MILITARY MEDICINE, J73. 7:677, 2008

Satellite Vegetation Index Data as a Tool to Forecast PopulationDynamics of Medically Important Mosquitoes at Military

Installations in the Continental United States

Seih C. Britch, PhD*; Kenneth J. Linthicum, PhD*; Assaf Anyamha, PhDf; Compton J. Tucker, PhDf;Edwin W. Pak, MScf; Francis A. Maioney, Jr., MScp, Kristin Cobb, MSc§; Erin Stanwix, MSc§;Jeri Humphries, MSc§; Alexandra Spring, MSc\\; Benedict Pagac, MSc\\; Melissa Miller, MSc\\

ABSTRACT The United States faces many existing and emerging mosquito-borne disease threats, such as West Nilevirus and Rift Valley fever. An important component of strategic prevention and control plans for these and othermosquito-borne diseases is forecasting the distribution, timing, and abundance of mosquito vector populations.Populations of many medically important mosquito species are closely tied to climate, and historical climate-populationassociations may be used to predict future population dynamics. Using 2003-2005 U.S. Army Center for HealthPromotion and Preventive Medicine mosquito surveillance data, we looked at populations of several known mosquitovectors of West Nile virus, as well as possible mosquito vectors of Rift Valley fever virus, at continental U.S. militaryinstallations. We compared population changes with concurrent patterns for a satellite-derived index of climate(normalized difference vegetation index ) and observed instances of population changes appearing to be direct responsesto climate. These preliminary findings are important first steps in developing an automated, climate-driven, earlywarning system to Hag regions of the United States at elevaled risk of mosquito-borne di.sease transmission.

INTRODUCTIONFaced with mosquito-borne diseases such as West Nile virus(WNV) and emerging threats such as Rift Valley fever (RVF)virus, which could arrive in the United States at any timealong a variety of pathways,'^ we should constantly improveour ability to predict the population dynamics ofthe mosquitovectors of these pathogens.' Predictions of changes in mos-quito populations would greatly enhance estimates of the riskof spread of mosquito-borne viruses.''"^ For instance, regionspredicted to have larger populations of competent vectorsduring outbreaks would likely experience more severe mos-quito-borne virus transmission. Classic range maps^ displaythe total region where a mosquito species might be found butare not designed to show long-term annual or seasonal trends,habitat patchiness, or spatial variations in population density.Military and civilian public health and vector-control agen-cies have surveillance systems to sample mosquitoes at thelocal level, which capture much of this heterogeneity,'*'̂ butthese data are not analyzed temporally on larger scales orused to forecast population abundances.'"

•U.S. Department of Agriculture, Agricultural Research Service, Centerfor Medical, Agricullural, and Veterinary Entomology. Gainesville, FL32608.

tNASA-Goddard Space Flight Center, Biospheric Sciences Branch,Greenbelt, MD 20771.

iU.S. Army Center for Health Promotion and Preventive Medici ne-West,Fort Lewis, WA 98433.

SU.S. Army Center for Health Promotion and Preventive Medicine-South, Fort McPherson, GA 30330.

IIU.S. Army Center for Health Promotion and Preventive Medicine-North, Fort George G. Meade, MD 20735.

This manuscripl was received for review in May 2007. The revisedmanuscript was accepted for puhhcation in April 2(K)8.

Many potentially confounding, varying factors may influ-ence measurements of mosquito population changes from onetime period to another. On one hand, an array of constantlyinteracting and changing abiotic environmental parameters,including a suite of climate effects, perpetually affect mos-quito population samples. On the other hand, human factorssuch as irrigation, application of pesticides, or particularsurveillance methods may variously affect our perception of amosquito population. Finally, population samples vary withthe outcomes of the bionomic properties of the mosquitospecies themselves, such as flight range, quality and avail-ability of breeding habitat and hosts, and presence and abun-dance of predators. Despite these challenges, a proven methodfor predicting mosquito population dynamics is to examineclimate-population relationships. In Africa, climate data mea-sured by satellites are used to predict conditions precedingproduction of large populations of mosquitoes and thus theearliest stages in a RVF epizootic." '- This effective earlywarning system in Africa was developed by looking at long-temi associations between outbreaks of RVF. the mosquitoesthat transmit RVF virus, and climate, in this report, ourobjective was to look at short-term climate-population asso-ciations for selected U.S. mosquito species, to assess thepotential for a similar in-depth analysis. We focused onpopulations being influenced by climate, not to discount theeffects of confounding factors but as a starting point inidentifying potential predictive factors. We explored relation-ships between a satellite climate index and mosquito popu-lations in selected regions of the United States, using mos-quito surveillance data gathered by the U.S. Army Center forHealth Promotion and Preventive Medicine (USACHPPM)."

MILITARY MEDICINE, Vol. 173, July 2008 677

Report Documentation Page Form ApprovedOMB No. 0704-0188

Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering andmaintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, ArlingtonVA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if itdoes not display a currently valid OMB control number.

1. REPORT DATE 2008 2. REPORT TYPE

3. DATES COVERED 00-00-2008 to 00-00-2008

4. TITLE AND SUBTITLE Satellite Vegetation Index Data as a Tool to Forecast PopulationDynamics of Medically Important Mosquitoes at Military Installations inthe Continental United States

5a. CONTRACT NUMBER

5b. GRANT NUMBER

5c. PROGRAM ELEMENT NUMBER

6. AUTHOR(S) 5d. PROJECT NUMBER

5e. TASK NUMBER

5f. WORK UNIT NUMBER

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) U.S. Army Center for Health Promotion and PreventiveMedicine-West,Fort Lewis,WA,98433

8. PERFORMING ORGANIZATIONREPORT NUMBER

9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)

11. SPONSOR/MONITOR’S REPORT NUMBER(S)

12. DISTRIBUTION/AVAILABILITY STATEMENT Approved for public release; distribution unlimited

13. SUPPLEMENTARY NOTES

14. ABSTRACT see report

15. SUBJECT TERMS

16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF ABSTRACT Same as

Report (SAR)

18. NUMBEROF PAGES

8

19a. NAME OFRESPONSIBLE PERSON

a. REPORT unclassified

b. ABSTRACT unclassified

c. THIS PAGE unclassified

Standard Form 298 (Rev. 8-98) Prescribed by ANSI Std Z39-18

NDV¡ and Mosquito Populations

In particular, we sought to identify instances of populationpatterns that suggested a response to climate. If such re-sponses exist, then we can build on those instances in ongo-ing analyses to determine whether similar current or futureclimate conditions lead to expected population abundances orwhether other factors need to be identified.

METHODSUSACHPPM mosquito surveillance data were availablefor 2003-2005 from three regional commands, that is.Center for Health Promotion and Preventive Medicine-North, Center for Health Promotion and Preventive Med-icine-South, and Center for Health Promotion and Preven-tive Medicine-West. The raw data were in an electronicspreadsheet format, divided according to year, and presentedthe number of adult female mosquitoes collected and trap-night values in daily, weekly, or monthly reports according toinstallation. There was variation among installations in sam-pling months, but sampling was generally performed inspring, summer, or autumn months. Collections were con-ducted by using Centers for Disease Control and Preventionlight traps with or without carbon dioxide. New Jersey lighttraps, gravid traps, or counterflow-geometT7 traps baited withoctenol. No installation sampled throughout the year, andapproximately one-third of installations submitted data in all3 years. Because the satellite climate data we used were in amonthly format, we compiled mosquito surveillance reportsand calculated a trap-night metric for each species in eachmonth at each installation. All mosquito catch numbers werelogarithmically transformed before analysis. Although manydozens of mosquito species were detected at USACHPPMinstallations, we limited analyses to small groups of speciesknown to be important in WNV transmission'"* and hypoth-esized to be important in RVF virus transmission in theUnited States'*^"'̂ (Table I).

We used a subset of the USACHPPM data to investigateclimate-population relationships across an operationally andecologically meaningful cross-section of the United States.

We chose four installations in Georgia, namely. Fort Ben-ning, Fort Stewart. Fort Gillem. and Fort McPherson, becausethis region could be particularly susceptible to the iu-rival ofemerging mosquito-bome pathogens. Factors contributing tothis susceptibility include nearly year-round mosquito cli-mate, high concentrations of seaports and airports, frequentmilitary mobilizations, and geographic proximity to islandnations outside the continental United States. We chose threeother installations, that is. Fort Riley. Fort Lewis, and YumaProving Ground, because they could provide insights regard-ing potential climate-population relationships in widely sep-arated regions (the Midwest plains, the Pacific Northwest,and the desert Southwest, respectively). Data from Fort Gil-lem and Fort McPherson were combined because the instal-lations are close (<20 km) and data were sometimes reportedas pooled values for the two installations.

We obtained monthly North American normalized differ-ence vegetation index (NDVI) satellite climate data sets for1981-2005 from the Goddard Space Flight Center."*-" TheNDVI measures the greenness of the earth, capturing in oneindex the combined effects of temperature, humidity, insola-tion, elevation, soils, land use. and precipitation on vegeta-tion. There is an almost-linear relationship between NDVIvalues and precipitation in semi-arid areas of Africa.-'-' andrelationships between NDVI values and increases in locust--'and mosquito -̂* populations in Kenya have been documented.In ecological terms, vegetation is the productivity on whichall animal life depends directly or indirectly and shouldcorrelate with temporal and spatial variations in mosquitopopulations. The NDVI is unitless. with a theoretical range of- 1 to -f I. but most values fall between 0 and 0.7. Positivevalues near 0 indicate bare soil with little or no vegetation,whereas values near 0.7 indicate dense vegetation. RawNDVI data are reported in the form of a matrix, or coverage,of measurements taken from 8-km squares of the earth'ssurface. To provide a view of the relative long-term magni-tude of NDVI, we also calculated the numerical difference

TABLE I. Focal Species Drawn from 2003-2005 USACHPPM Mosquito Surveillance Data

Mosquito Species

Aedes atbopictuxAedes ve.xansAnopheles cruciansCoquiHemdiii perturbansCuli seta melanuraCule.x erythrothoraxCulex pipiensCulex salinariusOchteroiatus soilicitansOchleroiaius iriseriatus

Typical Habitat

Tree holes; peridomesticFkHidwater poolsVegetated pools and pondsHeavily vegetated bodies of waterTree holes and swamp.sCattail marshes and pondsStagnant water poolsFresh or foul water poolsSalt marshesTree holes: peridomestic

WNV

XX

XXXXXXX

RVF Virus

X

XXX

Location

Fort Gillem/Fort McPhersonFort Lewis, Fort Riley. Fort Gillem/Fort McPhersonFort StewartFort StewartFort Gillem/Fori McPherson. Fort StewartYuma Proving GroundFort LewisFort Gillem/Fort McPherson, Fon BenningFort Gillem/Fort McPherson. Fort StewartFort Riley

Case

2

1,422

1.231312

Species were studied because of their status as either competent vectors for WNV in the United States or potentially compeient vectors for RVF vim.s inthe United States. It should be noted ihat most potential RVF virus vectors listed here are also vectors of WNV. Except for the bird-feeder Ciiti.seuimelanura. which is important in maintaining WNV in the wild, all species in this table feed on atid potentially transmit viruses to humans. Case refersto one of the four kinds of climate-population patterns observed in this study.

678 MILITARY MEDICINE, Vol. 173, July 2008

NDV! and Mosquito Populations

(i.e.. the anomaly value) between the observed NDVl and a25-year tnean NDV! for each month. For the anomaly NDVI,0 indicates no difference from the 25-year average for thatmonth, negative numbers indicate below-average greenness(dry periods), and positive numbers indicate above-averagegreenness (v̂ 'et periods).

With a combination of geocoded data from the Bureau ofTransportation Statistics'' and the U.S. Geological SurveyNational Atlas,-'' we mapped polygons of USACHPPM mil-itary installations in a geographic information system with allraw and anomaly North American NDVI coverages. Using^coprocessing tools in the geographic information system, wecalculated average monthly raw NDVI and anomaly NDVIvalues for each focal USACHPPM installation. Finally, weplotted monthly 2003-2005 raw NDVI and anomaly NDVIvalues with monthly population densities of the focal mos-

quito species for each installation and qualitatively assessedassociations of NDVI values with changes in mosquito pop-ulations.

RESULTSSelected plots of 2003-2005 USACHPPM mosquito surveil-lance data and NDVI data at the focal installations are givenin Figure 1. Overall, we observed several instances of mos-quito populations appearing to be influenced by variations inclimate, as measured with the NDVI. These instances can beplaced into the following groups: case 1, unusually dry peri-ods coinciding with the appearance of a mosquito species notcollected at any other time in the sample period (Fig. lA);case 2, unusually dry periods coinciding with populationsreaching unusually high densities (Fig. IB); case 3. unusuallywet periods, especially over late winter/early spring months.

V

Ft. Gillem/Ft. McPherson. GAOi^lerotaius satlicMansNOVI Anomaly

- - - NOVI Adua'

Sample Month

Ft. Lewis, WA

AMas vexansNDVl AnomalyNDVI Actual

! r

I I 1 I

i I

L^R

Sample Month

Ft. Gillem/Ft. McPherson. GA1 CulisetB melanura— • — NDVI Anomaly

I 'Sample Month

Ft. Stewart, GA

l/iH-

I 1 i — NOVI Anomaly— NOVI Actual

i i •

- -n

; î i i

Sample Month

M

FIGURE 1. Representative 2O()3-2(K)5 mosquito population data (biirs) superimposed on curves of raw NDVI (dashed lines) and anomaly NDVI (solidlines) dala lor seven local USACHPPM installations. Yearly and monthly breaks are shown with vertical dotted lines, and the NDVI zero-line provides quickreference lo monthly anomalies above or below the 25-year average greenness for that location. Thick horizontal lines below the .t-axis indícale the rangeof months sampled at that installation in a given year. Il should be noted that, at most installations, peaks and troughs in raw NDVI values generallycorresponded lo summer (wet and warm) and winter (cool and dry) months, respectively. For Yuma Proving Ground, the NDVI curve typically was flat,reflecting year-round dry conditions. At all installalions. anomaly NDVI values revealed fine-scale, month-to-month, climate effects that were hidden in thebroad seasonal patterns of raw NDVI data. A, Case I. B. Case 2. C. Case 3. D, Case 4.

MILITARY MEDICINE, Vol. 173. July 2008

NDVI and Mosquito Populations

B Ft. Lewis, WA

\'.:~.^\ CMxplplam '•— • — NDVI Anomaly \- - - NDVIAetual I

i l

i I I

Ft. Stewart, GA; ^

crucians* NDVI Anomaly

— — — NDVlAcluali i i

Sample Monthf K

Sample Month

Ft. Riley, 1̂

! i IHi

' —i Ochlûmmlus triserlBlua— — NDVI Anomaly- -« - NDVI Actual

VSI

: ' !

î

I •• !

- G .

.1

Ft. Stewart, GA

V

CuMssfa mBiamiNDVI AnomalyNDVI Actual

Sample Month Sample Monlh

Sample Month

FIGURE 1. (Continued).

coinciding with populations reaching unusually high densities(Fig. lC); case 4, populations appearing to rise and to fallsynchronously with alternating wet and dry periods (Fig. ID).There are variations in climate-population relationships ineach of these cases. For instance, in case 1, a dry month maycoincide with the month in which the rare species is observed,as with Culiseta melanura at Fort Gillem/Fort McPherson

0.1z

• a AS

Ft. Gillem/Ft. McPherson, GA

Sample Month

(Fig. 1 A), or the dry month may precede months in which thespecies is first detected, as with Ochieroraius sollicitans at thesame location (Fig. lA). The anomalously dry periods at FortLewis and Fort Stewart both consisted of sustained drynessover three NDVI months (Fig. 1 A), and the appearance of therare species was sustained for more than one of those drymonths.

680 MILITARY MEDICINE, Vol. 173. July 2008

NDVI and Mosquito Populations

Ft. Gillem/Ft. McPherson. GA

Sample Month Sample Month

Yuma Proving Ground, AZL = — ^ 1 C i f to «rytticotttoran— . — NDVI Anomaly- - - NDVI Actual

I i

1! A

! M i i ' !

H I KSample Month

Ft. Gillem/Ft. McPherson, GA

— • — NDVIAnomaiy— • - NDVI Actual

I ¡ 1 h1. !

I i

Hii I \r

Ft. Riley, KS

• LI

• — NDVI Actual

\ Ml 1

t ;

l/i

Sample Month Sample Month

FIGURE 1. (Continued).

Case 2 is similar to case 1 except that the species appearsat some density in the same season in at least one other year.As with case 1, there are variations in the patterns for case 2.The majority of instances of case 2 involve either rare orabundant species that at least double in density during shortor sustained anomalously dry periods (Fig. IB). These in-stances include Culex pipiens at Fort Lewis, Ochlerotatus

MILITARY MEDICINE, Vol. 173. July 2008

triseriatus at Fort Riley, and Anopheles crucians, C. mela-nura, and Coquillettidia perturbans at Fort Stewart. Oneinstance of case 2, that of Aedes albopictus at Fort Gillem/Fort McPherson, shows a relationship in which populationsof moderate abundance increased to unusually high peaks inthe wet months immediately following large downswings inmoisture. For this species at this locatioti, the peaks appeared

est

NDVI and Mosquito Populations

to be associated with the patterns of moisture and greennessrather than a particular month.

For certain species in certain locations, the case 3 climate-population relationship may be the most worthy of furtherexploration for the development of forecasts of populationabundances, ln particular, excessive moisture and vegetationdevelopment late in the year appear to give rise to unusualpopulation abundances the following spring. The most strik-ing example is Culex erythrothorax at Yuma Proving Ground(Fig. lC).

Case 4 demonstrates that populations of at least one speciesmay respond to climate in comparable ways in two ecologicallydifferent regions. Increases and decreases in populations ofAedes vexan.s at kith Fort Gillem/Fort McPherson and FortRiley appeared to be synchronized with positive and negativechanges in moisture and vegetation (Fig. 10). The synchro-nization is easily recognized for Aedes vexans at Fort Gillem/Foit McPherson where population changes closely trackanomaly NDVI variations. However, the pattern is less ob-vious at Fort Riley. Here, a relatively wet May in 2004followed by nomial and above-normal moisture into Septem-ber coincided with very high population numbers, whereas anormal spring and early summer followed by late summerdryness in 2003 led to smaller populations. Importantly, in allof these ecologically diverse installations, a sudden changefrom dry to wet corresponds to rapid increases in Aedesvexim.s trap counts, as observed June-July 2004 at FortGillem/Fort McPherson and May-June 2005 at Fort Riley.

DISCUSSIONThis first qualitative analysis of NDVI and 2003-2005USACHPPM mosquito surveillance data yields observa-tions that hold promise for future climate-based modelsdeveloped to forecast population dynamics of medicallyimportant mosquitoes. All mosquito species included in thisstudy either are known to be associated with important U.S.arboviruses such as WNV, Eastern equine encephalitis virus,or St. Louis encephalitis virus'"" or are strong candidates fortransmitting emerging threats such as RVF'-* '̂̂ or chikungu-nya' (Table I). Therefore, any ability to predict relative abun-dances of populations of tbese species in coming seasonswould be an important tool in allocating limited resources fordisease surveillance and control.

For future research, it Is important to emphasize that theseNDVi data are monthly summaries and data from finer tem-poral scales (I.e., 10-day and 15-day summaries) are alsoavailable. Although a dry montb may coincide with theappearance of a rare species (case I), the NDVI signal mayactually be visible late in the previous month or early in thecurrent month and so be useful operationally as a warningregarding potential mosquito-borne disease activity. Futurestudies should also consider mosquito bionomics, as outlinedin Table I. Populations of peridomestic container breederssuch as A. (dbopictus can theoretically peak during dry or wetperiods, reflecting the human activity of irrigation in dry

times (especially sustained dry times) but also reflecting thefact tbat rain can fill containers left around houses and otherbuildings.-^ In contrast, A. vexans is a floodwater mosquitowitb population densities expected to be closely tied to rain-fall, especially repeated fiooding, and thus positive NDVIanomalies. Depending on its timing and intensity, rainfall canlead to increased vegetation without flooding, but in someyears high NDVI values can indicate elevated A. ve.xanspopulations if there is flooding. At tbe peak of tbe growingseason, rainfall patterns can lead to frequent fiooding withoutconcurrent increases in vegetation, whicb may unlink NDVIvalues and population changes. For some Aedes species, too-frequent flooding may prevent habitat from drying enough foreggs to be conditioned for batching. In tbese situations, tbeelement of tbe NDVI that is most informative is not tbeanomaly value as much as tbe number of consecutive monthswitb anomaiou.sly bigh NDVI values. Many of the Cidexspecies in this study are expected to emerge at some pointafter rainfall and flooding of new or previously dry areas, andthey could emerge constantly in warm seasons without flood-ing if ponds or bodies of foul water do not dry out. Theirpatterns of emergence may not always track rainfall andNDVI values, because their babitat also includes bodies ofwater fiooded by human activity, such as containment ponds,storm drains, and peridomestic containers.

On its own, the basic analysis in tbis study would not beenough for operational planning, although the findings are animportant first step. For instance, our analyses highligbted thespecies in Table I because changes in their populations ap-peared to be related to climate; we did not report results forpopulations of several other medically important species thatdid not have such associations with NDVI, such as Ochlero-¡atus canadensis at Fort Benning and Fort Stewart and Ochle-rotatus taeniorhynchus at Fort Stewart and Yuma ProvingGround. This does not mean that those species are not relatedto climate but simpiy indicates tbat tbe 2003-2005 populationsamples from tbose locations were not informative witb re-spect to NDVi. Also, this survey of a range of installationshelped identify locations with promising climate-populationrelationships for a maximal number of species, such as FortGillem/Fort McPberson with five species across all casepatterns, compared witb Fort Riley witb two species witbonly two case patterns. With tbis information, we will narrowfuture efforts to key locations tbat can be developed assentinel installations, such as one of tbe bases in Georgia,especially because of tbe geographic and cultural propertiesoutlined above.

Mosquito surveillance at military installations should becontinued or even augmented, to improve our ability toforecast mosquito population cbanges and to build an auto-mated sentinel system for conditions favorable for mosquito-bome diseases.-^ When integrated into routine, mosquito-borne disease surveillance and control plans, tbese satelliteclimate-based population forecasts could contribute to strate-gic preparation of military installations if exotic mosquito-

682 MILITARY MEDICINE, Vol. 173, July 2008

NDVI and Mosquito Populations

borne pathogens such as RVF virus are detected in the UnitedStates. Because mosquito surveillance catalogs all sampledmosquito species, these analyses and derivative productscould inform strategies against potential vectors for an arrayof emerging or established mosquito-borne pathogens.

ACKNOWLEDGMENTSWe thank CPT Hee Kim. Dr, Todd Walker. Or, Ed Evuns. MAJ SteveRichards, LTC Sonya Schleich, MAJ Anihony Schuster, and LTC ThomasBiirroughs for helping us acquire the USACHPPM data sets. Support forS.C.B. was provided by ihe U.S. Depanment of Agriculture through Agri-cultural Research Service funding for the postdocioral research associateprogram.

REFERENCES1. House JA, Turell MJ. Mcbus CA; Rift Valley fever: present status and

risk to the Western Hemisphere. Ann NY Acad Sei 1992; 653: 233-42.2. Reiter P. Fontenille D, Paupy C: Acdes alhopiaus as an epidemic vector

of chikungunya virus: another emerging problem? Lancet Infect Dis2006; 6: 463-4.

3. Sandhu HS, Thomas C, Nsubuga P. White ME: A global network forearly warning and response to infectious diseases and bioterrorism:applied epidemiology and training programs. 2001. Am J Public Health2O()3: 93: 1640-2.

4. Gubler DJ: The global emergence/resurgence ()f arbtivira] diseases aspublic health problems. Arch Med Res 2002; 33: 330-42.

.1. Woodrutf RE. Guest CS, Garner MG. et al: Predicting Ross River virusepidemics from regional weather data. Epidemiology 2002; 13: 384-93.

6. Bealy BJ: Control of arbovims diseases: i.s the vector the weak link?Arch Virol Suppl 2005; 73-88.

7. Darsie RF. Ward RA: Identification and Geographical Distribution of theMosquitoes of North America. North of Mexico. Gainesville. FL. Uni-versity of Florida Press, 21X15.

8. Kline Dl.: Mosquito populalion surveillance techniques. Tech Bull Flor-ida Mosq Control Assoc 2006; 7: 2-8.

9. Xue R-D: Adult mosquito behavior in relation to operational mosquitocontrol programs. Tech Bull Florida Mosq Control Assoc 2006; 7:17-21.

10. Britch SC. Linthicum KJ: The value of long-term mosquito surveillancedata. Tech Bull Florida Mosq Control Assoc 2008: 8: 2-9.

I !. Linthicum KJ. Anyamba A. Tucker CJ. Ketley PW, Myers MF, PetersCJ: Climate and satellite indicators to forecast Rift Valley fever epidem-ics in Kenya, Science 1999; 285: 397-400.

12. Anyamba A. Chretien J-P. Small J, Tucker CJ. Linthicum KJ: Devel-oping global climate anomalies suggest potential disease risks for 2006-2007. Int J Health Geogr 2007; 5: 60.

13. U.S. Army Center for Health Promotion and Preventive Medicine:

Entomological Sciences Program, 2008. Available at http://chppm-www.apgea,army.mil/ento: accessed March II, 2008.

14. Turell MJ. Dohm DJ. Sardelis MR, O'Guinn ML, Andreadi.'i TG.Blow JA: An update on the potential of North American mosquitoes(Diptera: Culicidae) to transmit Wesi Nile virus. J Med Enlomol2005: 42: 57-62.

15. Turell MJ. Gargan T, Bailey CL: Culex pipiens as a vector of Rift Valleyfever. Am J Epidemiol 1982; 116: 556.

16. Turell MJ, Bailey CL, Beaman JR: Vector competence of a Houston, TXstrain of Aedes alhoplctus for Rifi Valley fever virus. J Am MosqControl Assoc 1988; 4: 94-6.

17. Gargan TP, Clark GG, Dohm DJ. Turell MJ, Bailey CL: Vector potentialof selected North American mosquito species for Rift Valley fever vims.Am J Trop Med Hyg 1988; 38: 440-6.

18. Pinzón J: Usitig HHT to successfully uncouple seasonal and inlerannualcomponents in remotely sensed data. Presented at the 6th World Mul-ticonference on Systematics, Cybernetics, and Informatics (SCI 2002),July 14-18, 2002. Orlando, FL.

19. Pinzón J, Brown ME, Tucker CJ: EMD correction of orbital driftartifacts in satellite dala siream. In: Hilbert-Huang Transform and UsApplications, pp 167-83. Edited by Huang NE. Shen SSP. Hackensack.NJ. World ScienliHc. 2005.

20. Tucker CJ. Pinzón JE. Brown ME. el al: An extended AVHRR 8-kmNDVI data set compatible with MODIS and SPOT vegetation NDVIdata. Int J Remote Sens 2005; 26: 4485-98.

21. Nicholson SE. Davenpon ML. Malo AR: A comparison of the vegeta-tion response to rainfall in the Sahel and east Africa, using normalizedditterence vegetation index from NOAA AVHRR. Ctini Change 1990;17: 209-41.

22. Malo AR. Nicholson SE: A study of rainfall and vegetation dynamics inthe African Salie! using normalized difference vegetation index. J AridEnviron 1990: 19: 1-24.

23. Hielkema JU, Roffey J, Tucker CJ: Assessment of ecological conditionsassociated with the 1980/81 Desert Locust plague upsurge in WestAfrica using environmental satellite data. Ini J Remote Sens 1986; 7:1609-22.

24. Linthicum KJ. Bailey CL, Davie.s FG. Tucker CJ: Detection of RiftValley fever viral activity in Kenya by satellite remote-sensing imagery.Science 1987; 235: 1656^9.

25. Bureau of Transportation Statistics: Geographic Information Program.2(X)8. Available at http://www.bts.gov/programs/geographic_iiiformation-services; accessed March 11, 2008.

26. U.S. Geological Survey: National Atlas, 2008. Available at http://www-atlas,usgs.gov; accessed March It. 2008.

27. Britch SC. Linthicum KJ. Anyamba A. Tucker CJ. Pak EW. MosquitoSurveillance Team: Long-term surveillance data and patterns of invasionby Aedes albopictus in Florida. J Am Mosq Control Assoc 2008; 24:115-20.

28. Linthicum KJ. Anyamba A. Britch SC. et al: A Rifl Valley fever risksurveillance system for Africa using remotely sensed data: putential foruse on other continents. Vet Ital 2007; 43: 663-74.

MILITARY MEDICINE, Vol. 173, July 2008 683