section 3.3.
Considering the scan-dependent feature of HIRAS observed brightness temperature, all 29 HIRAS FORs of each pair are used for regression respectively. The CESI regression coefficients of α and β for 26 pairs (modified method) and 24 pairs (original method) are shown in Figure 10 and Figure 11, respectively, and these coefficients vary regularly with the scan angle. When all pairs of CESIs from both modified and original methods are obtained, they can be used for cloud detection comparison.
To illustrate the effectiveness of this improved cloud detection scheme proposed for HIRAS, cloud products from the AIRS onboard the Aqua satellite are used in this study. In the past few decades, the AIRS science team has made great achievements in cloud product retrievals from AIRS and the advanced microwave sounding unit instrument suite [33]. Ice cloud optical depth and cloud top pressure from the AIRS (version 6) products are used in this study because CESIs can well reflect the ice water content in the atmosphere and AIRS cloud products are particularly well suited to revealing valuable information about cloud altitudes [34].
Due to HIRAS observations in higher latitudes are affected by noise, the linear regression relationship between LWIR and SWIR brightness temperature is not very ideal. This study is conducted only over the tropics (30° S–30° N). Figure 12 shows the tropical distributions of AIRS ice cloud optical depth (Figure 12a) and cloud-top pressure (Figure 12b) during the daytime on February 2, 2019. We can see that clouds with high cloud-top pressures (above 400 hPa) are mainly concentrated in 20° S-20° N and ice water are mainly found in these high-altitude clouds. Comparing the distributions of two CESIs by modified method (MM) shown in Figure 12c,d, CESIMM-26 can capture features in the distributions of AIRS ice cloud optical depth better than CESIMM-6. That’s because different CESIs located at different peak WF heights and cloud-sensitive levels provide cloud information within different layers of the atmosphere. The AIRS ice cloud optical depth represents the clouds contained in all layers of atmosphere and CESIMM-26 reflects the clouds above 765 hPa. But CESIMM-6 can only provide information of clouds above 321 hPa. This explains why Figure 12a,d are more consistent.
Next, we take a few more CESIs as examples to compare the cloud detection effects of the modified and original method (OM). CESIMM-21 and CESIOM-19: Though the peak WFs of them are both located around 628 hPa, the WF broadness differences of them are quite different (Figure 7). We can explore the influence of WF broadness difference on cloud detection from comparison of these two CESIs. CESIMM-24 and CESIOM-22: Their WF broadness differences are very close, and the only difference between them is that their paired SWIR channels are different (SWIR 2035 and SWIR 2031). For one LWIR channel, we can study the influence of different paired SWIR channels on cloud detection effects. Figure 13 shows the distributions of 4 CESIs above during the daytime on 2 February 2019. Obviously, CESIMMs obtained by modified method (Figure 13c,d) can capture more features of cloud than original method (Figure 13a,b). The distributions of CESIMM are more consistent with AIRS ice cloud optical depth and can give a more detailed description of the cloud distributions.
To show the distributions of CESIMM and CESIOM in more detail, we select data from two smaller areas (black boxes in Figure 12a,b) to compare by drawing them into scatter plots. Figure 14 shows the scatter plots of CESIMM versus CESIOM from the AIRS/HIRAS overlapping area over the two domains mentioned above with colors indicating the AIRS ice cloud optical depth. It can be seen that when there are few ice particles in the atmosphere, values of all these CESIs are close zero. In other words, CESIs are insensitive to clear sky. In general, as the amount of ice particle increases, so do values of CESIs. However, there are quite a few ice clouds that could be captured by CESIMMs but show no reaction in CESIOMs (circles scattering to the right). The same phenomenon exists in Figure 15. Both CESIMM and CESIOM show no reaction to the clouds near the surface (dark red circles). But when it comes to the upper atmosphere, CESIMMs can capture more features of ice clouds with high cloud-top pressure than CESIOMs (blue circles scattering to the right). According to the discussions above, the improved cloud detection scheme developed in this study can capture features in the distributions of AIRS-retrieved ice cloud optical depth and cloud-top pressure better than the original method.

5. Discussion

The improved cloud detection scheme developed in this study can capture features of ice clouds better than the original method. By adding more restrictions, the sensitivities to clouds of paired LWIR/SWIR channels are more similar and this method has more advantages in the identification of cloud height. However, this cloud detection scheme still has some shortcomings: (1) For now it can only be used for cloud detection over tropics when applied to HIRAS; (2) The determination of clear-sky thresholds for CESIs is a difficult problem. Once the CESI thresholds are determined, the modified CESIs could thus be used for cloud detection when assimilating HIRAS data for NWP. Our future research will focus on the determination of CESI thresholds for selecting clear-sky HIRAS data, bias estimates of HIRAS observations under clear-sky conditions, and assimilation of HIRAS data in the Global/Regional Assimilation and Prediction Enhanced System model.

6. Conclusions

The FY-3D HIRAS is the first Chinese operational hyperspectral infrared sounder providing high spectral resolution data for weather applications. It represents a major step forward in Chinese infrared sounding development. An improved cloud detection scheme using brightness temperature observations from paired HIRAS LWIR and SWIR channels is developed here. When pairing LWIR and SWIR channels, the concept of WF broadness is proposed as an additional criterion to find LWIR and SWIR channels with similar WF peak altitudes. A set of height-dependent thresholds for differences of the cloud-sensitive levels between LWIR and SWIR channels is obtained and used for pairing. Before calculating the CESIs of the paired LWIR/SWIR channels, HIRAS clear-sky brightness temperature observations derived by collocating with MODIS cloud mask products are used for developing the linear regression models for all selected pairs. This approach avoids the simulation errors caused by profiles and ensures a sufficient amount of data for regression. The CESIs for the 26 (modified method) and 24 (original method) pairs are applied to HIRAS observations and compared with AIRS cloud products. The variations in CESIs in the lower atmosphere are consistent with ice water contents at all altitudes, and CESIs with higher peak WFs are sensitive to ice clouds in the upper troposphere. In the future, our research will focus on the determination of CESI thresholds for selecting clear-sky HIRAS data for NWP assimilation.

Author Contributions

Conceptualization, X.Z. and X.X.; methodology, X.Z.; software, X.X.; validation, X.X. and X.Z.; formal analysis, X.X.; investigation, X.X.; resources, X.Z.; data curation, X.X.; writing—original draft preparation, X.X.; writing—review and editing, X.X.; visualization, X.X.; supervision, X.Z.; project administration, X.Z.; funding acquisition, X.Z. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the National Key R&D Program of China (Grant 2018YFC1507004), special fund for the development of China Meteorological Administration numerical weather prediction systems (GRAPES), and the Postgraduate Research and Practice Innovation Program of Jiangsu Province (Grant KYCX20_0916).

Acknowledgments

We would like to acknowledge the suggestions given by reviewers and editor.

Conflicts of Interest

We declared that we have no conflicts of interest to this work. We declare that we do not have any commercial or associative interest that represents a conflict of interest in connection with the work submitted.

References

  1. Aumann, H.H.; Gregorich, D.T.; Gaiser, S.L.; Chahine, M.T. Application of Atmospheric Infrared Sounder (AIRS) data to climate research. Proc. SPIE Int. Soc. Opt. Eng. 2004, 5570, 202–208. [Google Scholar]
  2. Blumstein, D.; Tournier, B.; Cayla, F.R.; Phulpin, T.; Fjortoft, R.; Buil, C.; Ponce, G. In-flight performance of the infrared atmospheric sounding interferometer (IASI) on METOP-A. Proc. SPIE 2007, 6684, 66840. [Google Scholar] [CrossRef]
  3. Pougatchev, N.; August, T.; Calbet, X.; Hultberg, T.; Oduleye, O.; Schlüssel, P.; Stiller, B.; Germain, K.S.; Bingham, G. IASI temperature and water vapor retrievals—Error assessment and validation. Atmos. Chem. Phys. Discuss. 2009, 9, 6453–6458. [Google Scholar] [CrossRef] [Green Version]
  4. Aumann, H.; Chahine, M.; Gautier, C.; Goldberg, M.; Kalnay, E.; McMillin, L.; Revercomb, H.; Rosenkranz, P.; Smith, W.; Staelin, D.; et al. AIRS/AMSU/HSB on the aqua mission: Design, science objectives, data products, and processing systems. IEEE Trans. Geosci. Remote Sens. 2003, 41, 253–264. [Google Scholar] [CrossRef] [Green Version]
  5. Klaes, K.D.; Cohen, M.; Buhler, Y.; Schlüssel, P.; Munro, R.; Luntama, J.-P.; Von Engeln, A.; Clérigh, E.Ó.; Bonekamp, H.; Ackermann, J.; et al. An Introduction to the EUMETSAT Polar system. Bull. Am. Meteorol. Soc. 2007, 88, 1085–1096. [Google Scholar] [CrossRef]
  6. Hilton, F.; Armante, R.; August, T.; Barnet, C.; Bouchard, A.; Camy-Peyret, C.; Capelle, V.; Clarisse, L.; Clerbaux, C.; Coheur, P.-F.; et al. Hyperspectral Earth Observation from IASI: Five Years of Accomplishments. Bull. Am. Meteorol. Soc. 2012, 93, 347–370. [Google Scholar] [CrossRef]
  7. Han, Y.; Revercomb, H.E.; Cromp, M.; Gu, D.; Johnson, D.B.; Mooney, D.; Scott, D.; Strow, L.L.; Bingham, G.E.; Borg, L.; et al. Suomi NPP CrIS measurements, sensor data record algorithm, calibration and validation activities, and record data quality. J. Geophys. Res. Atmos. 2013, 118, 12734–12738. [Google Scholar] [CrossRef]
  8. Carminati, F.; Xiao, X.; Lu, Q.; Atkinson, N.; Hocking, J. Assessment of the Hyperspectral Infrared Atmospheric Sounder (HIRAS). Remote Sens. 2019, 11, 2950. [Google Scholar] [CrossRef] [Green Version]
  9. Qi, C.; Wu, C.; Hu, X.; Xu, H.; Lee, L.; Zhou, F.; Gu, M.; Yang, T.; Shao, C.; Yang, Z.; et al. High Spectral Infrared Atmospheric Sounder (HIRAS): System Overview and On-Orbit Performance Assessment. IEEE Trans. Geosci. Remote Sens. 2020, 58, 4335–4352. [Google Scholar] [CrossRef]
  10. Wu, C.; Qi, C.; Hu, X.; Gu, M.; Yang, T.; Xu, H.; Lee, L.; Yang, Z.; Zhang, P. FY-3D HIRAS Radiometric Calibration and Accuracy Assessment. IEEE Trans. Geosci. Remote Sens. 2020, 58, 3965–3976. [Google Scholar] [CrossRef]
  11. Hilton, F.; Atkinson, N.C.; English, S.J.; Eyre, J.R. Assimilation of IASI at the Met Office and assessment of its impact through observing system experiments. Q. J. R. Meteorol. Soc. 2009, 135, 495–505. [Google Scholar] [CrossRef]
  12. McNally, A.P.; Watts, P.D.; Smith, J.A.; Engelen, R.; Kelly, G.A.; Thépaut, J.N.; Matricardi, M. The assimilation of AIRS radiance data at ECMWF. Q. J. R. Meteorol. Soc. 2006, 132, 935–957. [Google Scholar] [CrossRef] [Green Version]
  13. Li, Z.; Shen, H.; Wei, Y.; Cheng, Q.; Yuan, Q. Cloud detection by fusing multi-scale convolutional features. ISPRS Ann. Photogramm. Remote Sens. Spat. Inf. Sci. 2018, IV3, 149–152. [Google Scholar] [CrossRef] [Green Version]
  14. Wu, X.; Pan, B. Utilizing Multilevel Features for Cloud Detection on Satellite Imagery. Remote Sens. 2018, 10, 1853. [Google Scholar] [CrossRef] [Green Version]
  15. Mahajan, S.; Fataniya, B. Cloud detection methodologies: Variants and development—A review. Complex Intell. Syst. 2020, 6, 251–261. [Google Scholar] [CrossRef] [Green Version]
  16. Amato, U.; Lavanant, L.; Liuzzi, G.M.; Masiello, G.; Serio, C.; Stuhlmann, R.; Tjemkes, S. Cloud mask via cumulative discriminant analysis applied to satellite infrared observations: Scientific basis and initial evaluation. Atmos. Meas. Tech. 2014, 7, 3355–3372. [Google Scholar] [CrossRef]
  17. Saunders, R.W.; Kriebel, K.T. An improved method for detecting clear sky and cloudy radiances from AVHRR data. Int. J. Remote Sens. 1988, 9, 123–150. [Google Scholar] [CrossRef]
  18. Frey, R.A.; Ackerman, S.A.; Liu, Y.; Strabala, K.I.; Zhang, H.; Key, J.; Wang, X. Cloud Detection with MODIS. Part I: Improvements in the MODIS Cloud Mask for Collection 5. J. Atmos. Ocean. Technol. 2008, 25, 1057–1072. [Google Scholar] [CrossRef]
  19. Kopp, T.J.; Thomas, W.; Heidinger, A.K.; Botambekov, D.; Frey, R.A.; Hutchison, K.D.; Iisager, B.D.; Brueske, K.; Reed, B. The VIIRS Cloud Mask: Progress in the first year of S-NPP toward a common cloud detection scheme. J. Geophys. Res. Atmos. 2014, 119, 2441–2456. [Google Scholar] [CrossRef]
  20. McNally, A.P.; Watts, P.D. A cloud detection algorithm for high-spectral-resolution infrared sounders. Q. J. R. Meteorol. Soc. 2003, 129, 3411–3423. [Google Scholar] [CrossRef]
  21. Menzel, W.P.; Schmit, T.J.; Zhang, P.; Li, J. Satellite-Based Atmospheric Infrared Sounder Development and Applications. Bull. Am. Meteorol. Soc. 2018, 99, 583–603. [Google Scholar] [CrossRef]
  22. Smith, W.L.; Platt, C.M.R. Comparison of Satellite-Deduced Cloud Heights with Indications from Radiosonde and Ground-Based Laser Measurements. J. Appl. Meteorol. 1978, 17, 1796–1802. [Google Scholar] [CrossRef] [Green Version]
  23. Wielicki, B.A.; Coakley, J.A. Cloud Retrieval Using Infrared Sounder Data: Error Analysis. J. Appl. Meteorol. 1981, 20, 157–169. [Google Scholar] [CrossRef] [Green Version]
  24. Wylie, D.P.; Menzel, W.P.; Woolf, H.M.; Strabala, K.I. Four Years of Global Cirrus Cloud Statistics Using HIRS. J. Clim. 1995, 7, 1972–1986. [Google Scholar] [CrossRef] [Green Version]
  25. Lin, L.; Qin, Z.; Weng, F. Combining CrIS double CO2 bands for detecting clouds located in different layers of the atmosphere. J. Geophys. Res. Atmos. 2017, 122, 1811–1827. [Google Scholar] [CrossRef]
  26. Aumann, H.H.; Chen, X.; Fishbein, E.; Geer, A.; Havemann, S.; Huang, X.; Liu, X.; Liuzzi, G.; DeSouza-Machado, S.; Manning, E.M.; et al. Evaluation of Radiative Transfer Models With Clouds. J. Geophys. Res. Atmos. 2018, 123, 6142–6157. [Google Scholar] [CrossRef]
  27. Saunders, R.; Matricardi, M.; Brunel, P. An improved fast radiative transfer model for assimilation of satellite radiance observations. Q. J. R. Meteorol. Soc. 1999, 125, 1407–1425. [Google Scholar] [CrossRef]
  28. Li, X.; Zou, X. Bias characterization of CrIS radiances at 399 selected channels with respect to NWP model simulations. Atmos. Res. 2017, 196, 164–181. [Google Scholar] [CrossRef]
  29. Carrier, M.J.; Zou, X.; Lapenta, W.M. Identifying Cloud-Uncontaminated AIRS Spectra from Cloudy FOV Based on Cloud-Top Pressure and Weighting Functions. Mon. Weather Rev. 2007, 135, 2278–2294. [Google Scholar] [CrossRef]
  30. Xu, L.; Fang, S.; Niu, R.; Li, J. Cloud detection based on decision tree over tibetan plateau with modis data. ISPRS Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 2012, 1, 535–538. [Google Scholar] [CrossRef] [Green Version]
  31. Han, Y.; Qin, Z.; Weng, F. Cloud and precipitation features of Super Typhoon Neoguri revealed from dual oxygen absorption band sounding instruments on board FengYun-3C satellite. Geophys. Res. Lett. 2015, 42, 916–924. [Google Scholar] [CrossRef]
  32. Saunders, R.; Hocking, J.; Turner, E.; Rayer, P.; Rundle, D.; Brunel, P.; Vidot, J.; Roquet, P.; Matricardi, M.; Geer, A.; et al. An update on the RTTOV fast radiative transfer model (currently at version 12). Geosci. Model Dev. 2018, 11, 2717–2737. [Google Scholar] [CrossRef] [Green Version]
  33. Kahn, B.H.; Hanii, T.; Stephens, G.L.; Qing, Y.; Julien, D.; Gerald, M.; Manning, E.M.; Heymsfield, A.J. The Atmospheric Infrared Sounder. Atmos. Chem. Phys. 2018, 18, 10715–10739. [Google Scholar] [CrossRef] [Green Version]
  34. Weisz, E.; Li, J.; Menzel, W.P.; Heidinger, A.K.; Kahn, B.H.; Liu, C.-Y. Comparison of AIRS, MODIS, CloudSat and CALIPSO cloud top height retrievals. Geophys. Res. Lett. 2007, 34, 17811. [Google Scholar] [CrossRef]
Figure 1. Radiative Transfer for TOVS (RTTOV) simulated brightness temperatures (black curves with open circle) and WF peaks (gray curves) for HIRAS LWIR and SWIR channels at CO2 bands using the U.S. standard profile.
Figure 1. Radiative Transfer for TOVS (RTTOV) simulated brightness temperatures (black curves with open circle) and WF peaks (gray curves) for HIRAS LWIR and SWIR channels at CO2 bands using the U.S. standard profile.
Remotesensing 12 04171 g001
Figure 2. Spatial distributions of brightness temperature observations (unit: K) of (a) HIRAS SWIR channel 2007 (2382.5 cm−1, peak WF at 133 hPa), (b) CrIS SWIR channel 1238 (2382.5 cm−1, peak WF at 133 hPa), (c) HIRAS SWIR channel 1759 (2227.5 cm−1, peak WF at 535 hPa), and (d) CrIS SWIR channel 1176 (2227.5 cm−1, peak WF at 535 hPa) at descending nodes on 1 January 2019.
Figure 2. Spatial distributions of brightness temperature observations (unit: K) of (a) HIRAS SWIR channel 2007 (2382.5 cm−1, peak WF at 133 hPa), (b) CrIS SWIR channel 1238 (2382.5 cm−1, peak WF at 133 hPa), (c) HIRAS SWIR channel 1759 (2227.5 cm−1, peak WF at 535 hPa), and (d) CrIS SWIR channel 1176 (2227.5 cm−1, peak WF at 535 hPa) at descending nodes on 1 January 2019.
Remotesensing 12 04171 g002
Figure 3. WF profiles (curves) and WF peaks (symbols) of LWIR channel 143 (thick solid curve with open circles) and four SWIR channels satisfying the criterion: 2023 (thin solid curve with open circles), 1715 (dashed curve with triangles), and 1751 (dashed curve with squares).
Figure 3. WF profiles (curves) and WF peaks (symbols) of LWIR channel 143 (thick solid curve with open circles) and four SWIR channels satisfying the criterion: 2023 (thin solid curve with open circles), 1715 (dashed curve with triangles), and 1751 (dashed curve with squares).
Remotesensing 12 04171 g003
Figure 4. The WF broadness B L W = p 1 , L W p 2 , L W and B S W = p 1 , S W p 2 , S W for LWIR channel 143 (solid curve) and SWIR channel 1751 (dashed curve). The difference of WF broadness Δ B = | B L W B S W | between LWIR channel 143 and SWIR channels 2023, 1715 and 1751, whose WFs are shown in Figure 3, is 86, 227 and 255 hPa, respectively.
Figure 4. The WF broadness B L W = p 1 , L W p 2 , L W and B S W = p 1 , S W p 2 , S W for LWIR channel 143 (solid curve) and SWIR channel 1751 (dashed curve). The difference of WF broadness Δ B = | B L W B S W | between LWIR channel 143 and SWIR channels 2023, 1715 and 1751, whose WFs are shown in Figure 3, is 86, 227 and 255 hPa, respectively.
Remotesensing 12 04171 g004
Figure 5. Flow chart summarizing three major steps in the HIRAS LWIR/SWIR pairing procedure. Modifications added in this study to original method are indicated in bold.
Figure 5. Flow chart summarizing three major steps in the HIRAS LWIR/SWIR pairing procedure. Modifications added in this study to original method are indicated in bold.
Remotesensing 12 04171 g005
Figure 6. Vertical distribution of LWIR (circle and square) and SWIR (upward-pointing triangle and downward-pointing triangle) peak WF of 26 (modified method) and 24 (original method) pairs. Grey markers indicate the same paired LWIR and SWIR channels.
Figure 6. Vertical distribution of LWIR (circle and square) and SWIR (upward-pointing triangle and downward-pointing triangle) peak WF of 26 (modified method) and 24 (original method) pairs. Grey markers indicate the same paired LWIR and SWIR channels.
Remotesensing 12 04171 g006
Figure 7. Vertical distribution of WF broadness for (a) modified method’s 26 pairs and (b) original method’s 24 pairs.
Figure 7. Vertical distribution of WF broadness for (a) modified method’s 26 pairs and (b) original method’s 24 pairs.
Remotesensing 12 04171 g007
Figure 8. Local time distributions of (a) MODIS and (b) HIRAS from ascending nodes and time difference between them from (c) ascending nodes and (d) descending nodes during 0000–2400 UTC 1 January 2019. The black lines in (b) are the tracks of the MODIS (dashed) and HIRAS (solid).
Figure 8. Local time distributions of (a) MODIS and (b) HIRAS from ascending nodes and time difference between them from (c) ascending nodes and (d) descending nodes during 0000–2400 UTC 1 January 2019. The black lines in (b) are the tracks of the MODIS (dashed) and HIRAS (solid).
Remotesensing 12 04171 g008aRemotesensing 12 04171 g008b
Figure 9. Distribution of the (a) MOD035 cloud mask at a 5-km resolution at descending node on January 1. Bright green dots are collocated HIRAS clear-sky observations. Scatter plots of HIRAS brightness temperature of LWIR 98 and SWIR 1783 are from (b) clear-sky and (c) cloudy observations in this area. Black lines indicate the linear regression between LWIR and SWIR brightness temperature under clear-sky condition (solid) and cloudy condition (dashed).
Figure 9. Distribution of the (a) MOD035 cloud mask at a 5-km resolution at descending node on January 1. Bright green dots are collocated HIRAS clear-sky observations. Scatter plots of HIRAS brightness temperature of LWIR 98 and SWIR 1783 are from (b) clear-sky and (c) cloudy observations in this area. Black lines indicate the linear regression between LWIR and SWIR brightness temperature under clear-sky condition (solid) and cloudy condition (dashed).
Remotesensing 12 04171 g009
Figure 10. Regression coefficients (a,b) α and (c,d) β of the linear regression model ( T b , r e g r e s s i o n S W I R = α T o b s L W I R + β ) for 29 FORs of the 26 pairs with modified method.
Figure 10. Regression coefficients (a,b) α and (c,d) β of the linear regression model ( T b , r e g r e s s i o n S W I R = α T o b s L W I R + β ) for 29 FORs of the 26 pairs with modified method.
Remotesensing 12 04171 g010
Figure 11. Regression coefficients (a,b) α and (c,d) β of the linear regression model ( T b , r e g r e s s i o n S W I R = α T o b s L W I R + β ) for 29 FORs of the 24 pairs with original method.
Figure 11. Regression coefficients (a,b) α and (c,d) β of the linear regression model ( T b , r e g r e s s i o n S W I R = α T o b s L W I R + β ) for 29 FORs of the 24 pairs with original method.
Remotesensing 12 04171 g011aRemotesensing 12 04171 g011b
Figure 12. Spatial distributions of (a) AIRS ice cloud optical depth (color interval for ln(optical depth) is 0.3), (b) AIRS cloud-top pressure, (c) CESIMM-6 (~321 hPa) and (d) CESIMM-26 (~765 hPa) at ascending nodes on 2 February 2019. Color interval for ln(CESI) is 0.126.
Figure 12. Spatial distributions of (a) AIRS ice cloud optical depth (color interval for ln(optical depth) is 0.3), (b) AIRS cloud-top pressure, (c) CESIMM-6 (~321 hPa) and (d) CESIMM-26 (~765 hPa) at ascending nodes on 2 February 2019. Color interval for ln(CESI) is 0.126.
Remotesensing 12 04171 g012aRemotesensing 12 04171 g012b
Figure 13. Spatial distributions of (a) CESIOM 19 (~628 hPa), (b) CESIOM 22 (~718 hPa), (c) CESIMM 21 (~628 hPa) and (d) CESIMM 24 (~718 hPa) at ascending nodes on 2 February 2019. Color interval for ln(CESI) is 0.126.
Figure 13. Spatial distributions of (a) CESIOM 19 (~628 hPa), (b) CESIOM 22 (~718 hPa), (c) CESIMM 21 (~628 hPa) and (d) CESIMM 24 (~718 hPa) at ascending nodes on 2 February 2019. Color interval for ln(CESI) is 0.126.
Remotesensing 12 04171 g013
Figure 14. Scatter plots of (a,b) CESIMM-21 versus CESIOM-19 (~628 hPa) and (c,d) CESIMM-24 versus CESIOM-22 (~718 hPa) from the AIRS/HIRAS overlapping area at ascending nodes over domain 1 (left panels) and domain 2 (right panels) on 2 February 2019. The colors indicate AIRS ice cloud optical depth (Figure 12a).
Figure 14. Scatter plots of (a,b) CESIMM-21 versus CESIOM-19 (~628 hPa) and (c,d) CESIMM-24 versus CESIOM-22 (~718 hPa) from the AIRS/HIRAS overlapping area at ascending nodes over domain 1 (left panels) and domain 2 (right panels) on 2 February 2019. The colors indicate AIRS ice cloud optical depth (Figure 12a).
Remotesensing 12 04171 g014
Figure 15. Scatter plots of (a,b) CESIMM-21 versus CESIOM-19 (~628 hPa) and (c,d) CESIMM-24 versus CESIOM-22 (~718 hPa) from the AIRS/HIRAS overlapping area at ascending nodes over domain 1 (left panels) and domain 2 (right panels) on 2 February 2019. The colors indicate AIRS ice cloud top pressure (Figure 12b).
Figure 15. Scatter plots of (a,b) CESIMM-21 versus CESIOM-19 (~628 hPa) and (c,d) CESIMM-24 versus CESIOM-22 (~718 hPa) from the AIRS/HIRAS overlapping area at ascending nodes over domain 1 (left panels) and domain 2 (right panels) on 2 February 2019. The colors indicate AIRS ice cloud top pressure (Figure 12b).
Remotesensing 12 04171 g015
Table 1. Number of LWIR and SWIR pairs selected from channels at CO2 bands of HIRAS within different pressure layers by original method (second column) and when adding the criterion based on the WF broadness in this study (third column) after step 1.
Table 1. Number of LWIR and SWIR pairs selected from channels at CO2 bands of HIRAS within different pressure layers by original method (second column) and when adding the criterion based on the WF broadness in this study (third column) after step 1.
Pressure (hPa)NumberNumber
250–2752522
275–3004833
300–3253314
325–350186
350–3754421
375–400137
400–425109
425–45066
450–475126
475–500157
500–525127
525–5503214
550–5752017
575–60088
600–62566
625–65096
650–67564
700–72522
725–75022
750–77511
Table 2. Pair number, differences of the WF peak ( Δ p W F p e a k ), WF broadness ( Δ P b r o a d n e s s ), cloud-sensitive level ( Δ p C S ) and the RMSE between the finally paired LWIR and SWIR HIRAS channels during two pairing steps.
Table 2. Pair number, differences of the WF peak ( Δ p W F p e a k ), WF broadness ( Δ P b r o a d n e s s ), cloud-sensitive level ( Δ p C S ) and the RMSE between the finally paired LWIR and SWIR HIRAS channels during two pairing steps.
PairStep 1Step 2
Δ p W F p e a k   ( hPa ) Δ P b r o a d n e s s   ( hPa ) Δ p C S   ( hPa ) RMSE (K)
112282311.5
2250169.8
3380188.6
42774149.2
53851108.6
64231199.4
7032611.3
8451014613.6
948783612.8
103519515.8
111744115.3
120113315.0
13183615.1
1437343213.7
150107312.8
16191861612.1
174017819.0
182016278.3
1940134411.0
20057914.3
212257114.1
2244134614.7
23018088.9
240392612.5
2523661111.9
2647158355.8
Table 3. Channel number, wavenumber, WF peak and cloud-sensitive level ( p C S ) for the final paired 26 LWIR an SWIR HIRAS channels using the modified method in this study.
Table 3. Channel number, wavenumber, WF peak and cloud-sensitive level ( p C S ) for the final paired 26 LWIR an SWIR HIRAS channels using the modified method in this study.
PairLWIRSWIR
Channel NumberWave-Number (cm−1)Peak WF (hPa) p C S , L W   ( hPa ) Channel NumberWave-number (cm−1)Peak WF (hPa) p C S , S W   ( hPa )
188704.37525336817992252.500241391
2111718.75026640717992252.500241391
391706.25027940917992252.500241391
494708.12529344017912247.500266454
595708.75030744417912247.500266454
698710.62532150117832242.500279520
799711.25033650920152387.500336503
8105715.00036664017352212.500321716
9108716.87539965217592227.500351718
10146740.62541570520192390.000450710
11107716.25043370920192390.000450710
12109717.50045067720192390.000450710
13127728.75046870420192390.000450710
14131731.25048767820192390.000450710
15124726.87550682717552225.000506830
16136734.37552581417552225.000506830
17142738.12554581417152200.000585815
18140736.87556582217152200.000585815
19125727.50058584617512222.500545850
20123726.25060686720272395.000606876
21120724.37562887520272395.000606876
22151743.75065087020272395.000606876
23168754.37567289720312397.500672889
24154745.62571892320352400.000718897
25150743.12574190820352400.000718897
26167753.75076593220352400.000718897
Table 4. Same as Table 3 except for the 24 pairs using the original method.
Table 4. Same as Table 3 except for the 24 pairs using the original method.
PairLWIRSWIR
Channel NumberWave-Number (cm−1)Peak WF (hPa) p C S , L W   ( hPa ) Channel NumberWave-Number (cm−1)Peak WF (hPa) p C S , S W   ( hPa )
188704.37525336817992252.500241391
2111718.75026640717992252.500241391
391706.25027940917992252.500241391
494708.12529344017912247.500266454
595708.75030744417912247.500266454
698710.62532150117832242.500279520
799711.25033650920152387.500336503
8146740.62541570520192390.000450710
9107716.25043370917322205.000415758
10109717.50045067720192390.000450710
11116721.87546874120192390.000450710
12139736.25048779717192202.500487840
13124726.87550682717192202.500487840
14133732.50052581917192202.500487840
15145740.00054577817152200.000585815
16143738.75056582617152200.000585815
17125727.50058584617152200.000585815
18149742.50060686517152200.000585815
19153745.00062882517152200.000585815
20151743.75065087020272395.000606876
21168754.37567289720312397.500672889
22154745.62571892320312397.500672889
23158748.12574192520352400.000718897
24167753.75076593220352400.000718897
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Xia, X.; Zou, X. Development of CO2 Band-Based Cloud Emission and Scattering Indices and Their Applications to FY-3D Hyperspectral Infrared Atmospheric Sounder. Remote Sens. 2020, 12, 4171. https://0-doi-org.brum.beds.ac.uk/10.3390/rs12244171

AMA Style

Xia X, Zou X. Development of CO2 Band-Based Cloud Emission and Scattering Indices and Their Applications to FY-3D Hyperspectral Infrared Atmospheric Sounder. Remote Sensing. 2020; 12(24):4171. https://0-doi-org.brum.beds.ac.uk/10.3390/rs12244171

Chicago/Turabian Style

Xia, Xinlu, and Xiaolei Zou. 2020. "Development of CO2 Band-Based Cloud Emission and Scattering Indices and Their Applications to FY-3D Hyperspectral Infrared Atmospheric Sounder" Remote Sensing 12, no. 24: 4171. https://0-doi-org.brum.beds.ac.uk/10.3390/rs12244171

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop