Comparison of AIRS Satellite, ERA Model and Radiosonde Convective Available – Potential Energy Data in the Southern Great Plains Region

[fusion_old_tabby title=”Abstract”]

High priority must be given to research for Earth remote sensing applications, especially relating to the forecasting of severe weather, due to its destructive consequences. Convective Available Potential Energy (CAPE) is routinely used to characterize convection as having moderate or severe potential. Relating a climatology of CAPE to near real time observations from meteorological sensors on new weather satellites is a valuable tool in assessing the risk of severe weather. Satellite data products from AQUA AIRS were used to compute a ten year climatology for the Southern Great Plains region. CAPE was computed from vertical profiles of pressure, temperature, and dew point temperature from high vertical resolution AIRS soundings (101 levels) using the SHARPpy algorithm used by the National Weather Service Storm Prediction Center (NWP). It was found that numerical estimates of CAPE are sensitive to the vertical smoothing of the temperature and moisture profile. In addition, error in the surface parcel dew point estimate degrades the accuracy of CAPE, but can be corrected when the satellite estimated surface dew point is limited to agree with the radiosonde measurement to within 1 °C. Further improvements in estimating CAPE will allow us to take full advantage of this satellite data for near-real time forecasting.

[fusion_old_tabby title=”Full Text”]

Introduction

A common goal of the severe storm science community is to obtain accurate information in a timely manner regarding atmospheric stability. This information can be used to communicate predictions of severe weather events. To better inform how to make these predictions, a ten year record of upper air sounding profiles from the Department of Energy Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP) site was used to create a climatology of Convective Available Potential Energy (CAPE) [11]. CAPE is a measurement of the amount of energy a parcel of air contains, which is effectively the buoyancy of that parcel. CAPE is commonly used by meteorologists as an indicator of severe weather. The seasonal variation in CAPE and dew point is shown in Fig. 1. We then look at the ability of satellite observations to characterize the CAPE probability distribution function at the 1:30 am/pm overpass times as a function of distance from the SGP site.

FIG. 1 | NASA Aqua MODIS imagery May 31, 2013 at 18:30 UTC (1:30 pm local time)

FIG. 1 | NASA Aqua MODIS imagery May 31, 2013 at 18:30 UTC (1:30 pm local time)

FIG. 2 | CAPE computed using SHARPpy from model ECMWF ERA-Interim at 18 UTC (top) and from NASA AIRS L2 v6 18:30 UTC (bottom). Circle symbol marks Nonnan, Oklahoma near El Reno. Star symbol marks the location of the DOE ARM SGP site.

FIG. 2 | CAPE computed using SHARPpy from model ECMWF ERA-Interim at 18 UTC (top) and from NASA AIRS L2 v6 18:30 UTC (bottom). Circle symbol marks Nonnan, Oklahoma near El Reno. Star symbol marks the location of the DOE ARM SGP site.

The ability to measure vertical profiles of water vapor from space at times when ground-based upper air soundings are not available can fill an important need in short-range weather prediction. New satellite observations allow for the retrieval of water vapor measurements with higher vertical resolution than was previously available. In order to demonstrate the advantages of these new data opportunities, it’s important to look at a practical application. Supercell thunderstorm events, like the El Reno, Oklahoma tornado on May 31, 2013, are examples of just how dangerous and unpredictable tornados can be. Fig. 2 demonstrates the severe convection from this disastrous storm. The El Reno case study will be used to illustrate the potential value of estimating CAPE values from satellite soundings.Fig. 3 compares the regional CAPE for the El Reno case study between European Centre for Medium-Range Weather Forecasts (ECMWF), ERA-Interim model fields, and NASA Atmospheric Infrared Sounder (AIRS) L2 v6 satellite observations.

The ECMWF reanalysis is a European climate model that assimilates observations to give a numerical description of the atmosphere. While AIRS and ERA data are very similar to each other, there are notable differences in the precise location of the most extreme CAPE values with the satellite observation of peak CAPE slightly west of the model analysis. Note that El Reno is just west of Norman, Oklahoma and is where the supercell formed that produced the El Reno tornado showing the close proximity to the radiosonde launch site. Fig. 4 provides an example comparison of the vertical soundings of temperature and dew point temperature at the DOE ARM SGP site (just north of El Reno) at about noon on that day. Note that the radiosonde profile has much higher vertical resolution than either the NWP model or the satellite retrieved profile from AIRS.

National Weather Service (NWS) forecasters are currently using Geostationary Operational Environmental Satellite (GOES) sounder products for a range of applications, for which they are obtaining positive results. These products include estimates of total precipitable water vapor (TPW) and atmospheric stability indices, such as convective available potential energy (CAPE) and lifted index (LI). Infrared observations from geostationary orbit capture the diurnal cycle of surface skin temperature with data
collected over the continental United States every hour. These geostationary data profiles can contribute to weather warnings [10] . However, the limited number of infrared spectral channels fundamentally limits the vertical resolution of the existing GOES sounder thermodynamic products, and utilizing other sensors could fill this gap. Unlike the current GOES sounder, new high spectral resolution infrared sensors on polar orbiting weather satellites (POES) can sense the atmospheric boundary layer at specific times of day (about 10:30 am/pm and 1:30 am/pm). Forecasters could make use of CAPE estimates from operational satellite sounders, such as CrIS and IASI, on JPSS and METOP platforms during the most unstable daytime period considering there are no operational ground level observations during this time.

FIG. 3 | Temperature and dewpoint temperature vertical profiles from ARM Vaisala RS92 Radiosonde (red), ECMWF ERA-Interim (black) and NASA AIRS L2 Version 6 (blue) at the DOE ARM SGP site on 31 May 2013 at about 18:00 UTC (noon).

FIG. 3 | Temperature and dewpoint temperature vertical profiles from ARM Vaisala RS92 Radiosonde (red), ECMWF ERA-Interim (black) and NASA AIRS L2 Version 6 (blue) at the DOE ARM SGP site on 31 May 2013 at about 18:00 UTC (noon).

For example, the Atmospheric Infrared Sounder (AIRS) has been used to provide quantitative information about the lower atmosphere [4]. Software to process AIRS data in near real-time has been included in the IMAPP direct broadcast software package [1] [13]. Near real-time data assimilation of polar orbiting advanced sounder products into rapidly updated NWP models have the potential to provide a positive impact for future warnings on forecasts [10]. The objective of this paper is to make a quantitative assessment of CAPE derived from high spectral resolution infrared sounders. Satellite overpasses of the Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) program Southern Great Plains (SGP) site provide the validation dataset needed to make this assessment [8]. This analysis makes use of a combination of research radiosondes launched from the DOE ARM SGP site, ECMWF model reanalysis fields, and coincident NASA AIRS satellite observations. A characterization is made of the temporal, spatial, and measurement accuracy of CAPE derived from hyperspectral infrared weather satellites. An analysis of CAPE sensitivity
to these errors will help outline whether or not utilizing near real-time hyperspectral satellite soundings of temperature and water vapor from direct broadcasting instruments is a plausible path for accurate severe weather forecasting.

Methods

To obtain a statistically useful range of CAPE values in the U.S. Southern Great Plains, vertical profiles of pressure, temperature, and water vapor were obtained for the time period January 1, 2005 to December 31, 2014 for a region centered on the ARM SGP central facility. The ARM site was chosen for this study because routine radiosonde launches at 6 Coordinated Universal Time (UTC) and 18 UTC are within 1.5 hours of the nominal satellite overpass times of the AQUA satellite. Values of CAPE were computed for each vertical profile using software consistent with that of the NWS Storm Prediction Center (SPC) [7]. Only cases with radiosonde profiles having CAPE greater than 50 J/kg were included in the analysis. This threshold was used to eliminate the large number of zero (or small) CAPE values that are not relevant for severe weather. To investigate spatial sampling issues, CAPE values from AIRS and ERA were selected within a radius of 50, 150, and 250 km of the ARM SGP central facility. The selected CAPE values for a time and space region are used to create histograms using a uniform bin size of 50 J/kg. Normalizing by the sum of the histogram creates a probability distribution function (PDF). PDF at the 25th, 50 th, 67 th, 75 th, 95 th and 99th percentiles were tabulated to quantify the characteristics of each CAPE distribution shown in Fig. 5.

To analyze the dependence of CAPE on the vertical resolution of the temperature and water vapor profiles, the radiosonde profiles were smoothed with a vertical boxcar function at width values (1 3 5 7 9 15 21 27 35 41 47 53) of 75 meters per layer. The CAPE computed from the smoothed profile was differenced from the original radiosonde CAPE for each profile. The following equation defines CAPE [2] [9], where g is the acceleration due to gravity, LFC is the level of free convection, EL is the equilibrium level, Tparcel is the temperature of the parcel, and Tenv is the environmental temperature.

Equation

The forecasted parcel is a parcel estimate at the expected time of convection. This paper utilizes the surface parcel method for calculating CAPE. In particular, this study uses the SHARPpy software routines described in Halbert [7]. SHARPpy is a python software library that can be used by the research community and is derived from the SHARP software used operationally at the Storm Prediction Center [7].

Results

An analysis was performed to understand the ARM radiosonde, ERA, and AIRS CAPE sensitivity to spatial, temporal, vertical, and measurement error.

Spatial Sampling Error

For the satellite product, a spatial sampling error can exist when the AIRS profile coincident with the SGP site location is invalid (e.g. overcast and the closest valid profile is some distance away). To quantify this issue an analysis was created for CAPE values within a radius of the ARM SGP site. The spatial sampling error for the ERA interim can be neglected because the model grid is continuous over the domain of interest [7]; however, the ERA was analyzed in the same circular region for consistency with the AIRS analysis. There is no significant spatial sampling error in the AIRS product due to invalid retrievals when using the quality control criteria given in Table 1 ( Supplementary figures).

Temporal Sampling Error

emporal sampling error can be an important error in CAPE estimation given the rapid boundary layer changes due to surface heating during the day and cooling after sunset. For this reason, operational radiosondes launched at 0 and 12 UTC (6 am and 6 pm) are not ideal for of assessment of CAPE during mid-day in the SGP region. The ARM SGP site was chosen for this study of AIRS CAPE because research grade radiosondes are launched at 6 and 18 UTC (about midnight and noon local time). The afternoon satellite overpass is at about 1:30 pm (19:30 UTC) with some variation from day to day. The radiosonde launch time and Aqua satellite overpass time difference is typically less than 2 hours. To further minimize this relatively small temporal sampling error, the radiosonde data was interpolated to each Aqua satellite overpass times. The ERA Interim analyses are available only at 0, 6, 12, and 18 UTC. For this study, the 6 and 18 UTC ERA analysis fields are used without time interpolation and thus represent the atmospheric state about 1 to 2 hours prior to the satellite overpass, but they are time coincident with the ARM SGP radiosonde launches.

Vertical Resolution Error

The ARM radiosondes were used to investigate the dependence of CAPE on vertical resolution of the temperature and moisture sounding. In the ARM product file used for this study, the radiosonde data has been interpolated to 200 height bins with a spacing of 75 meters and two additional bins at 2 m and 30 m. A boxcar smoother was applied to each vertical profile for a range of boxcar full widths, and CAPE was recomputed for each smoothed profile. Results are summarized in Fig. 5. There is an 11% reduction in CAPE for a full-width vertical smoothing of 1,000 meters. A vertical smoothing of 2,000 m causes a reduction in CAPE values of about 18%. AIRS has a reduction of 17% at the 50th percentile, which is roughly consistent with the expected vertical resolution of about 2 km for retrieved water vapor profiles inherent to the hyperspectral infrared [2]. The ERA model 50th percentile is biased by –23% relative to radiosonde profiles, which may be due to a vertical smoothing inherent in the NWP model, particularly with respect to the vertical layering of water vapor. This is apparent in the dew point profiles of the case study examples shown in Fig. 4.

Vertical smoothing of the radiosonde profiles can occasionally lead to a temporary increase in CAPE values as indicated by the positive outliers. Fig. 7 illustrates the effect of smoothing a profile containing a nocturnal temperature inversion. The surface parcel temperature increases with smoothing in this case, which leads to an increase in the CAPE value computed from the smoothed profile. Investigation of these anomalous cases of increasing CAPE shows they are all profiles at night containing a temperature inversion.

Measurement Error

he case study analysis revealed a sensitivity of CAPE to surface parcel temperature and dew point temperature (i.e. an error in the surface parcel estimate could cause errors in the computed CAPE). In order to quantify this error, the ARM radiosonde surface temperature and dew point temperature were used as references to compute the error in the surface parcel estimates from the closest AIRS retrieved profiles and ERA reanalysis profiles to the ARM SGP launch site. The mean differences are less than 1 °C in each comparison with a standard deviation of about 2 °C; however, this good, statistical agreement in the mean disguises an error when CAPE is non-zero. Table 2 (Supplementary figures) shows the result of analyzing the 10-year matchup dataset for the subset of cases with CAPE greater than a minimum cutoff. The most notable feature in Table 2 is how the error in surface dew point changes from near zero for all CAPE values to –2 °C for the subset with CAPE greater than a minimum value of 50 J/kg. This is error is the same for both AIRS and ERA. As the CAPE minimum threshold increases, the error grows for both ERA and AIRS with ERA exceeding –5 °C and AIRS exceeding –7 ° C for the CAPE values greater than 2500 J/kg. Surface air temperature error also grows with increasing CAPE, but the error is less than half as large as the dew point temperature error. Table 2 shows very similar behavior between ERA and AIRS for CAPE up to 1500 J/Kg. For higher CAPE values, the AIRS bias error exceeds that of ERA although both have equally large standard deviations.

FIG. 6 | Vertical smoothing error by smoothing width.

FIG. 6 | Vertical smoothing error by smoothing width.

FIG. 7| An example of a smoothed soundings and an original radiosonde sounding on August 27, 2005 at the ARM SGP site containing a nocturnal temperature inversion.

FIG. 7| An example of a smoothed soundings and an original radiosonde sounding on August 27, 2005 at the ARM SGP site containing a nocturnal temperature inversion.

To characterize the extent to which errors in the surface parcel estimates lead to error in the derived CAPE estimates, a correlation coefficient was computed between AIRS and ARM radiosonde for nonzero CAPE values with a range of quality control criteria. Decreasing the AIRS cloud fraction cutoff increases the correlation with ARM radiosondes from 0.35 to 0.5, however the highest correlation (>0.8) is achieved only when the surface dew point of AIRS is within 1 C of the ARM site radiosonde. This is illustrated in Fig. 8. The left hand panels show the variation with cloud fraction, from 0.84) can be obtained even for AIRS cloud fractions up to 0.8 as long as the surface dew point estimate is within 1 degree of the truth. Fig. 8 also shows the comparison of AIRS and ERA with a similar restriction on cloud fraction and surface dew point error. The correlation coefficient between AIRS and ERA increases from 0.37 to 0.88 when the surface dew point temperatures agree to within 1 ° C independent of AIRS cloud fraction.

Discussion

In previous works, several authors validated AIRS retrieved temperature and moisture vertical profiles[5] [6] [3] [12]. Only a limited study has been published previously on the accuracy of CAPE derived from AIRS profiles relative to radiosondes [13]. That study commented on the lack of correlation of CAPE derived from AIRS to the CAPE derived from the small number of radiosonde profiles considered, but no explanation was provided for the result. For the current study, a long time series of AIRS and radiosonde
matchups was created to characterize the systematic biases and random characteristics of the hyperspectral infrared satellite retrievals estimations of CAPE. As shown in Fig. 5, the AIRS and ERA CAPE distributions share similar characteristics, including a smaller median value relative to ARM radiosondes. This under-estimate is consistent with the lower vertical resolution of the satellite and NWP products. The relatively poor correlation of AIRS and ERA CAPE, with matched ARM SGP radiosondes (0.35 and 0.5, respectively), is explained by an error in estimation of the surface parcel dew point temperature.

The systematic bias found in the AIRS derived CAPE and the ERA-Interim derived CAPE is consistent with the known reduced vertical resolution of NWP and satellite retrievals compared to radiosondes. However, the scatter in the AIRS and ERA-Interim CAPE values relative to radiosondes was shown to be primarily due to error in the estimate of the surface parcel dew point temperature. To account for this error and develop a correction method, a comparison between ASOS automated surface observations at Ponca City, OK (near the ARM SGP site) and AIRS retrieved surface temperature and dew point was conducted. A time series plot was created to see the seasonal variation in the two sets of data as shown in Fig. 9.

FIG. 8| AIRS vs. ARM radiosonde all data with AIRS cloud fraction less than 0.8 (upper left) and a subset with surface dew point within one degree (upper right), all data with AIRS cloud fraction less than 0.1(lower left) and a subset with surface dew point within one degree (lower right)

When comparing surface temperatures, AIRS is seeing higher temperatures in the winter than ASOS. In summer, when CAPE is high, there is a fair amount of scatter but no bias for AIRS 2 meter temperature as seen in Fig. 10. In contrast, the AIRS estimated dew point is drier than the ASOS estimated dew point by several degrees in the summer. This is consistent with what we found at the ARM site. The next step is to use the ASOS surface temp and dew point in updating the satellite CAPE calculation. Future work includes the use of ASOS surface temperature and dew point observations coincident with AIRS soundings to improve CAPE estimates in near-real time for the continental US east of the Rockies.

Conclusions

A comparison of CAPE was made for the U.S. Southern Great Plains region using a combination of DOE ARM radiosondes, ERA model reanalysis fields, and AIRS satellite observations. CAPE estimates were evaluated
for spatial, temporal, vertical resolution, and measurement errors. Numerical estimates of CAPE are sensitive to the vertical smoothing of the temperature and moisture profile. A vertical smoothing of 1-2 km leads to a reduction in the 50th percentile of CAPE by 10-20 percent. In addition, error in the surface parcel dew point estimate is found to degrade the accuracy of CAPE. For CAPE values greater than 50 J/kg,
both AIRS and ERA-Interim surface dew point temperatures are dry by 2 degrees compared to the surface radiosonde observations. This error increases to more than 5 °C for CAPE exceeding 2500 J/kg. Improvements of surface parcel dew point temperature can be expected to improve the CAPE estimates derived from both hyperspectral infrared satellite observations and NWP forecasts. This suggests that merging surface station meteorological data and available boundary layer profiling with satellite profiles could greatly improve the utility of the hyperspectral satellite sounding products and the NWP model fields. Timely and useful information on the evolution of the vertical structure of the atmosphere is available from the satellite overpasses at 10:30am (EUMETSAT/METOP) and at 1:30pm (NASA Aqua and NOAA/JPSS) and should be exploited for NWS forecasting applications.

Acknowledgements

The author wishes to thank Genevieve Burgess for assisting in the acquisition of AIRS products and discussions with Greg Blumberg on the use of the SHARPpy software. We acknowledge the support of NOAA grant NA10NES4400013. In addition, acknowledgment is made for the contribution of Dave Turner for the ARM data obtained from the DOE archive at http://www.archive.arm.gov.

[fusion_old_tabby title=”Figures”] [fusion_old_tabby title=”References”]
  1. Al SET. Dual-Regression Retrieval Algorithm for Real-
    Time Processing of Satellite Ultraspectral Radiances. J
    Appl Meteor Climatol. 2012; 1455–1476. doi:10.1175/
    JAMC-D-11-0173.1
  2. Bedka, S., Knuteson, R., Revercomb, H., Tobin, D.,
    & Turner D. An assessment of the absolute accuracy
    of the Atmospheric Infrared Sounder v5 precipitable
    water vapor product at tropical, midlatitude, and
    arctic ground. J Geophys Res Atmos. 2010; 181–202.
    doi:10.1175/2009JAMC2286.1
  3. Botes D, Mecikalski JR, Jedlovec GJ. Atmospheric
    Infrared Sounder ( AIRS ) sounding evaluation
    and analysis of the pre-convective environment. J
    Geophys Res Atmos Geophys Res,. 2012;117: 1–22.
    doi:10.1029/2011JD016996
  4. Chahine MT. Improving Weather Forecasting and
    Providing. Bull Amer Meteor Soc. 2006; 911–927.
    doi:10.1175/BAMS-87-7-911
  5. Divakarla MG, Barnet CD, Goldberg MD, McMillin
    LM, Maddy E, Wolf W, et al. Validation of Atmospheric
    Infrared Sounder temperature and water vapor
    retrievals with matched radiosonde measurements
    and forecasts. J Geophys Res Atmos. 2006;111: 1–20.
    doi:10.1029/2005JD006116
  6. Fetzer EJ, Lambrigtsen BH, Eldering A, Aumann HH,
    Chahine MT. Biases in total precipitable water vapor
    climatologies from Atmospheric Infrared Sounder and
    Advanced Microwave Scanning Radiometer. J Geophys
    Res Atmos. 2006;111: 1–14. doi:10.1029/2005JD006598
  7. Halbert KT, Blumberg WG, Marsh PT. SHARPpy:
    Fueling the Python Cult. Prepr 5th Symp Adv Model
    Anal Using Python. 2015; 1–3.
  8. Hart, J. A., J. Whistler, R. Lindsay and MK. NSHARP,
    version 3.10. Storm Predict Center, Natl Centers Environ
    Predict. 1999;
  9. Rasmussen EN, Blanchard DO. A Baseline Climatology
    of Sounding-Derived Supercell andTornado Forecast
    Parameters. Weather Forecast. 1998;13: 1148–1164.
    doi:10.1175/1520-0434(1998)0132.0.
    CO;2
  10. Stensrud DJ. Convective-scale warn-on-forecast
    system. Bull Amer Meteor Soc. 2009; 1487–1499.
    doi:10.1175/2009BAMS2795.1
  11. Stokes GM, Schwartz SE. The Atmospheric Radiation
    Measurement (ARM) Program: Programmatic
    Background and Design of the Cloud and Radiation
    Test Bed. Bulletin of the American Meteorological
    Society. 1994. pp. 1201–1221. doi:10.1175/1520-
    0477(1994)0752.0.CO;2
  12. Tobin DC, Revercomb HE, Knuteson RO, Lesht BM,
    Strow LL, Hannon SE, et al. Atmospheric Radiation
    Measurement site atmospheric state best estimates
    for Atmospheric Infrared Sounder temperature and
    water vapor retrieval validation. J Geophys Res Atmos.
    2006;111: 1–18. doi:10.1029/2005JD006103
  13. Weisz E, Sr WLS, Smith N. Advances in simultaneous
    atmospheric profile and cloud parameter regression
    based retrieval from high-spectral resolution radiance
    measurements. J Geophys Res -Atmospheres. 2013;118:
    6433–6443. doi:10.1002/jgrd.50521
[fusion_old_tabbyending]
2016-10-16T23:52:36+00:00 April 16th, 2016|