use of gps to adjust radisondes: validation using airs

L.M. McMillin NOAA/NESDIS/ORA

Use of GPS to Adjust Radisondes: Validation using AIRS

Larry M. McMillin, Seth I. Gutman, Jiang Zhao,

M. K. Rama Varma Raja, James G.Yoe


Main Points

• Surface based GPS measurements can be valuable for– Removing biases from radiosonde moisture reports

– Improving the initial state of numerical models

• Steps required to achieve the potential value– Locate a GPS at each radiosonde site

– Make the measurements that are taken (sometimes for other purposes) available for moisture measurements in a timely fashion

• AIRS validation– 3way total water vapor validations


GPS radiosonde approach background

• Radiosondes are subject to calibration errors– Vaisala packaging issue

• There can be significant radiosonde to radiosonde biases– The United States has a policy of flying radiosondes from at least two

different vendors at any one time

• Procedures have been developed at the ARM sites– Use the upward looking microwave to adjust the individual radiosonde

reports

• The microwave sensors are expensive and rare– The GPS water vapor has many of the same features, but is cheaper and

widely available


GPS radiosonde approach background

• GPS sensors are installed at many locations for other purposes– Many of these could make the water vapor measurements

– Some countries/agencies have a policy of not making the reports available

– There is no standard procedure reporting and distributing the information

– Needs national/international decisions and policy (WMO)

• In the United States, some GPS sites are at or close to existing radiosonde sites

• These have been used in this study


GPS radiosonde approach

• Take the GPS IPW and use the radiosonde IPW to calculate an adjustment factor by taking the ratio

• Apply the ratio to all the radiosonde layer precipitable water values

• Compare the adjusted profile to the values retrieved by the AIRS instrument

• Note – GPS takes continuous measurements– Use the GPS at the radiosonde time for the radiosonde comparison

– Use the GPS at the satellite time for the satellite comparison

– Use the GPS to track changes in moisture with time• Discard cases with large changes

• Use the GPS to adjust for the time changes


GPS radiosonde approach

• Evaluate the 3 U.S. RAOB types – RS80-37H

– VIZ B2

– RS MSS


Figure 1. Locations of the GPS stations located near radiosonde stations and used in this study. Stations marked with x’s are not used because local terrain conditions made the results differ. Stations marked without a radiosonde type had no AIRS matches and could not be used.


Figure 3 RMS and bias as a function of pressure for the Meteorological Sounding System (MSS) made by the Space Data Corporation. The lines labeled as RGPS and AGPS show the match with radiosondes and AIRS after adjustments. The AGPS has the smaller time difference.


Figure 4 RMS and bias as a function of pressure for the VIZ-B2 radiosonde. The lines labeled as RGPS and AGPS show the match with radiosondes and AIRS after the GPS has been used to adjust the radiosonde. The AGPS has a smaller time difference.


Figure 5 RMS and bias as a function of pressure for the Vaisala RS80-57H radiosonde. The lines labeled as RGPS and AGPS show the match with radiosondes and AIRS after the GPS has been used to adjust the radiosonde. The AGPS has the smaller time difference.


Costs for a GPS sensor

• Here are the approximate costs for a GPS water vapor sensor:– Receiver & antenna: $5,500

– PC for data logging and communications: $1,000

– antenna installation: $100 > $10,000 depending on the type of monument and desired degree of stability

• Communications: depends on what's available.

• Collocated surface met sensors:– $5,500 and up depending on what's wanted.

• Average total cost: – $15,000 per system + recurring costs for maintenance and communications.


Recommendation 1

• Place a GPS sensor at all U.S. radiosonde stations

• Recommendation 1a– North America

• Recommendation 1b– North and South America

• Recommendation 1c– The world


GPS water vapor can be valuable input to numerical models

• Note that satellite measurements based on emitted atmospheric radiation are subject to surface effects over land

• GPS sensor can offer a supplemental coverage over land areas

• Data are taken at frequent time intervals

• The next series of slides have been provided by Seth Gutman

• They illustrate the potential benefits of using the GPS IPW values in the model

• Note for this and the cases just discussed, the IPW has the greatest effect on the surface layers because they dominate the IPW values


Overview

The GPS-Met project started in 1993 as a collaboration between FSL, UCAR and NCSU to determine how well and under what circumstances GPS could be used to measure IPW.

It has evolved into a collaboration between FSL, other NOAA organizations, other federal, state and local government agencies, universities, and the private sector.

Major accomplishments include: - specification of the GPS-IPW observation

accuracy and error covariance; - development of real-time data

processing techniques for operational weather forecasting; - verification of positive impact

on Wx forecast accuracy; -definition and exploration of new applications, including radiosonde moisture sounding QC and cal/val of satellite moisture soundings.


Project Linkages

Ground-BasedGPS-Met

11/94

RefractivityModels

NWPModels

AircraftObservations

OtherObservingSystems

SubjectiveForecasting

Interagency-InternationalCollaboration

ARM SGP - 94 SMO – 03 ARM NSA - 03 MLO - 04

ClimateObservations

NWP ModelVerification

RadiosondeMoisture

Obs

SatelliteMoistureProducts

ObservationVerification

ClimateModel

Verification


GPS-Met Network

318 sites operating with 28 in checkout or waiting for antenna positions


Multi-year study with the 60km RUC indicates that GPS makes a small but consistent positive impact on short-term weather forecast accuracy:

Primarily at the lower levels where most of the moisture resides - IPW more correlated w/ low-level moisture

Magnitude of impact consistently increases with the number of stations

RH forecast improvement is greatest in the cool months when convection is less frequent and the moisture distribution is more synoptic scale.

Impact on precipitation forecast accuracy generally increases with precipitation amount threshold

GPS NWP Impact Tests

Verification areaNo. Stations 18 56 67 100+ 200+

Year 1998-99 2000 2001 2002 2003Level % improvement (normalized by total error)850 1.5 3.8 3.9 5.0 5.4700 1.1 4.1 6.3 6.5 7.0500 0.7 2.1 2.0 2.4 3.1400 0.3 0.1 -0.4 -0.5 1.0

Mean (850-400) 0.9 2.5 2.9 3.3 4.1Mean (850-500) 1.1 3.3 4.1 4.6 5.2


Impact of GPS-IPW increases as the number of GPS Observations Increase

Positive impact to 500 hPa, but largest Impact at 700 and 850 hPa

With GPS Without GPS

Improved PBL in RUC 20 increases 3-h CAPE forecast by 500 J/kg, but with GPS, the CAPE exceeds 1000 J/kg. Only model to forecast these storms!


SITE ST GPS RAOB DIST (km) VERT (m)18 - Sterling VA zdc1 iad 15.3 13.019 - Caribou ME carm car 0.3 6.6

Table 2. Other GPS-Met Sites evaluated for the Radiosonde Replacement Program

SITE ST GPS RAOB DIST (km) VERT (m)1 - Jacksonville FL jxvl jax 0.5 1.92 - Key West FL kwst eyw 0.6 3.93 - Tallahassee FL talh tlh 0.6 2.34 - Blacksburg VA blkv rnk 0.8 5.45 - Hilo HI hilo hto 0.8 0.16 - Flagstaff AZ fst1 fgz 0.9 22.17 - Salt Lake City UT slcu slc 1 5.68 - Cape Canaveral FL ccv3 xrb 1.2 6.29 - Pago Pago AS aspa stu 1.6 48.210 - Anchorage AK anc1 anc 2.9 3.211 - Grand Junction CO mc01 gjt 3.2 17.812 - Lake Charles LA mcne lch 5.6 12.113 - Bismark ND bsmk bis 7.1 67.514 - Point Barrow AK sg27 brw 7.7 0.115 - Corpus Christi TX txcc crp 8 116 - Vandenberg AFB CA van1 vbg 8.5 13.717 - Gaylord MI nor2 apx 9.3 53.5

Table 1. GPS-Met sites within 10 km of an Upper-Air site

GPS Radiosonde Moisture QC for RRS


GPS Radiosonde Moisture QC for RRS

13

2

4

6

7

910

11

12

13

14

15

17

18

19

8

16

5

Non NWS U/A Site

NWS U/A Site

RRS Study Sites


13

- B

ism

ark

5 -

Hilo

17 -

Gay

lord

10 -

Anc

hor

age

6 -

Fla

gst

aff

7 -

Sal

t La

ke C

ity

11 -

Gra

nd

Junc

tion

9 -

Pa

go P

ago

3 -

Tal

lah

asse

e

1 -

Jac

kson

ville

14

- P

oint

Ba

rro

w

4 -

Bla

cksb

urg

12

- L

ake

Cha

rle

s

15 -

Cor

pus

Chr

isti

16 -

Va

nde

nbe

rg A

FB

2 -

Key

We

st

8 -

Cap

e C

ana

vera

l

-1 .0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

IPW

V (

cm)

-1 .0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

-0.13 -0 .05 -0 .03 -0 .03 -0 .02 -0 .02 -0 .02 -0.01 -0.01 -0.00 0.04 0.06 0.10 0.10 0.12 0.13 0.26-0.12 -0 .04 -0 .04 -0 .03 -0 .02 -0 .03 -0 .01 -0.03 -0.02 -0.00 0.04 0.05 0.09 0.08 0.06 0.11 0.24

0.19 0.33 0.11 0.10 0.09 0.11 0.07 0.22 0.13 0.14 0.09 0.12 0.21 0.26 0.29 0.23 0.350.22 0.35 0.19 0.13 0.15 0.17 0.11 0.47 0.20 0.22 0.12 0.21 0.36 0.33 0.34 0.28 0.39

One year, all sites sorted by mean difference (sonde-GPS)

M ean and R M S of a ll observations

M ean and R M S of observations m inus outlie rs > 2 R M SD


GPS for Space Wx forecasting

Global satellite Cal/Val Global

Climate Monitoring

International DGPS sites have GPS, need Sfc Met & improved Comms

International GPS Service sites have GPS, need Sfc Met & improved Comms

GCOS/GUAN sites need GPS & improved Comms

Additional Links


3 way comparisons – AIRS validation

• Radiosondes, AIRS and GPS can all provide a measurement of IPW

• comparing all 3 has significant advantages over a 2 way comparison– When an anomaly occurs, the suspect is easily identified

– Nose estimates are more robust


3 way comparisons – AIRS validation

• An important question for scatter plots– Best prediction

– Best estimate of the underlying relationship

• Since all the values have errors, the best fit slope is always biased towards the independent variable

• For the best estimate, rotate both variables by 45 degrees, do the least squares fit, and rotate the solution back.– Errors then force the solution towards the 45 degree line

– Effect of errors on the slope is reduced


GPS Comparison Descriptions

• AIRS/GPS/radiosonde moisture comparisons– GPS gives only Integrated Precipitable Water (IPW)

– GPS observations are continuous – AIRS and radiosondes are not• Use AIRS time for AIRS

• Use RABO time for radiosondes

• Discard cases with large changes in GPS TPW between the two times

• Note – for the AIRS matches with GPS adjusted radiosondes, use the AIRS time for the GPS

– Note - This is equivalent adjusting the radiosonde at the radiosonde time, and then using the GPS to correct for the change in moisture with time


Figure 2. Scatter plots of total precipitable water for three instruments, AIRS, GPS, and radiosondes. The values in the upper left hand corner are least squares fit. The values in the bottom right define the estimate of the true slope. The line shown in the figures is the 45 degree line.


IPW Conclusions

• All three instruments compare well

• The GPS versus radiosonde has the best fit, but only by a very small amount (.961 versus .943 for AIRS/GPS and .947 for AIRS versus radiosonde).

• AIRS compares well to both


Summary and Conclusions

• When the GPS is used to validate AIRS in a three way AIRS, radiosonde, GPS comparison, the following can be observed– The differences between any two are roughly the same meaning each has

about the same error

• A GPS measurement at each radiosonde site can reduce calibration errors and produce a more uniform moisture effort for a combined network

• The equipment is relatively cheap

• GPS data by itself can be a valuable input to forecast models

use of gps to adjust radisondes: validation using airs

Documents

radiosonde time

radiosonde stations

radiosonde ipw

radiosonde comparisonuse

significant radiosonde

radiosonde sitemake

h radiosonde

radiosonde type