Search

Advanced Microwave Scanning Radiometer-2 (AMSR2) is a successor of AMSR for Earth-Observation System (AMSR-E), while the third generation of AMSR (AMSR3) will be launched in the near future. The AMSR2 soil moisture (SM) product is also an important component of the SM operational products system (SMOPS) datasets that are operationally produced by National Oceanic and Atmospheric Administration (NOAA). The refinement of the NOAA AMSR2 SM data product can not only benefit the past AMSR-E and the upcoming AMSR3 but also improve the SMOPS data quality. In this second article of the two-part series, the extreme gradient boosting (XGB) model was trained using the AMSR2 6.925-, 10.65-, 18.7-, and 36.5-GHz brightness temperature (Tb) measurements in dual polarizations, ancillary maps, and the vegetation index datasets and in turn used to predict the daily global AMSR2 SM retrievals from 2012 to 2021. Validation results show that the refined AMSR2 SM retrievals (AMSRr) show an overwhelming advantage in data accuracy over the currently operational AMSR2 (AMSRc) product. Compared to the AMSRc, the developed AMSRr presents a significant improvement on data availability. Results also indicate that the refined AMSR2 datasets are comparable with the latest version SM active passive (SMAP) SM product. Based on this study, higher quality AMSRr SM data product will be operationally produced in the NOAA and will eventually benefit the NOAA SMOPS blended SM product and its users.

Soil moisture (SM) is an important variable for hydrological, meteorological, and agricultural studies and applications. The SM Operational Products System (SMOPS) was developed by the National Oceanic and Atmospheric Administration (NOAA)-National Environmental Satellite, Data, and Information Service (NESDIS) to operationally provide an integrated satellite SM data product. The Advanced Microwave Scanning Radiometer 2 (AMSR2) SM retrieval is an important component of the currently operational SMOPS. This study is proposed to refine the AMSR2 data product using an optimal machine-learning model, and this first paper of the two-part series is to intercompare the six commonly used machine-learning models including multiple linear regression (MLR), regression tree (RRT), random forest (RFT), gradient boosting (GBR), extreme gradient boosting (XGB), and artificial neural network (ANN). Results indicate that all of the six approaches can preserve the reference data information beyond the training time period, which ensures they predict past and future satellite retrievals without a new training procedure. Relative to other models, the XGB method is more successful to respect to the reference data SM Active Passive (SMAP) and the in situ observations from the U.S. Department of Agriculture Soil Climate Analysis Network (SCAN). It has a good implication for the implementation of the XGB model to refine the AMSR2 SM retrievals in the second paper.

Abstract Passive microwave observations and derived products from the National Oceanic and Atmospheric Administration (NOAA) Polar-orbiting Operational Environmental Satellite (POES) Advanced Microwave Sounding Unit (AMSU) are now widely used in [...]

The primary signal used in all current passive microwave precipitation retrieval algorithms over land is the depression of the instantaneous brightness temperature (TB) caused by ice scattering. This study presents a new methodology to retrieve instantaneous precipitation rate over land by using TB temporal variation (ΔTB) at 19 GHz, which primarily reflects the surface emissivity variation due to the precipitation impact. As a proof‐of‐concept, we exploit observations from five polar‐orbiting satellites over the Southern Great Plains of the United States. Results show that ΔTBat 19 GHz correlate well with the instantaneous precipitation rate. Further analysis shows that ΔTBat 19 GHz is better correlated with the precipitation rate when multiple satellite observations are used due to the much shorter revisit time for a certain location. The retrieved instantaneous precipitation rate over Southern Great Plains from ΔTBat 19 GHz reasonably agrees with the surface radar observations, with the correlation, the root‐mean‐square error and the bias being 0.49, 2.39 mm/hr, and 6.54%, respectively. Future work seeks to combine the ice scattering signal at high frequencies and this surface emissivity variation signal at low frequencies to achieve an optimal retrieval performance. Current precipitation estimation technique via satellite passive microwave observations links the hydrometers in the air to the surface precipitation intensity. That is, the cold brightness temperature (TB) at high‐frequency channels (e.g., 85 GHz) indicates heavy precipitation. The TB observations from low‐frequency channels such as 19 GHz are largely discounted. This study presents a new idea to link the surface condition variation to the precipitation intensity, by using TB temporal variation (δTB) at 19 GHz from five polar‐orbiting satellites. Results show that δTBat 19 GHz correlate well with the precipitation rate. The estimated instantaneous precipitation rate over the Southern Great Plains of United States from δTBat 19 GHz reasonably agrees with the ground radar observations, with the correlation, the root‐mean‐square error, and the bias at 0.49, 2.39 mm/hr, and 6.54%, respectively. Observations from five polar‐orbiting satellites are used to derive TB temporal variation at 19 GHzTB temporal variation at 19 GHz correlates well with the instantaneous precipitation rateTB temporal variation at 19 GHz primarily reflects the surface emissivity variation due to the precipitation impact

Snowfall rate retrieval from spaceborne passive microwave (PMW) radiometers has gained momentum in recent years. PMW can be so utilized because of its ability to sense in‐cloud precipitation. A physically based, overland snowfall rate (SFR) algorithm has been developed using measurements from the Advanced Microwave Sounding Unit‐A/Microwave Humidity Sounder sensor pair and the Advanced Technology Microwave Sounder. Currently, these instruments are aboard five polar‐orbiting satellites, namely, NOAA‐18, NOAA‐19, Metop‐A, Metop‐B, and Suomi‐NPP. The SFR algorithm relies on a separate snowfall detection algorithm that is composed of a satellite‐based statistical model and a set of numerical weather prediction model‐based filters. There are four components in the SFR algorithm itself: cloud properties retrieval, computation of ice particle terminal velocity, ice water content adjustment, and the determination of snowfall rate. The retrieval of cloud properties is the foundation of the algorithm and is accomplished using a one‐dimensional variational (1DVAR) model. An existing model is adopted to derive ice particle terminal velocity. Since no measurement of cloud ice distribution is available when SFR is retrieved in near real time, such distribution is implicitly assumed by deriving an empirical function that adjusts retrieved SFR toward radar snowfall estimates. Finally, SFR is determined numerically from a complex integral. The algorithm has been validated against both radar and ground observations of snowfall events from the contiguous United States with satisfactory results. Currently, the SFR product is operationally generated at the National Oceanic and Atmospheric Administration and can be obtained from that organization. A 1DVAR‐based snowfall rate algorithm has been developed for passive microwave radiometers aboard polar‐orbiting satellitesThe algorithm has been validated against both radar snowfall estimates and gauge observations with satisfactory resultsThe snowfall rate product is being produced by the National Oceanic and Atmospheric Administration operationally at near real time

A prototype precipitation retrieval algorithm over land has been developed by utilizing 4 year National Mosaic and Multi‐Sensor Quantitative Precipitation Estimation and Special Sensor Microwave Imager/Sounder coincident data sets. One of the unique features of this algorithm is using the ancillary parameters (i.e., surface type, surface temperature, land elevation, and ice layer thickness) to stratify the single database into many smaller but more homogeneous databases, in which both the surface condition and precipitation vertical structure are similar. It is found that the probability of detection (POD) increases about 8% and 12% by using stratified databases for rainfall and snowfall detection, respectively. In addition, by considering the relative humidity at lower troposphere and the vertical velocity at 700 hPa in the precipitation detection process, the POD for snowfall detection is further increased by 20.4% from 56.0% to 76.4%. The better result is evident in both ends of the retrieved rain rate when the stratified databases are used, especially when the rain rate is greater than 30 mm/h. Similarly, the retrieved snowfall rate using stratified databases also outperforms that using single database. The correlation between retrieved and observed rain rates from stratified databases is 0.63, while it is 0.42 using the single database. The root‐mean‐square error is reduced by 50.3% from 2.07 to 0.98 by using stratified databases. The retrieved snow rates from stratified database are also better correlated with observations and possess smaller root‐mean‐square error. Additionally, the precipitation overestimation from the single database over the western United States is largely mitigated when the stratified databases are utilized. It is further demonstrated that over the majority of the stratified databases, the relationship between precipitation rate and brightness temperature is much closer to that from the corresponding category in the validation databases, rather than that from the single database. Therefore, overall superior performance using the stratified databases for both the precipitation detection and retrieval is achieved. Single database is stratified into smaller and more homogeneous databasesResults from stratified databases greatly outperform that from single databaseIncluding relative humidity and vertical velocity is beneficial for snow detection

The quality of humidity measurements from global operational radiosonde sensors in upper, middle, and lower troposphere for the period 2000–2011 were investigated using satellite observations from three microwave water vapor channels operating at 183.31±1, 183.31±3, and 183.31±7 GHz. The radiosonde data were partitioned based on sensor type into 19 classes. The satellite brightness temperatures (Tb) were simulated using radiosonde profiles and a radiative transfer model, then the radiosonde simulated Tb's were compared with the observed Tb's from the satellites. The surface affected Tb's were excluded from the comparison due to the lack of reliable surface emissivity data at the microwave frequencies. Daytime and nighttime data were examined separately to see the possible effect of daytime radiation bias on the sonde data. The error characteristics among different radiosondes vary significantly, which largely reflects the differences in sensor type. These differences are more evident in the mid‐upper troposphere than in the lower troposphere, mainly because some of the sensors stop responding to tropospheric humidity somewhere in the upper or even in the middle troposphere. In the upper troposphere, most sensors have a dry bias but Russian sensors and a few other sensors including GZZ2, VZB2, and RS80H have a wet bias. In middle troposphere, Russian sensors still have a wet bias but all other sensors have a dry bias. All sensors, including Russian sensors, have a dry bias in lower troposphere. The systematic and random errors generally decrease from upper to lower troposphere. Sensors from China, India, Russia, and the U.S. have a large random error in upper troposphere, which indicates that these sensors are not suitable for upper tropospheric studies as they fail to respond to humidity changes in the upper and even middle troposphere. Overall, Vaisala sensors perform better than other sensors throughout the troposphere exhibiting the smallest systematic and random errors. Because of the large differences between different radiosonde humidity sensors, it is important for long‐term trend studies to only use data measured using a single type of sensor at any given station. If multiple sensor types are used then it is necessary to consider the bias between sensor types and its possible dependence on humidity and temperature. Investigating the quality of operational radiosonde humidity profilesMost humidity sensors fail to measure mid‐upper tropospheric humidityDdifferent radiosonde sensors introduce temporal and spatial inhomogeneity

The Special Sensor Microwave/Imager (SSM/I), first placed into operation in July 1987, has been making measurements of earth-emitted radiation for over eight years. These data are used to estimate both atmospheric and surface hydrological parameters and to generate a time series of global monthly mean products averaged to a 1° lat × 1° long grid. Specifically, this includes monthly estimates of rainfall and its frequency, cloud liquid water and cloud frequency, water vapor, snow cover frequency, and sea ice frequency. This study uses seasonal mean values to demonstrate the spatial and temporal distributions of these hydrological variables. Examples of interannual variability such as the 1993 flooding in the Mississippi Valley and the 1992–93 snow cover changes over the United States are used to demonstrate the utility of these data for regional applications.

A Special Sensor Microwave/Imager (SSM/I) algorithm is developed to measure both cloud liquid water path (LWP) and cloud frequency (CF) over the oceans. For climate analysis, the LWP and CF parameters are computed on pentad and monthly timescales. Comparisons are made between cloud frequencies obtained from microwave and visible/infrared measurements. It is shown that the SSM/I CF correlates with International Satellite Cloud Climatology Program low- and middle-level cloudiness. Interannual variations of monthly LWP are found to be strongly correlated with El Niño and La Niña events. In general, positive LWP anomalies are associated with positive SST anomalies. However, positive LWP anomalies may also occur in regions of negative SST anomalies. This is probably due to an increase in warm top rain clouds, produced from low-level convergence. When pentads of outgoing longwave radiation data are compared to the LWP, they both show the detailed structure for atmospheric intraseasonal oscillations at 30–60-day periods. However, there are some interesting differences. Finally, as an important application, the monthly LWP is compared with simulations from a general circulation model. While the simulation captures the locations of observed maxima and minima, there is a large discrepancy between the model and measurement for the Northern Hemisphere in summer.

Rainfall estimates from spaceborne microwave radiometers form the foundation of global precipitation data sets. Since the beginning of the satellite microwave rainfall estimation era in the 1980s, the primary signature leveraged over land for these estimates has been the brightness temperature (TB) depression due to ice particle scattering. Contrary to this practice, time series analyses based on observations from two spaceborne radars and two spaceborne radiometers reveal a TB increase at H19 due to raindrop emission as the primary cloud particle signature over desert terrain. Low surface emissivity supports the use of liquid raindrop emission as the primary signature over desert surfaces. In these regions, the surface rain rate better correlates with the liquid raindrop emission signal than with the scattering induced by ice further aloft, suggesting a new potential for improving rainfall estimation over deserts by exploiting the liquid raindrop emission signature. Global rainfall mapping is a very challenging task since ground‐based observations from radars or rain‐gauges are limited or unavailable over many regions (e.g., desert). Satellite observations provide the unique opportunity to map the rainfall over these regions. Deserts cover about one third of the Earth's land surface, while the current satellite rainfall estimate techniques over deserts relate the ice particles aloft with surface rain rates, leading to large estimation uncertainties. This study for the first time reports the liquid raindrop signature over deserts from the satellite observations, which could represent a fundamental algorithm change over deserts when estimating rainfall from space. Time series analyses reveal a TB increase at H19 due to raindrop emission as the primary cloud signature over desertsLow surface emissivity supports the use of raindrop emission as the primary signature over desertsSurface rain rate better correlates with the raindrop emission signal than with the scattering induced by ice further aloft

We compared two fast radiative transfer models, Community Radiative Transfer Model (CRTM) and Radiative Transfer for TIROS Operational Vertical Sounder (RTTOV), with the LBL model Atmospheric Radiative Transfer Simulator (ARTS). We used the measurements from Advanced Technology Microwave Sounder (ATMS) and the Global Precipitation Measurement Microwave Imager (GMI) for evaluation of the radiative transfer models. The models in comparison with the observations and each other performed very well with a mean difference less than 0.5 K for the temperature sounding channels operating near the oxygen absorption band at 60 GHz. There was a difference of up to 1 K among the models as well as compared with the observations for humidity sounding channels operating around water vapor absorption line at 183 GHz. The mean difference between the simulations and observations was up to 6 K for surface sensitive channels. Water vapor and surface sensitive channels also showed to be more sensitive than the temperature sounding channels to the spectroscopy models used to calculate the absorption coefficients. There was a small difference, less than 0.1 K, between brightness temperatures calculated using traditional boxcar and actual Sensor or Spectral Response Functions, except for a difference of 0.25 K for ATMS Channel 6. Double difference technique showed about 1 K difference between water vapor channels from ATMS instruments onboard N20 and National Polar‐orbiting Partnership (NPP). However, comparison of a new version of ATMS/NPP observations recently generated using an enhanced calibration algorithm with ATMS/N20 observations showed that the differences between the two instruments are less than 0.5 K after improving the ATMS/NPP calibration. Radiative transfer models are used to simulate satellite observations from input atmospheric profiles and surface parameters. These models have a wide range of applications, including being used as forward model to assimilate satellite observations into numerical weather prediction models or for calibration and validation of satellite measurements. Radiative transfer models are subject to errors and uncertainties and therefore need to be evaluated. We compared three microwave radiative transfer models versus each other as well as satellite measurements. The results showed that the models were consistent with each other and with observations for microwave temperature sounding channels. However, the models were less consistent with each other and with the observations for the water vapor and window channels. The difference between the models and observations was affected by several factors such as spectral response functions, spectroscopy databases, and emissivity models. Microwave radiative transfer models generally agree with each other and observationsSpectroscopy models largely impact simulations for window and water vapor channelsSensor response function and emissivity show some important effect on calculated brightness temperatures