The extreme climate conditions and the sparse number of research stations in Antarctica limit the number of meteorological records over this area. Satellite radiometric measurements and ground-based GPS measurements can therefore improve the amount of available water vapour information. Combining the Zenith Total Delay (ZTD) time series from 6 Antarctic GPS stations and surface meteorological data, we have determined Precipitable Water Vapour (PW) variations with a 2-hour temporal resolution for a period of 5 years. Data from the Advanced Microwave Sounding Unit (AMSU-B) on board the NOAA-15 satellite cover most parts of Antarctica but with observations limited to merely few times a day. GPS and AMSU-B data sets are therefore complementary with respect to time and space. We present a cross validation between PW results from the two independent retrieval algorithms using one year data. Additionally, we compare the observed PW from AMSU-B and GPS with the National Centre for Environmental Prediction (NCEP) reanalysis. All three data sets are highly correlated. The mean differences between the three data sets are station dependent and vary from -1.7 to +1.2 mm. A large part of the bias may result from pressure uncertainties affecting the GPS PW estimates. GPS and AMSU-B as independent data sources are confirmed to be accurate methods for PW estimation for the dry Antarctic atmosphere. PW results from the NCEP analysis corresponds, in general, well to the PW observations at the investigated stations along the Antarctic coast. The results obtained at the station O’Higgins differ from those of the other station. O’Higgins is located at the Antarctic Peninsula and has a more humid environment than the coast of the main Antarctic continent, which may explain the peculiar behaviour of this station.
GPS analysis requires statistical information on temporal variations of precipitable water vapor content (PWC), since temporal variability of PWC is assumed in the analyses. Spatial scales of PWC variations also are necessary for incorporation of the analyzed GPS data into numerical weather prediction systems. The objective of the present study is to investigate temporal and spatial characteristics of the PWC variations in Japan. For this purpose, slant-path PWC was observed with ground-based microwave radiometers (MWRs) for the directions of GPS satellites in the eastern part of the Kanto Plain during several observation periods in 2000 and 2001. Three components, vertical PWC, horizontal gradient and higher-order inhomogeneity, were retrieved from the slant-path observation data. Deviations of vertical PWC from ten-day averages (hereafter referred to as vertical deviations) were then calculated in order to remove seasonal changes of this component. Simultaneous observation data at three MWR sites also were used for the rough estimation of horizontal scales of the three PWC components. The results show: 1) The vertical deviations marked were about 20 times as large amplitudes as the other two components, while the variations due to the gradient had even smaller amplitudes than the inhomogeneity; 2) The vertical deviations had large spectral power against periods around 5-6 days and 8-9 days, while the gradient was dominated by diurnal variations; and, 3) It was roughly estimated that the vertical deviations (gradient) had the horizontal scales of several hundred (several ten) kilometers. The horizontal scale of the PWC inhomogeneity was considered to be less than 10 km. The vertical deviations (the gradient) were considered to be closely related with large-scale meteorological disturbances (local-scale circulations) on the basis of the above temporal and spatial characteristics.
During the Tsukuba GPS (Global Positioning System) Dense Network Campaign Observations in the autumn of 2000, we performed 3 hourly upper air soundings using two types of radiosondes. One was the Meisei RS2-91, operational sonde of the Japan Meteorological Agency (JMA) made by Meisei Electric Co. Ltd. with an independent thin-film capacitive sensor, and the other was Vaisala RS80-15G with A-type Humicap humidity sensor (RS80-A hereafter). It was found that the Precipitable Water Vapor (PWV) measured by Vaisala RS80-A were clearly smaller (dry bias) by 3-4 mm (about 4-6%) than those by JMA RS2-91 and those analyzed from GPS observations. The dry bias error of Vaisala RS80-A was also confirmed with dual sonde balloon flights equipped with both Vaisala RS80-A and JMA RS2-91 instruments, and near simultaneous flights of two types of radiosondes. It was found that the dry bias error was reduced by the new tight protective cap for the humidity sensor, but substantial amount of dry bias error still remained. We further compared humidity measurements by the two radiosondes to a chilled-mirror dew-point hygrometer using a two-pressure type humidity calibration chamber. The experiment was conducted under room temperature conditions. It showed that relative humidity as measured by the Vaisala RS80- A was about 5 to 15% smaller than that of the dew-point hygrometer. The bias was larger in high humidity conditions. Although the dry bias of Vaisala RS80-A radiosondes has been pointed out in several other studies, the dry bias revealed in this study is much larger than previously reported, indicating that a problem still exists in RS80-A humidity measurements taken in moist air conditions such as Japan. The RS2-91 showed only small dry bias relative to the dew-point hygrometer (less than 4%). However, the Aerological Observatory of the JMA reported that they improved the RS2-91 humidity sensor since 1999 (Shibue et al. 2000). RS2-91 sondes manufactured before and in 1999 had a dry bias with the same order of magnitude as the Vaisala RS80-A. There have been several reports (e.g., Ohtani et al. 2000) that the PWV derived by GEONET (Japanese nationwide GPS array) and other campaign GPS observations agreed well with those by JMA or Vaisala radiosondes. The implication of the dry bias errors of the radiosonde to the previous GPS studies is that the past GPS analysis in Japan produced a drier atmosphere than actual conditions. It is recommended that the past GPS data, including the GEONET data, be reanalyzed using the most advanced GPS analysis methods.
In order to study small-scale water vapor variations over distances from a few km to 20 km, two campaign observations with a dense GPS network were carried out for 2.5 months in total at Tsukuba, Japan. For the observations 79 GPS antennas were installed at 75 sites within a 20 km by 20 km square area, at 1 to 3 km intervals. The PCV models provided by the US National Oceanic and Atmospheric Administration (NOAA) were applied to remove unmodeled phase center variation (PCV) specific to GPS antenna type. In addition, new PCV maps (MPS map) were constructed for all the antennas by stacking one-way postfit residuals over both campaign periods, to remove not only azimuth dependent PCV, but also the errors due to multipath effects. After MPS maps were introduced into the analysis, strong elevation dependence as well as azimuth dependence of postfit phase residuals, almost disappeared for all the antennas. In addition, the time variations in postfit residuals which were common to all the GPS sites, were subtracted to remove satellite orbits and/or clock errors. This led to the accurate estimate of slant path delay (SPD), which enabled the SPD to be applied to tomography analyses of water vapor (Seko et al. 2003). The horizontal scale of SPD was estimated using correlation distributions. As a result, the horizontal scale of the zenith total delay, the gradient component, and the postfit residual may be roughly considered as 644±120 km, 62±23 km, and 2-3 km, respectively. Improvement of the postfit residuals following the application of MPS maps also showed a positive impact on PWV estimation. Systematic biases of GPS derived PWV between different antenna types (Trimble and Ashtech) were reduced, resulting in a better agreement of GPS PWV, with RMS errors of 2.0 mm or less relative to PWV by rawinsonde or water vapor radiometer observations. The distribution of time-averaged PWV estimated at the 75 GPS sites showed a systematic pattern which has a negative correlation with the antenna height of each site.
The characteristics of post-fit residuals computed using three types of software have been investigated, and the behavior of multipath errors in the post-fit phase residuals evaluated. GPS data observed during the Tsukuba GPS dense net campaign were analyzed, and the post-fit phase residuals of the linear combination of the L1 and L2 GPS phase measurements carrier wave (LC), computed using the three software packages were studied. The post-fit phase residuals were stacked, and their mean value for each site was obtained, with a resolution of 1° × 1° for the azimuthal and elevation angles, respectively. The skymaps of the stacked post-fit phase residuals showed similar patterns among the three types of software. A random error index was introduced to evaluate random errors in the post-fit phase residuals, which showed twice as large errors using the point positioning strategy, as those obtained using the double difference strategy. The oscillation patterns shown in the stacking map were similar to those simulated using the Elosegui's multipath model (1995). This suggests that the pattern arose from multipath errors. When the stacking maps were introduced into the post-fit phase residuals, the impact of the resulting post-fit phase residuals was significant at sites that showed large multipath errors. The point positioning strategy still contained random errors that were twice as large as the double difference strategy errors. Multipath errors induced biases in the zenith tropospheric delay, which resulted in larger biases of the absolute slant tropospheric delay reconstruction at lower elevation angles. This suggests the importance of introducing stacking maps in GPS analysis.
Observations at low elevation angles are gathered and used for high accuracy GPS analyses. Introducing tropospheric gradient parameters in the estimation process allows it to take into account of azimuthal variations of tropospheric refraction. This measure significantly improves the repeatability of station coordinates. Most of the mean gradients for stations in the northern and the southern hemisphere show a similar behavior. By combining several years of data from a global solution results a consistent pattern.
Atmospheric delays in GPS processing are estimated by fitting parameters of atmospheric models to the observed delays. GPS-related error sources include multipath effects and phase center variations, as well as uncertainties in atmospheric models. An atmospheric model must accurately capture the delay distribution if errors are to be minimized. Large GPS position errors occurred on 7 March 1997 on the Izu Peninsula and at Hatsu-shima, about 100 km southwest of Tokyo, concomitant with a mountain lee wave. The coinciding of the large position errors and the mountain lee wave suggests that small-scale fluctuations in water vapor and air density associated with lee waves could cause large position errors. This study used a high-resolution non-hydrostatic model to simulate the mountain lee wave. Slant delays for the GPS satellites were calculated from the reproduced water vapor and air density fields using a ray-tracing method. The atmospheric delays are obtained by fitting the atmospheric models to the reproduced slant delays, instead of the actual delay data. The atmospheric models were evaluated by determining the difference in position error between the reproduced delays and the model-fitted delays. The mountain lee wave case was used to evaluate three different atmospheric models. The first, the “constant model”, has only zenith delay as an unknown parameter. Large position errors occurred when the constant model was used to fit the data in this case. The “linear gradient model”, adds two horizontal gradient parameters to the zenith parameter, and yielded significantly reduced horizontal position errors. Horizontal and vertical position errors were reduced further with a “second order model”, which adds second-order terms. Evaluation of the atmospheric models using the mountain lee wave case indicated that 1) the linear gradient model cannot express complicated atmospheric disturbances; and, 2) a second-order model reduces horizontal and vertical position errors.
Since 1994, the NOAA Research-Forecast Systems Laboratory (NOAA/FSL) has been evaluating the utility of ground-based Global Positioning System (GPS) remote sensing techniques for operational weather forecasting, climate monitoring, atmospheric research, and other applications such as satellite calibration and validation. Techniques have been developed to acquire, process, distribute GPS integrated precipitable water vapor (IPW) retrievals and ancillary surface meteorological observations every 30-minutes with less than 15 minute latency. Techniques to assimilate these observations into the research version of the Rapid Update Cycle (RUC) numerical weather prediction assimilation/model system running hourly at NOAA/FSL have been developed, and the impacts of these observations on shortrange weather forecast accuracy have been evaluated since 1998 using a 60-km version of the system. These assessments consist of data denial experiments (parallel runs with and without GPS water vapor observations) to determine the impact that GPS-derived integrated (or total column) precipitable water vapor (IPW) retrievals have on short-range moisture and precipitation forecasts. The experiments have been conducted over a portion of the central United States that, from a meteorological perspective, is one of the best-observed areas on Earth. While this greatly facilitates the impact assessments, it also presents a special challenge to a new observing system under evaluation, such as GPS-Met, since relatively few measurements have to “compete” with an enormous number of other (conventional and nonconventional) observations of similar and related parameters. Despite this, five years of experiments in- dicate more or less continuous improvements in 3-hour relative humidity forecasts at pressure levels below 500 hPa. The greatest skill is seen during the cold season when moisture changes are dominated by synoptic-scale weather systems. Perhaps the most significant result is that the impact in improved forecast skill from assimilation of GPS-IPW data has increased each year as the number of stations has increased, suggesting that further increases in the network density over the United States will result in further forecast improvement.
Within the framework of the Helmholtz Association Strategy Project “GPS Atmosphere Sounding” (GASP), an operational monitoring of integrated water vapor was established using 170 GPS sites in Germany and neighboring countries. The product, which can be obtained within 12-15 minutes of computer time on a single Linux PC, is generated each hour with a 30-minute time resolution and an accuracy of±1-2 mm in the precipitable water vapor. The GPS estimates are regularly validated using collocated instruments and the local model (LM) of the German Weather Service (DWD). First experiments for numerical weather predictions are performed at DWD and showed 2% improvement for the relative humidity in a 12-hour forecast, whilst the impact on the precipitation forecast over 24 hours is mixed up to now.
The GPS meteorology has been the driving force behind the near real-time (NRT) GPS analyses during the past few years. High availability of precise near real-time GPS orbits is essential for obtaining reliable solutions in almost all applications. This is especially true when estimating the troposphere parameters. Besides the demand on high accuracy, the orbit products should stand out for their stability in production as well as for their completeness in set of satellites. The most likely candidates for the applicable orbits are products of the International GPS Service (IGS). We will outline, at first, the results of our monitoring activity of the IGS ultra-rapid orbits provided beside their routine usage in Geodetic Observatory Pecny´ (GOP). Thereafter, we will explain our procedure for NRT orbit determination using an efficient technique of combining short orbit arcs into the final orbits over 2-3 days. Our official orbit results, as well as testing variants of the combination, were evaluated using the IGS final and ultrarapid orbit products. Finally, we will demonstrate the direct application of our orbits in the NRT GPS analyses for sensing the regional and global troposphere.
The operational data analysis of the GPS radio occultation experiment aboard the German CHAMP (CHAllenging Minisatellite Payload) satellite mission is described. Continuous Near-Real-Time processing with average time delay of ∼5 hours between measurement and provision of analysis results is demonstrated. A delay of less than 3 hours is reached for individual events. This is made possible by using an operationally operated ground infrastructure, consisting of a polar downlink station, a globally distributed fiducial GPS ground network, a precise orbit determination facility, an automated occultation processing system and an advanced data center (the Information System and Data Center at GFZ, ISDC). The infrastructure was installed within the CHAMP and the German GPS Atmosphere Sounding Project (GASP). More than 120,000 globally distributed occultation measurements were automatically analysed during 2001 and 2002. A set of∼46,000 vertical profiles of refractivity, temperature and water vapor is validated with meteorological analyses from the European Centre for Medium-Range Weather Forecasts (ECMWF) and data from the global radiosonde network. The mean temperature bias in relation to the analyses is less then 0.4 K between 10 and 35 km, the mean deviation of the refractivity is <0.5%. A height dependent standard deviation of ∼1 K at 10 km and ∼2 K at 30 km is observed. This result is confirmed by comparing ∼6,000 CHAMP occultations with corresponding radiosonde measurements. A negative bias of the refractivity in relation to the analyses up to ∼5% in the Tropics is found in the lower troposphere. It corresponds to mean meridional dry biases of the specific humidity up to ∼30%. It is shown, that the application of a heuristic retrieval method, based on the Canonical Transform method and the sliding spectral approach, reduces the refractivity bias on average by a factor of ∼2. The corresponding bias in the specific humidity is reduced by a factor of ∼3. In mid-latitudes almost no more refractivity bias out of the planetary boundary layer is observed. This is shown by a comparison of CHAMP refractivity and water vapor profiles with radiosonde data. More than 50,000 globally distributed electron density profiles were automatically derived during 2001/2002. A validation study including 1,004 comparisons with corresponding ionosonde data yields a bias of 0.18 MHz and 13.4 km for foF2 and hmF2 respectively. The standard deviation is 1.28 MHz (foF2) and 46.8 km (hmF2).
Structure and propagation of equatorial Kelvin waves during May 2001 and December 2002 are observed from the temperature profiles in the upper troposphere and the lower stratosphere using CHAMP and SAC-C GPS radio occultation data. Kelvin waves derived from temperature fluctuations characterize eastward phase propagation in time-longitude section and eastward phase tilts with height in altitude- longitude section between 10 and 30 km. The phase progression spans the range indicating the continuity of Kelvin waves from the upper troposphere to the lower stratosphere. Results show that near the tropopause, Kelvin waves fluctuate temperature by 2 K in general, with wave periods of 12.5-14 days for zonal wave number 1 and 9.3-11.0 days for wave number 2, and vertical wavelengths of 7.6- 8.5 km in 2001 and of 4.4-5.8 km in 2002.
We have analysed the gravity wave activity using temperature profiles retrieved from the GPS occultation experiments on board the SAC-C and CHAMP satellites. By dividing the latitude and longitude ranges, here considered in individual cells, the variability of wave energy as a function of altitude observed in both hemispheres during October, November and December 2001 is presented. Some features were repeatedly observed during the three months. Nevertheless, as most available occultations were occurred during November, we begin showing results from this month. Significant differences detected during October are only pointed out at midlatitudes, provided that sufficient number of occultations are available. A significantly larger wave activity at 40-60°N with respect to 40-60°S is detected during this month. Differences between both the hemispheres are discussed and compared with previous results. Wave activity enhancements at equatorial regions above Brazil, Indonesia and India are found to be correlated with outgoing longwave radiation data. Vertical fluctuations with lengths above and below 3.5 km are considered separately, in order to identify the contribution of Kelvin and Rossby-gravity waves. In both cases, for long and short fluctuations and at the upper troposphere as well as at the lower stratosphere, there is a height interval of several kilometers where a systematic enhanced wave activity is observed. Its average height is progressively decreasing with increasing latitude, same as the height variation in the tropopause location. Longitudinal enhancements are also detected mainly around the equator and at midlatitudes. A clear signature is observed locally in the Southern Hemisphere (SH) at midlatitudes and 70-65°W, for vertical wavelengths longer than 3.5 km. It corresponds to mountain waves forced by the Andes Range, observed whenever occultations are available in this cell.
GPS radio occultation (RO) measurements from a low-earth orbiting (LEO) satellite can determine profiles of atmospheric temperature in the troposphere and stratosphere with high vertical resolution. The RO technique can also provide electron density perturbations in the ionospheric E region. We discuss in this review the application of GPS occultation data for the studies of the dynamical structure of the troposphere, stratosphere and ionosphere. By analyzing RO data obtained by the GPS/MET (GPS/ Meteorology) experiment, the detailed thermal structure near the tropical tropopause has been described. The GPS/MET temperature data have also been used to determine the global distribution of atmospheric gravity wave energy in the stratosphere. These studies indicate enhanced wave activity over regions of tropical convection, in particular, around the Indonesian Archipelago. In addition, orographic generation of atmospheric waves is recognized over the Andean mountain range, whose effects reach the ionosphere, producing the sporadic E layers.
Downward looking (DL) GPS receivers from platforms within Earth's atmosphere (e.g. from a mountain top) can observe GPS satellites and produce an estimate of atmospheric refractivity profiles through the radio occultation technique. The main observations are the radio wave bending angles as a function of the impact parameter. The DL technique provides bending angles at both negative and positive elevation angles. Either a least-squares ray tracing method or the Abel inversion technique can be used for the retrieval of the refractivity from the bending angles. A least-squares method (LS) can be applied directly to the set of positive and negative elevation angles. It allows the retrieval of the refractivity to about 2 km above the receiver height. The Abel inversion operates on a profile of partial bending angles found by subtracting the positive elevation measurement from the negative ones with the same impact parameter. This paper summarizes the promises and limitations of the two inversion techniques using simulated data. The analysis uses both dry and wet atmospheres from a climatological model and real radiosonde data. The results show that both LS and Abel techniques are capable of retrieving the refractivity. The paper also highlights the differences between the two algorithms and the limitations.
Downward-looking (DL) Global Positioning System (GPS) occultation experiments from the top of Mt. Fuji were carried out as a joint project between Kyoto University, Meteorological Research Institute in Japan, and the Jet Propulsion Laboratory (JPL) in the U.S.A. from July 10 to September 25, 2001, in order to obtain temperature, water vapor and pressure profiles near the Earth's surface. A TurboRogue SNR-8000 GPS receiver and a choke ring antenna were installed at the Mt. Fuji weather station located at an altitude of about 3.8 km. Applying an Abel inversion, these DL observations can provide refractivity profiles over an area south of Mt. Fuji. This paper shows temperature, relative humidity, and pressure profiles derived from these refractivity profiles using a one-dimensional variational technique (1D-Var). The derived profiles show agreement with the Mt. Fuji weather station observations within 1.7°C, 1.2%, and 1.0 hPa at the receiver altitude.
Precipitable water vapor (PWV), which was obtained from the nation wide GPS (Global Positioning System) network over Japan, operated by the Geographical Survey Institute (GSI), was assimilated into the meso data assimilation (DA) system of the Japan Meteorological Agency (JMA). Two different methods were examined; one is an optimum interpolation (OI) method, and the other is a 4-dimensional variational (4D-Var) method. Using the analysis data derived from both systems, a number of forecast experiments for rainfall events in Baiu and summer seasons were carried out using the JMA meso scale numerical weather prediction model (MSM). Remarkable improvements in rainfall forecast were seen in several cases, both for the OI and 4D-Var experiments. A statistical score of the 4D-Var experiments, however, showed that the impact of GPS PWV was almost neutral for rainfall forecast, and no substantial improvements were obtained. One of the reasons might be that GPS sites used for the experiments were too few, and sparsely distributed compared to the rainfall systems. Another reason specific for the 4D-Var is that, although the 4D-Var improved PWV analyses, it sometimes modified vertical profiles of water vapor significantly, which brought about different static stability from the first guess of the model (MSM) or from the observations. These results suggested the importance of correct assimilation of vertical profiles of water vapor.
Observation system experiments with JMA MesoScale model, were performed for precipitable water data derived from TMI (TRMM Microwave Imager) and ground-based GPS observation, by using a fourdimensional variational assimilation method. Since GPS data exists over land only, and TMI data are available only over ocean, use of both data can provide information about water vapor complementally over whole analysis domain. Although the number of experiments is not sufficient yet, the results so far suggest that the complementary use of TMI and GPS precipitable water data, can improve the precipitation forecast of the model.
A study on the impact of zenith total delays (ZTD's) obtained from observations by ground based Global Positioning System (GPS) receivers in Europe on the skill of numerical weather prediction (NWP) is presented. The ZTD depends mainly on the local pressure at the GPS site and the integrated water vapour in the column above. Both pressure and humidity information are vital to NWP models, but it is mainly as a source of humidity information the ZTD's are expected to be significant to NWP models. Given the currently meagre amount of humidity observations the inclusion of ZTD's in the data assimilation, determining the starting states for NWP forecasts, is expected to improve forecast skill. Groups in Europe, the US, and Japan are carrying out impact and other studies to assess whether such expectations can be fulfilled. We here present results from parallel runs, with and without ZTD data in the data assimilation, for the period of February 2002, using European ZTD data from the COST Action 716 data sample. Data from 117 GPS stations are included in the assimilations. The runs are performed using the spectral version of the HIRLAM local area NWP model and its variational data assimilation system, HIRVDA, in 3DVar mode. The simulations have been carried out at both 0.45 and 0.15 degree resolution. It is found that statistical verification against observations (heights, winds, temperatures, humidities) from stations on the EWGLAM list indicates mainly neutral impact of GPS ZTD's in this period, with the exception that systematic improvements are seen in the geopotential heights. A detailed comparison of NWP forecasts of precipitation is performed against 12 hour rain gauge observations. Both contingency tables and in particular detailed comparisons (by eye) of precipitation maps show that for this period inclusion of ground based GPS ZTD observations improve the prediction of strong precipitation. In combination with our previous results, (Vedel and Huang 2003), and those reported by other groups, this solidifies the conclusion that GPS ZTD data will improve NWP forecast skill of precipitation. Questions remain to be solved before ground based GPS data can be used in an optimal way in NWP models. These include better knowledge of the errors and error correlations of both the ZTD observations and of the NWP first guess field, as well as better ways of comparing model and observed precipitation. The new European project TOUGH (Targeting Optimal Use of GPS Humidity Observations in Meteorology) will address some of these issues.
This study conducted data assimilation experiments using the operational mesoscale four-dimensional variational data assimilation (4D-Var DA) system of the Japan Meteorological Agency (JMA). Experiments investigated the impacts of GPS-derived water vapor and Doppler radar-derived radial wind (RW) on precipitation prediction for a heavy rain event on 21 July 1999. RW data were obtained from Doppler radars at Narita and Haneda airports. GPS data were obtained from the GPS Earth Observation Network (GEONET) of the Geographical Survey Institute (GSI). Both precipitable water vapor (PWV) and slant water vapor (SWV), which is the amount of water vapor integrated along the slant path between GPS receivers and GPS satellites, were derived from the GPS data. Because SWV contains three-dimensional water vapor distribution information (Seko et al. 2000), we anticipated that assimilating SWV data would more accurately reproduce the moist air inflow at lower layers. Comparisons between observed and model-predicted precipitation regions helped define the impacts of assimilating RW, PWV, and SWV into the model. If the assimilated data included only conventional meteorological data, the model yielded small precipitation regions over a mountainous area far from Tokyo. If the assimilated data included both GPS-derived water vapor data and conventional data, lowlevel inflow in the model was more humid and precipitation occurred along the low-level convergence zone. Because the predicted position of the convergence zone differed from observations, however, the position of the precipitation region was not reproduced correctly. When RW and conventional data were assimilated into the model, low-level northerly flow was reproduced in the northwest of Tokyo. This northerly flow intensified the low-level convergence where precipitation was observed, and the position of the forecasted precipitation was more similar to that of observations. In this model run, however, lowlevel inflow from the south was less humid than observed, and precipitation onset was delayed by 1 hour. If GPS-derived water vapor data, RW data, and conventional data were all simultaneously assimilated, the precipitation position was modeled correctly, and precipitation onset occurred as observed. Comparisons between the vertical cross sections of analyzed water vapor fields and first-guess water vapor fields helped measure the impact of data assimilation on the modeled water vapor distribution. When PWV, RW, and the conventional data were assimilated, water vapor on the windward side of the low-level inflow decreased. In contrast, water vapor in the low-level inflow did not decrease when SWV data were used, instead of PWV data. For this rainfall event, the assimilation of RW and GPS-derived water vapor data improved the precipitation prediction. Assimilation of SWV data improved the representation of the vertical water vapor distribution.
This paper assesses the impact on short-range quantitative precipitation forecasts (QPFs) of assimilating zenith total delay (ZTD) and rainfall observations associated with a winter storm that occurred from 5-6 December 1997 in southern California. Assimilation of hourly rainfall improves the threat score by more than 300% within the assimilation window, but such an improvement drops quickly to 30% or lower beyond this window. The assimilation of ZTD observations does not produce a rainfall distribution as close to the observations as does the assimilation of rainfall within the assimilation window (only a 34% improvement). However, improvement in the QPFs beyond the window from the ZTD experiment is comparable to that from the rainfall experiment. Assimilation of ZTD and rainfall observations modifies the thermodynamic structures of the atmosphere, favoring development of precipitation in the observed rainy areas. The horizontal and vertical wind velocities are also adjusted consistent with the precipitation process. A spectral analysis of observed and simulated hourly rainfall, as well as the model forecast difference with and without data assimilation, indicates that rainfall assimilation adjusts the model variables on smaller scales (25 to 50 km) while the ZTD assimilation adjusts the model variables mainly on larger scales (> 50 km).
In this paper, we describe the GPS radio occultation (RO) inversion process currently used at the University Corporation for Atmospheric Research (UCAR) COSMIC (Constellation Observing System for Meteorology, Ionosphere and Climate) Data Analysis and Archive Center (CDAAC). We then evaluate the accuracy of RO refractivity soundings of the CHAMP (CHAllenging Minisatellite Payload) and SACC (Satellite de Aplicaciones Cientificas-C) missions processed by CDAAC software, using data primarily from the month of December 2001. Our results show that RO soundings have the highest accuracy from about 5 km to 25 km. In this region of the atmosphere, the observational errors (which include both measurement and representativeness errors) are generally in the range of 0.3% to 0.5% in refractivity. The observational errors in the tropical lower troposphere increase toward the surface, and reach ∼3% in the bottom few kilometers of the atmosphere. The RO observational errors also increase above 25 km, particularly over the higher latitudes of the winter hemisphere. These error estimates are, in general, larger than earlier theoretical predictions. The larger observational errors in the lower tropical troposphere are attributed to the complicated structure of humidity, superrefraction and receiver tracking errors. The larger errors above 25 km are related to observational noise (mainly, uncalibrated ionospheric effects) and the use of ancillary data for noise reduction through an optimization procedure. We demonstrate that RO errors above 25 km can be substantially reduced by selecting only low-noise occultations. Our results show that RO soundings have smaller observational errors of refractivity than radiosondes when compared to analyses and short-term forecasts, even in the tropical lower troposphere. This difference is most likely related to the larger representativeness errors associated with the radiosonde, which provides in situ (point) measurements. The RO observational errors are found to be comparable with or smaller than 12-hour forecast errors of the NCEP (National Centers for Environmental Prediction) Aviation (AVN) model, except in the tropical lower troposphere below 3 km. This suggests that RO observations will improve global weather analysis and prediction. It is anticipated that with the use of an advanced signal tracking technique (open-loop tracking) in future missions, such as COSMIC, the accuracy of RO soundings can be further improved.
Challenging Minisatellite Payload (CHAMP) radio occultation (RO) observations during a two-week period are assimilated into global analyses using the National Center for Environmental Prediction (NCEP) three-dimensional variational (3D-Var) system with a recently improved observation operator for assimilating GPS bending angle data. The NCEP 3D-Var system used in this research is suboptimal since Advanced Microwave Unit (AMSU) radiances are not included in our experiments. Analyses with and without CHAMP observations are compared with each other and with collocated conventional radiosonde and dropsonde data, which are excluded from both experiments. Zonal mean temperature, humidity and surface pressure differences between the GPS analyses and NO-GPS analyses are examined. The GPS analyses in the Southern Hemisphere show higher temperatures than the NO-GPS analyses, particularly in the mid- and high latitudes. The GPS analyses show drier air in the lower troposphere and more moist air in the middle troposphere compared to the NO-GPS analyses. The surface pressure is slightly increased (maximum 0.8 hPa) in the Southern Hemisphere and decreased (maximum 0.25 hPa) in the Northern Hemisphere due to the inclusion of GPS observations. Compared with the collocated independent soundings, the large cold bias (as large as 2.5 K) in the NCEP Southern Hemisphere analyses produced without CHAMP observations is significantly reduced. On average, a 20% mean error reduction in the temperature analysis is obtained in the Southern Hemisphere when CHAMP data are included. Degradations in the surface pressure analysis found from previous the GPS/Meteorology data assimilation studies are greatly reduced. The differences between the surface pressure analysis errors with and without CHAMP data are less than 0.8±1.5 hPa. Comparisons of numerical forecasts initialized with analyses produced with and without CHAMP occultations display a small improvement in the forecasts in the tropics and the Southern Hemisphere associated with the use of the CHAMP observations.
We present the concept, some of the approaches used, and the capabilities of the technique referred to as GPS tomography. It is used for retrieval of the 3-dimensional distribution of the refractivity due to atmospheric water vapor. We discuss the presently used methods for retrieval of the primary observable in the GPS tomography, the slant path delay, as well as their shortcomings. Comparisons of GPS slant delays to independent data from a microwave radiometer are included. >From a tomographic point of view we concentrate on the capabilities to retrieve the vertical structure of the wet refractivity. For this purpose we present and apply two methods for tomographic inversion. Both are based on the Kalman filtering technique, where the expected statistical behavior of the refractivity is utilized. The difference between the two is in the way the covariance matrix of the Kalman filter is constructed. We base our study on simulated and real data from the ground network of 8 GPS receivers operating in Göteborg, Sweden. The results demonstrate that at present the limitations of the GPS tomographic technique are errors in the retrieved wet slant delays and their poor geometric distribution.
Observations using a dense network of 75 Global Positioning System (GPS) receivers were carried out in Tsukuba, Japan, in October-November 2000 and July-September 2001. We have applied a tomography technique on the slant path propagation delay of the GPS signals to derive the temporal and spatial distribution of water vapor. In particular, we have employed the moving cell method, which was originally developed by Seko et al. (2000) for a beta-mesoscale precipitation system associated with the Baiu front. Because we are interested in local-scale phenomena, we tried to determine the water vapor distribution with a time resolution of ten minutes. The tomographic analyses resolve the water vapor variations during the meteorological events studied here.
The Tsukuba GPS Dense Net campaign that took place in the autumn of 2000 and in the summer of 2001 measured the meso-γ scale distribution of water vapor. As part of the campaign, 75 GPS receivers and 20 automatic meteorological observation systems (AMOS) recorded water vapor variations associated with a thunderstorm on 1 August 2001. The three-dimensional water vapor distribution in the area was estimated from slant water vapor (SWV) data derived from GPS receivers using tomographic methods. SWV is the total amount of water vapor per unit area between a GPS receiver on the ground and a GPS satellite. The SWV data used in this study were obtained with sufficient accuracy by carefully removing multi-path effects, phase center variations of the GPS antenna, and other error sources. SWV was converted to a value that was projected onto the vertical direction (VSWV), so that the influence of the elevation angle on the slant path was removed. VSWV values from adjacent receivers to individual satellites were strongly correlated with each other. Variations in VSWV depended on the GPS receiver positions relative to the developing or moving thunderstorm. Our results indicate that SWV data can provide useful information about the water vapor distribution in the vicinity of thunderstorms. Correlations between the variation of VSWV and the precipitable water vapor (PWV) distribution around the GPS receiver were also calculated. The directions of the large values of VSWV corresponded to regions of high PWV. The three-dimensional water vapor distribution, estimated tomographically, agreed well with Doppler radar-observed reflectivity. Regions of high water vapor near the surface occurred on the northern side of a region of intense reflectivity. The more humid regions above 3 km corresponded to regions where reflectivity increased. The water vapor distribution estimated from the GPS showed an increase of water vapor above a height of 1 km, which preceded the appearance of radar echoes by about 20 minutes during the thunderstorm formation.
We characterize the temporal and spatial variation of the Zenith Wet Delay (ZWD) and the Zenith Total Delay (ZTD), estimated using a network of Global Positioning System (GPS) receivers. This characterization is important for the improvement and validation of atmospheric water vapor models, applicable in GPS meteorology, as well as in the navigation. We treat the estimates of Zenith Hydrostatic Delay (ZHD) and ZWD as realizations of random walk stochastic processes and we derive the corresponding parameters for different locations and measurement techniques for data acquired at intervals of 1 to 3 hours. The monthly standard deviation (StD) of the ZTD is less than 50 mm and does not exhibit a strong seasonal signature over the period 1997-1998 for any of the studied GPS sites. However, the StDs of the pairwise-differenced ZTD time series show a seasonal dependence, mainly due to the spatial variations of the ZWD, which should be considered when GPS data are assimilated in weather prediction models. We show the differences in typical spatial characteristics of ZHD and ZWD for the winter and summer seasons in North Europe. Finally, we describe the use of temporal structure functions for detection of rapid changes in ZTD.