Requirements for the Global Climatology Analysis Tool

Climate study of cotton growing within the CONUS

Francois Vandenberghe, David Hahn, Ming Ge, Tom Warner and Scott Swerdlin

Research Applications laboratory

NCAR, PO Box 3000, Boulder CO 80307

Final Report with verification

1. Introduction

Monsanto, a large US company specializing in seeds, genomics, and agricultural productivity, wished to obtain seasonal forecasts in all US counties where cotton is grown. More precisely, Monsanto is interested in daily forecasts of precipitation, temperature minima and maxima and relative humidity at the locations of 112 meteorological stations spread over the southern US (see locations on figure 1) for the period 1 May – 31 Aug 2005. Monsanto furthermore required that a database of these surface fields be created in order to easily perform regional queries, and generate frequency plots.

Because local effects due to terrain (drainage flow, etc.) and water bodies (see breeze, etc.) can be preponderant in defining the local weather on the time scale requested by Monsanto; high-resolution, or meso-scale, models are best to produce regional forecast and/or climatology. Meso-scale models usually work with increasing-resolution embedded domains. Each domain providing initial conditions and boundary forcing to the one it contains. Low resolution data from a global model, such at those run at the National Centers for Environmental Prediction are used to initialize and constrain on the boundaries the outer domain of the meso-scale model. This constant forcing of the global data through the boundaries ensures a consistency across the scales, while small scale features develop. Figure 1 shows such a configuration of two embedded domains that could be used to downscale the NCEP Global Forecast System (GFS) 1 degree (~100km) global forecast to 30km (outer domain over the US) and 10km (smaller domain over the cotton growing region) gridded forecasts. Additional higher resolution (down to 1km) domains could be added locally to furthermore downscale global forecasts. More on meso-scale models and how they are currently used to generate high-resolution real-time forecasts and/or climatology for various applications can be found at http://www.rap.ucar.edu/projects.

Figure 1:

Locations of the 112 stations where daily temperature, humidity and precipitation forecasts were requested.

The white rectangles indicate potential computational domains for a meso-scale model

NCEP GFS products, however, do not extent beyond 180 hours and cannot be used for seasonal climate studies. NCEP offers other products specifically designed for climate applications. Those products are comprised of forecasts of temperature, humidity and accumulated precipitation monthly means. One data point per month is insufficient to drive a high-resolution meso-scale model as there is no representation of the diurnal cycle. Other seasonal forecast products exits, but they were either not publicly available or couldn’t made available within the time frame of the project. Consequently, a hybrid downscaling approach, combining NCEP seasonal forecasts and climatology of observed data, was used to generate daily minima and maxima of temperature and relative humidity with accumulated precipitations. The methodology is described in section 2. Results for the 112 stations are presented, in the form of a point and click map, in section 3. Forecasts for two selected sites representative of dry and wet weather are extracted, presented and commented on section 4. Conclusions are gathered in section 5.

2. Methodology

Several seasonal forecast products are available through different organisms. NCEP routinely produces seasonal forecasts and made publicly available the following products:

- Coupled Forecast System (CFS): Global monthly means forecast at 2.5 degree (~200km) resolution, 9 months ahead,

- Regional Spectral Model (RSM): Monthly means forecast at 50km resolution over the continental US, 4 months ahead.

RSM regional products are already downscaled from CFS seasonal forecasts and would have been very appropriate for the study. Unfortunately, at the time of the project, January 2005, RSM forecasts were not available for the period of interests (May-August 2005). Other products such as the global 12 months forecasts at 1 degree resolution can be obtained from the European Center for Medium-range Weather Forecasting (ECMWF) center. Those products, however, are not free and couldn’t be ordered and processed within the time frame of the project. Similarly, time was too short to be able to use the NCAR Community Climate System Model to generate seasonal forecasts useful for the study. CFS monthly means were, therefore, the only seasonal forecast products that could be used in this study.

The top panel of figure 2 shows the CFS mean 2 meter temperature forecast for the month of July 2005 produced in January 2005. For comparisons, we averaged the 6-hourly re-analysis data of last year NCEP 2-meter temperature forecast (NNRP products: NCAR/NCEP Reanalysis Project, http://dss.ucar.edu/datasets/ds090.0/) over the full month of July 2004 and plotted on the bottom panel of figure 2. Since both products, CFS and NNRP are on the same grid (T62: 2.5-degree, about 200km), we computed the difference and plotted the temperature trend on figure 3.

Figure 2:

Top panel: CFS monthly mean temperature 6 month forecast for July 05.

Bottom panel: NNRP 6-hourly temperature analyses averaged over the month of July 04.

Figure 3:

Difference 2005 CFS forecast minus 2004 NNRP mean temperature for July.

In order to add to the smooth CFS data the natural spatial and temporal variability that can be expected from a collection of daily meteorological measurements over a region as vast as the cotton belt, climatology information from real observations has to be used. Our approach was, therefore, to combine CFS mean forecasts with the natural variability that can be found in observations from historical database. Daily mean forecasts for May-August 2005 were derived by firstly interpolating in time the CFS monthly mean values between the 15^th of each month and secondly by interpolating in space the daily CFS gridded data at the station locations. A linear interpolation was used in both cases. A climatology of the natural diurnal variability that can be expected at the 112 stations locations was generated from the NCAR Automatic Data Processing observational database. This database contains all the surface observations collected by the weather service since 1975. For each station, the daily mean, minima and maxima were taken from the 3-hourly observational time series. The daily deviations, i.e. the maxima minus the mean and the minima minus the mean, were formed and added to the CFS daily local mean forecasts.

This hybrid method was applied to temperature and relative humidity only, 24 hour accumulated precipitations which do not follow diurnal cycles as marked as for temperature were solely derived from CFS forecasts. Because of time constraints, we were not able to use the cloud cover information available from both observation and model forecast to modulate the daily minima and maxima. This is obviously an improvement on the method that could be implemented in the future. For time reasons again, the observational climatology was built on the year 2004 only. To make sure that 2004 punctual local weather events were not included in the 2005 forecasts, a 3 day running mean filtering was applied to the local forecast time series of temperature and humidity. The diagram below describes the algorithm:

Figure 4:

Forecasts generation algorithm

To ensure that the strict deadlines of the project could be met, forecast products were generated and delivered to Monsanto in two phases. Purely model forecasts that result from the spatio-temporal interpolation of the CFS monthly means were firstly generated, reformatted into comma separated files and sent to Monsanto on January 27^th 2005. The final products with addition of observational information were sent to Monsanto in the same format on January 29^th, 2005.

3. Results for all stations (Hyperlinks)

Daily minima and maxima of temperature and relative humidity as well as 24hr accumulated precipitation forecasts were generated for the months of May to August 2005 according to the methodology described in section 2. Plots of the forecasts, in the format of time series, can be accessed though internet from the point and click map on figure 5. Because the stations density can be high in certain regions where it may be hard to discern the station names on the map, we have provided the full list of stations in text format with additional links to graphics in the appendix. More on the format and units of the plots accessible through the hyperlinks can be found in section 4, where sample of the plots are presented.

Note that for 7 stations: Wichita Jabara Airport, Sedgwick, KS (KAAO), Cape Girardeau Municipal Airport, Scott, MO (KCGI), Medicine Lodge, Barber, KS (KP28), Richmond International Airport, Henrico, VA (KRIC), Sacramento Executive Airport, Sacramento, CA (KSAC), Sacramento Metropolitan Airport, Sacramento, CA (KSMF) and Winfield Strother Field Airport, Cowley, KS (KWLD) observational data were missing from the NCAR ADP database and pure CFS forecast were used.

Figure 5:

Station locations with link to forecast data:

Select a site and press CTRL+click to access graphics of the forecast at site location.

4. Results for selected stations

As an illustration, we also present plots of the forecast data for Fayetteville regional airport, Cumberland, SC (KFAY latitude: 34.991N, longitude: 78.880W, elevation: 186feet/56m) and Chino airport San Bernardino, CA (KCNO latitude: 33.975N, longitude: 117.636W, elevation: 660feet/198m). These stations were chosen to represent dry and moist summer conditions. For each station we plotted the CFS forecast (left panel), the minima and maxima locally observed in 2004 during the same time period (center panel) and the forecasts according to the methodology described in section 2 (right panel). The later constitutes the products delivered to Monsanto. Temperatures are provided in degree Farenheit, relative humidity in % and accumulated precipitation in inches.

Figure 6:

Minima and maxima for temperature (red curves) and relative humidity (blue curves) with 24hr accumulated precipitation (green curves) at Fayetteville Airport, SC:

a) May-August 2005 from CFS monthly mean forecast (left panel),

b) May-August 2004 from daily observations (center panel),

c) May-August 2005 final forecast combining CFS 2005 mean forecast and 2004 observations diurnal cycle (right panel).

Figure 7:

Minima and maxima for temperature (red curves) and relative humidity (blue curves) with 24hr accumulated precipitation (green curves) at San Bernardino, CA:

a) May-August 2005 from CFS monthly mean forecast (left panel),

b) May-August 2004 from daily observations (center panel),

c) May-August 2005 final forecast combining CFS 2005 mean forecast and 2004 observations diurnal cycle (right panel).

6. Verification

The summer 2005 observations valid for the station became available in the NCAR archive by mid-September 2005. These observations were downloaded, the daily minima, maxima and means were computed for each station and compared to the forecasted data presented above. Comparisons are firstly visual, in which observed time series of daily temperature and relative humidity extrema were overlaid above the plots of the corresponding forecast. The mean difference (bias) and root mean square difference of daily extrema were also evaluated for each station over the period of May-August05. Those comparisons, however, were performed on temperature and relative humidity only. Precipitations were not verified, as in general bias and RMS are not good indicators of precipitation forecast skills.

Figure 8 provides a hyperlink to the verification plots for each station. The bias and RMS for both minima and maxima are given under each graph.

Figure 8: Station locations with link to verification data:

Select a site and press CTRL+click to access graphics of the forecast at site location.

As in the previous section, for illustration purpose, we also present plots of the verification data for Fayetteville regional airport, Cumberland, SC (KFAY latitude: 34.991N, longitude: 78.880W, elevation: 186feet/56m) and Chino airport San Bernardino, CA (latitude: 33.975N, longitude: 117.636W, elevation: 660feet/198m). These stations were chosen to represent dry and moist summer conditions. For each station we plotted the forecasts (plain line) and the summer 2005 observations (dotted lines) of temperature (upper panel) and relative humidity (lower panel) daily minima and maxima. Temperatures are provided in degree Farenheit, relative humidity in %

Figure 9:

Minima and maxima for temperature (red curves) and relative humidity (blue curves) for forecast (plain lines) and May-August 2005 observations (dotted lines) for Fayetteville regional airport, Cumberland, SC.

Figure 10:

Minima and maxima for temperature (red curves) and relative humidity (blue curves) for forecast (plain lines) and May-August 2005 observations (dotted lines) for Chino airport San Bernardino, CA.

Table 1 provides the mean bias and RMS calculated for the 4 month period (May-August 2005) and over the 112 stations.

7. Conclusion

Comparing 2005 CFS accumulated precipitation forecasts with the observed 2004 data, it is obvious that local daily forecasts solely based on CFS monthly mean precipitation forecasts lack of the natural variability that accompanies precipitation events. To represent this variability in the future, we envision to combine CFS forecast monthly value with stochastic modeling. The idea is that precipitation observation from the past years could be used to estimate the transition probability, i.e. the probability that it will rain two days consecutively. Using a Markov chain it is then possible to determine the day-to-day occurrence events of rain, while CFS monthly mean forecast could be used to the estimate the amount of precipitation.

Temperature and humidity plots clearly show the benefit of incorporating the observational variability due to the diurnal cycle for both temperature and relative humidity. It increases the range (distance between the minima and maxim curves) of temperature and moisture that can be expected on a daily basis. Temperature and precipitation forecasts verification shows a good agreement overall. It will be interesting to compare these verification results with NCEP medium range forecasts verification statistics to assess and rank the skill of the method.

In the future, NCEP is planning to make available RSM (50km) 7 month monthly mean forecasts. Although, these products still do not offer enough data to constraint a meso-scale model, they could be integrated into the framework described in this report to produce improved seasonal forecast on a regular and automatic basis.

8. Appendix: List of sites with hyperlinks to forecast data:

KMAI : MARIANNA MUNICIPAL AP, Jackson, FL, lat: 30.836N, lon: -85.184E, elev: 34m

KALI : ALICE INTERNATIONAL AP, Jim Wells, TX, lat: 27.741N, lon: -98.027E, elev: 52m

KATT : AUSTIN CAMP MABRY, Travis, TX, lat: 30.321N, lon: -97.760E, elev: 200m

KMAF : MIDLAND INTERNATIONAL AP, Midland, TX, lat: 31.943N, lon: -102.191E, elev: 872m

KMAE : MADERA MUNICIPAL AP, Madera, CA, lat: 36.988N, lon: -120.111E, elev: 77m

KJCT : JUNCTION KIMBLE COUNTY AP, Kimble, TX, lat: 30.511N, lon: -99.766E, elev: 533m

KPSX : PALACIOS MUNICIPAL AP, Matagorda, TX, lat: 28.725N, lon: -96.254E, elev: 3m

KP28 : MEDICINE LODGE ASOS, Barber, KS, lat: 37.284N, lon: -98.553E, elev: 467m

KDTS : DESTIN FORT WALTON BEACH AP, Okaloosa, FL, lat: 30.400N, lon: -86.472E, elev: 6m

KHSV : HUNTSVILLE INTL/JONES FIELD, Madison, AL, lat: 34.644N, lon: -86.786E, elev: 190m

KEET : ALABASTER SHELBY COUNTY AP, Shelby, AL, lat: 33.178N, lon: -86.782E, elev: 172m

KJBR : JONESBORO MUNICIPAL AP, Craighead, AR, lat: 35.831N, lon: -90.646E, elev: 79m

KTKI : MCKINNEY MUNICIPAL AP, Collin, TX, lat: 33.180N, lon: -96.590E, elev: 176m

KAEX : ALEXANDRIA INTERNATIONAL AP, Rapides, LA, lat: 31.335N, lon: -92.559E, elev: 25m

KCAE : COLUMBIA METRO ARPT, Lexington, SC, lat: 33.946N, lon: -81.122E, elev: 64m

KHDO : HONDO MUNICIPAL AP, Medina, TX, lat: 29.360N, lon: -99.174E, elev: 280m

KDTN : SHREVEPORT DOWNTOWN AP, Caddo, LA, lat: 32.543N, lon: -93.745E, elev: 54m

KRAL : RIVERSIDE MUNICIPAL AP, Riverside, CA, lat: 33.952N, lon: -117.439E, elev: 245m

KFAY : FAYETTEVILLE REGL AP GRANNIS F, Cumberland, NC, lat: 34.991N, lon: -78.880E, elev: 56m

KRIC : RICHMOND INTERNATIONAL AP, Henrico, VA, lat: 37.505N, lon: -77.320E, elev: 49m

KECG : ELIZABETH CITY COAST GUARD AIR, Pasquotank, NC, lat: 36.261N, lon: -76.175E, elev: 3m

KBLH : BLYTHE AIRPORT, Riverside, CA, lat: 33.619N, lon: -114.714E, elev: 120m

KCSM : CLINTON CLINTON-SHERMAN AP, Washita, OK, lat: 35.340N, lon: -99.200E, elev: 598m

KCSG : COLUMBUS METROPOLITAN ARPT, Muscogee, GA, lat: 32.516N, lon: -84.942E, elev: 119m

KLIT : LITTLE ROCK ADAMS FIELD, Pulaski, AR, lat: 34.727N, lon: -92.239E, elev: 78m

KICT : WICHITA MID-CONTINENT AP, Sedgwick, KS, lat: 37.655N, lon: -97.443E, elev: 402m

KSMF : SACRAMENTO METROPOLITAN AP, Sacramento, CA, lat: 38.696N, lon: -121.590E, elev: 7m

KDUG : DOUGLAS BISBEE-DOUGLAS INTL AP, Cochise, AZ, lat: 31.458N, lon: -109.606E, elev: 1251m

KLLQ : MONTICELLO MUNICIPAL AP, Drew, AR, lat: 33.638N, lon: -91.751E, elev: 82m

KSDL : SCOTTSDALE MUNI AP, Maricopa, AZ, lat: 33.623N, lon: -111.911E, elev: 448m

KESF : ALEXANDRIA ESLER REGIONAL AP, Rapides, LA, lat: 31.395N, lon: -92.296E, elev: 34m

KLAW : LAWTON MUNICIPAL AP, Comanche, OK, lat: 34.568N, lon: -98.416E, elev: 338m

KSJT : SAN ANGELO MATHIS FIELD, Tom Green, TX, lat: 31.351N, lon: -100.494E, elev: 583m

KAND : ANDERSON COUNTY AP, Anderson, SC, lat: 34.495N, lon: -82.709E, elev: 231m

KANB : ANNISTON METROPOLITAN AP, Calhoun, AL, lat: 33.587N, lon: -85.856E, elev: 181m

KSSF : SAN ANTONIO STINSON MUNI AP, Bexar, TX, lat: 29.339N, lon: -98.472E, elev: 175m

KMCE : MERCED MUNICIPAL AP, Merced, CA, lat: 37.285N, lon: -120.513E, elev: 46m

KHJO : HANFORD MUNICIPAL AP, Kings, CA, lat: 36.319N, lon: -119.629E, elev: 75m

KMCN : MACON MIDDLE GA REGIONAL AP, Bibb, GA, lat: 32.685N, lon: -83.653E, elev: 104m

KAGS : AUGUSTA BUSH FIELD, Richmond, GA, lat: 33.370N, lon: -81.965E, elev: 40m

KGWO : GREENWOOD LEFLORE ARPT, Carroll, MS, lat: 33.496N, lon: -90.087E, elev: 40m

KMKL : JACKSON MCKELLAR-SIPES REGL AP, Madison, TN, lat: 35.593N, lon: -88.917E, elev: 131m

KINK : WINK WINKLER CO AP, Winkler, TX, lat: 31.780N, lon: -103.201E, elev: 855m

KCGI : CAPE GIRARDEAU MUNICIPAL AP, Scott, MO, lat: 37.225N, lon: -89.571E, elev: 102m

KELP : EL PASO INTERNATIONAL AP, El Paso, TX, lat: 31.811N, lon: -106.376E, elev: 1194m

KBRO : BROWNSVILLE S PADRE ISL INTL A, Cameron, TX, lat: 25.914N, lon: -97.423E, elev: 7m

KHBR : HOBART MUNICIPAL AP, Kiowa, OK, lat: 34.989N, lon: -99.052E, elev: 474m

KAMG : ALMA BACON COUNTY AP, Bacon, GA, lat: 31.536N, lon: -82.507E, elev: 58m

KHKS : JACKSON HAWKINS FIELD, Hinds, MS, lat: 32.335N, lon: -90.222E, elev: 104m

KAUS : AUSTIN-BERGSTROM INTERNATIONAL, Travis, TX, lat: 30.183N, lon: -97.680E, elev: 146m

KLBB : LUBBOCK INTERNATIONAL AP, Lubbock, TX, lat: 33.666N, lon: -101.823E, elev: 991m

KWLD : WINFIELD STROTHER FIELD AP, Cowley, KS, lat: 37.168N, lon: -97.037E, elev: 350m

KLBX : ANGLETON/LAKE JACKSON BRAZORIA, Brazoria, TX, lat: 29.110N, lon: -95.462E, elev: 7m

KHKA : BLYTHEVILLE MUNICIPAL AP, Mississippi, AR, lat: 35.940N, lon: -89.831E, elev: 77m

KUZA : ROCK HILL YORK COUNTY AP, York, SC, lat: 34.987N, lon: -81.058E, elev: 203m

KTCL : TUSCALOOSA MUNICIPAL AP, Tuscaloosa, AL, lat: 33.212N, lon: -87.616E, elev: 51m

KLBT : LUMBERTON MUNICIPAL AP, Robeson, NC, lat: 34.610N, lon: -79.059E, elev: 38m

KVCT : VICTORIA REGIONAL AP, Victoria, TX, lat: 28.861N, lon: -96.930E, elev: 35m

KTRM : PALM SPRINGS THERMAL AP, Riverside, CA, lat: 33.628N, lon: -116.160E, elev: -34m

KCNM : CARLSBAD CAVERN CTY, Eddy, NM, lat: 32.337N, lon: -104.263E, elev: 985m

KHRL : HARLINGEN RIO GRANDE VALLEY IN, Cameron, TX, lat: 26.228N, lon: -97.654E, elev: 10m

KDCU : DECATUR PRYOR FIELD, Limestone, AL, lat: 34.653N, lon: -86.945E, elev: 180m

KSHV : SHREVEPORT REGIONAL ARPT, Caddo, LA, lat: 32.447N, lon: -93.824E, elev: 77m

KMEM : MEMPHIS INTERNATIONAL AP, Shelby, TN, lat: 35.056N, lon: -89.986E, elev: 77m

KPHX : PHOENIX SKY HARBOR INTL AP, Maricopa, AZ, lat: 33.428N, lon: -112.004E, elev: 337m

KFLO : FLORENCE REGIONAL AP, Florence, SC, lat: 34.194N, lon: -79.729E, elev: 44m

KRWI : ROCKY MOUNT WILSON AP, Nash, NC, lat: 35.855N, lon: -77.893E, elev: 48m

KAAO : WICHITA JABARA AP, Sedgwick, KS, lat: 37.750N, lon: -97.219E, elev: 430m

KGLH : GREENVILLE MID DELTA REGIONAL, Washington, MS, lat: 33.483N, lon: -90.986E, elev: 38m

KCEW : CRESTVIEW BOB SIKES AP, Okaloosa, FL, lat: 30.780N, lon: -86.522E, elev: 57m

KVLD : VALDOSTA REGIONAL AP, Lowndes, GA, lat: 30.782N, lon: -83.277E, elev: 60m

KTOI : TROY MUNICIPAL AP, Pike, AL, lat: 31.861N, lon: -86.012E, elev: 121m

KSPS : WICHITA FALLS MUNICIPAL ARPT, Wichita, TX, lat: 33.979N, lon: -98.493E, elev: 309m

KPIL : PORT ISABEL CAMERON COUNTY AP, Cameron, TX, lat: 26.166N, lon: -97.346E, elev: 5m

KMLU : MONROE REGIONAL AP, Ouachita, LA, lat: 32.516N, lon: -92.041E, elev: 24m

KMSL : MUSCLE SHOALS REGIONAL AP, Colbert, AL, lat: 34.744N, lon: -87.600E, elev: 164m

KEED : NEEDLES AP, San Bernardino, CA, lat: 34.767N, lon: -114.619E, elev: 271m

KIPL : IMPERIAL COUNTY AP, Imperial, CA, lat: 32.834N, lon: -115.579E, elev: -17m

KCDS : CHILDRESS MUNICIPAL AP, Childress, TX, lat: 34.427N, lon: -100.283E, elev: 594m

KIGM : KINGMAN AIRPORT, Mohave, AZ, lat: 35.258N, lon: -113.933E, elev: 1042m

KCLL : COLLEGE STATION EASTERWOOD FLD, Brazos, TX, lat: 30.589N, lon: -96.365E, elev: 92m

KEWN : NEW BERN CRAVEN CO REGL AP, Craven, NC, lat: 35.067N, lon: -77.050E, elev: 4m

KDHN : DOTHAN REGIONAL AP, Houston, AL, lat: 31.321N, lon: -85.450E, elev: 112m

KPNS : PENSACOLA REGIONAL AP, Escambia, FL, lat: 30.478N, lon: -87.187E, elev: 34m

KBFL : BAKERSFIELD MEADOWS FIELD, Kern, CA, lat: 35.434N, lon: -119.054E, elev: 149m

KCRP : CORPUS CHRISTI INTL ARPT, Nueces, TX, lat: 27.773N, lon: -97.513E, elev: 12m

KTUS : TUCSON INTERNATIONAL AP, Pima, AZ, lat: 32.131N, lon: -110.955E, elev: 776m

KJAN : JACKSON INTERNATIONAL AP, Rankin, MS, lat: 32.321N, lon: -90.078E, elev: 100m

KCRS : CORSICANA CAMPBELL FIELD, Navarro, TX, lat: 32.031N, lon: -96.399E, elev: 136m

KPNC : PONCA CITY MUNI AP, Kay, OK, lat: 36.737N, lon: -97.102E, elev: 304m

KCRE : NORTH MYRTLE BEACH GRAND STRAN, Horry, SC, lat: 33.816N, lon: -78.721E, elev: 10m

KOGB : ORANGEBURG MUNICIPAL AP, Orangeburg, SC, lat: 33.462N, lon: -80.858E, elev: 60m

KSGR : HOUSTON SUGARLAND MEM, Fort Bend, TX, lat: 29.622N, lon: -95.656E, elev: 24m

KGSO : GREENSBORO PIEDMONT TRIAD INTL, Guilford, NC, lat: 36.097N, lon: -79.944E, elev: 273m

KDMN : DEMING MUNICIPAL AP, Luna, NM, lat: 32.262N, lon: -107.721E, elev: 1310m

KCNO : CHINO AIRPORT, San Bernardino, CA, lat: 33.975N, lon: -117.636E, elev: 198m

KDAG : DAGGETT BARSTOW DAG AP, San Bernardino, CA, lat: 34.854N, lon: -116.786E, elev: 584m

KACT : WACO REGIONAL AP, Mclennan, TX, lat: 31.619N, lon: -97.228E, elev: 152m

KDVT : PHOENIX DEER VALLEY MUNI AP, Maricopa, AZ, lat: 33.688N, lon: -112.082E, elev: 443m

KSAC : SACRAMENTO EXECUTIVE ARPT, Sacramento, CA, lat: 38.507N, lon: -121.495E, elev: 4m

KMFE : MCALLEN MILLER INTL AP, Hidalgo, TX, lat: 26.184N, lon: -98.254E, elev: 30m

KLFK : LUFKIN ANGELINA CTY AP, Angelina, TX, lat: 31.236N, lon: -94.754E, elev: 87m

KTVR : TALLULAH VICKSBURG REGIONAL AP, Madison, LA, lat: 32.348N, lon: -91.030E, elev: 25m

KROW : ROSWELL INDUSTRIAL AIR PARK, Chaves, NM, lat: 33.307N, lon: -104.509E, elev: 1112m

KSAT : SAN ANTONIO INTL AP, Bexar, TX, lat: 29.533N, lon: -98.470E, elev: 246m

KONT : ONTARIO INTERNATIONAL AP, San Bernardino, CA, lat: 34.056N, lon: -117.600E, elev: 289m

KCUB : COLUMBIA OWENS DOWNTOWN AP, Richland, SC, lat: 33.971N, lon: -80.996E, elev: 64m

KEQY : MONROE AIRPORT, Union, NC, lat: 35.017N, lon: -80.621E, elev: 206m

KABI : ABILENE REGIONAL AP, Taylor, TX, lat: 32.411N, lon: -99.682E, elev: 545m

KBNA : NASHVILLE INTERNATIONAL AP, Davidson, TN, lat: 36.125N, lon: -86.676E, elev: 182m

KGZH : EVERGREEN MIDDLETON FIELD, Conecuh, AL, lat: 31.416N, lon: -87.044E, elev: 78m

KLVJ : HOUSTON CLOVER FIELD, Brazoria, TX, lat: 29.519N, lon: -95.242E, elev: 13m

9. Appendix B: List of sites with hyperlinks to verification data: