Report for DACS Contract Corresponding author: 1

(Project Number 00123945) Jonathan F. Day 2

Florida Medical Entomol. Lab, 3

Running head: Arbovirus Surveillance Program 200 9th

St. SE, Vero Beach, 4

Florida 32962, USA 5

Phone: (772) 778-7200 x 132 6

Fax: (772) 778-7205 7

E-mail: jfda@ufl.edu 8

9

10

Implementation of a Comprehensive and Sustainable Arbovirus 11

Surveillance Program at the County Level—Year 1 Preliminary Final Report 12

13

14

15

16

Jonathan F. Day and Gregory K. Ross 17

University of Florida, IFAS, Florida Medical Entomology Laboratory, 200 9th

St. SE, Vero Beach, FL 18

32962 19

20

21

22

23

24

25

26

27

2

ABSTRACT 28

New technologies have come online recently that have the potential of enhancing local arboviral 29

surveillance efforts and the forecasting of arboviral transmission events. These technologies utilize a 30

host of environmental data including Doppler radar-based total precipitation and Keetch Byram Drought 31

Indexing (KBDI), both at a 4 km2 resolution and MOSIAC modeling at an 11 km2 resolution. These 32

three environmental models have yet to be integrated into a single local arboviral surveillance protocol. 33

During the 2014 arboviral transmission season 36 sentinel chicken seroconversions, three EEE equine 34

cases, and four West Nile human cases were reported by the Volusia County Mosquito Control (VCMC). 35

The majority of the seroconversions occurred between July and September. Seroconversions were 36

reported at all 13 sentinel chicken sites maintained by VCMC suggesting that arboviral transmission was 37

widespread throughout the county. Here we propose a two-year research project. During the first year an 38

in-depth analysis of arboviral transmission in Volusia County during 2014 will be conducted focusing 39

on the daily Doppler radar-based precipitation totals, KBDI, and MOSIAC model signatures at each of 40

the 13 sentinel chicken flocks where arboviral transmission was reported. Environmental signatures will 41

be identified that indicated or preceded the 2014 transmission events based on all available data. During 42

the second year we will track the same environmental signatures in real-time at each sentinel chicken 43

site searching for the arboviral transmission signatures identified in the 2014 data set. Additional data, 44

such as mosquito abundance, gravid indices, exit trap data, and bird counts will be integrated into the 45

program during year two. Predicted transmission events will be verified by tracking sentinel chicken 46

seroconversions. The strengths of individual surveillance technologies or various combinations of the 47

three technologies will be evaluated and recommendations made concerning the best combination of 48

methods for arboviral surveillance at the local level. The goal of this project is to establish a 49

comprehensive, sustainable, and effective arboviral surveillance program at the mosquito control district 50

level by using all available environmental and biological surveillance data. The established program will 51

be used as an arboviral surveillance model for other MCDs throughout Florida. 52

53

3

KEY WORDS 54

Arboviral surveillance, environmental triggers, County-level disease surveillance 55

INTRODUCTION 56

The objective of this study is to establish a comprehensive, sustainable, and effective arboviral 57

surveillance program at the mosquito control program level by using all available environmental and 58

biological surveillance data. Arboviral transmission surveillance currently relies on three main 59

components: vertebrate sentinels, mosquito abundance data (including mosquito pooling for viral 60

isolation), and human cases. When used together, these three methods can provide a clear concise 61

picture of the current state of arboviral activity if the data are interpreted correctly. However, when these 62

methods are used to predict arboviral transmission, especially at the epidemic scale, major problems 63

with the surveillance methods becomes apparent. Sentinel chickens are unevenly distributed in space 64

and when seroconversions occur they provide little lead time for an operational mosquito control 65

response. Mosquito pooling, suffers from the same issues and is also expensive and labor intensive. 66

Human surveillance provides the data too late for mosquito control and health departments to have an 67

impact on virus transmission. 68

Historically, spatial and temporal arboviral surveillance results have been reported on a large 69

scale. For example, CDC Arbonet results are reported at the state level for West Nile positive sentinels, 70

wild birds, horses, and humans. To date, there has been little effort to integrate environmental factors 71

(rainfall and temperature) with the biological factors (mosquito abundance and age structure and 72

amplification host 73

abundance and age structure) that drive arboviral amplification on a county-level spatial scale. 74

The relationship between rainfall and vector abundance, reproduction, and dispersal has been 75

well documented (Shaman et al. 2004a, 2004b, 2005). Two predictive models have been developed for 76

the seasonal prediction of epidemic scale arboviral transmission (Day and Shaman 2008). The primary 77

model was developed in 2006 and estimates the real-time risk of SLEV/WNV amplification and 78

transmission based on Modeled Water Table Depth (MWTD). This model identifies areas of interest 79

4

(AOIs) in peninsular Florida. A second FMEL model was introduced in 2009. This model is based on 80

the Keetch-Byram Drought Index (KBDI) and estimates the real-time risk of arboviral amplification and 81

transmission. The model identifies AOIs throughout Florida. Both models estimate the risk of epidemic 82

scale arboviral transmission based on environmental factors that have been shown to be driving forces of 83

arboviral amplification and transmission in Florida and in other parts of North America. Different 84

environmental input variables are used for the datasets of each model, as are different ecological 85

assumptions specific to the biology of each major arbovirus found in Florida. Both models also assume 86

that there are populations of competent mosquito vectors and susceptible avian amplification hosts in the 87

AOIs identified as high risk by the models. 88

There are currently no available models that estimate the risk of sporadic arboviral amplification 89

and transmission on a local scale. Here we propose to build such a model for use by mosquito control in 90

Volusia County, Florida to monitor and predict arboviral transmission risk. This model will integrate 91

Doppler rainfall, KBDI, and MOSAIC models for use at a 4 to 11 km2 resolution around the 12 existing 92

sentinel chicken flocks located within the county. We know from past modelling efforts that accurate 93

real-time acquisition of precipitation and soil moisture indices are of critical importance for the tracking 94

of arboviral vectors including Culiseta melanura, Culex erraticus, Culex nigripalpus and Mansonia spp. 95

populations. Doppler radar provides a source of real-time precipitation data on a local scale. The well-96

established relationship between rainfall, vector dispersal, and arboviral amplification allows us to use 97

Doppler radar and the Keetch-Byram Drought Index (KBDI) (Keetch and Byram 1968) to track standing 98

and surface water as predictors of mosquito movement, mosquito reproduction, viral amplification, and 99

disease transmission risk (Shaman and Day 2005, Shaman et al. 2002). We propose that tracking of both 100

Doppler Radar-based Precipitation (DRP) totals and KBDI based can be used to predict vector mosquito 101

population movements and reproduction. 102

103

104

105

5

106

RESULTS 107

During the first year, the process of an in-depth analysis of Volusia County arboviral 108

transmission events and environmental signatures has been started. During the 2014 arboviral 109

transmission season 33 sentinel chicken West Nile seroconversions and four West Nile human cases 110

were reported by the Volusia County Mosquito Control (VCMC) (Table 1). The majority of the 111

seroconversions occurred between July and September. Seroconversions were reported at all 12 sentinel 112

chicken sites maintained by VCMC suggesting that arboviral transmission was widespread throughout 113

the county (Figures 2 - 3). In addition to the sentinel chicken arboviral seroconversions there were four 114

reported West Nile human cases reported in the county. Exact spatial and temporal environmental data 115

has been collected for each of these 33 transmission events. For each transmission event we have 116

calculated Doppler Radar-Based Precipitation (DRP), KBDI, and MOSAIC soil-moisture model data for 117

the three weeks prior to the reported transmission event. The National Weather Service (NWS) DRP was 118

download at: http://water.weather.gov/ahps/ in daily and hourly temporal formats. The data are at a 119

spatial resolution of 4km2 and projected in the Hydrologic Rainfall Analysis Project (HRAP) grid 120

coordinate system available for all of the lower 48 United States. Data was formatted for inclusion into a 121

GIS and weekly Doppler averages were calculated for 33 sites in Volusia county (Figure 5). Sites were 122

selected using a 4km buffer around the 12 sentinel sites to ensure that relevant environmental data was 123

selected (Figure 1). 124

To calculate the KBDI, 24-hour rainfall totals and the maximum daily temperature are needed. 125

The Florida Department of Forestry (DOF) combines traditional rainfall observations with data derived 126

from the National Weather Service's WSR88D (NEXRAD) radar network to provide a detailed view of 127

rainfall across the state to calculate the KBDI. Temperature is routinely measured at a number of sites 128

across Florida and since temperature is continuous, interpolation to regions that lack measurements is 129

straightforward. The calculation of KBDI by the Florida DOF is performed through an automated 130

procedure and a surface map for Florida is created and published daily. We have coordinated with the 131

6

Florida DOF to create a web portal for the real-time daily download of raw KBDI values for Florida and 132

portions of southern Georgia and southern Alabama. The web portal is available on the Florida DOF 133

website at http://www.fl-dof.com where downloads can be made at any time. We have automated the 134

downloading of KBDI datasets is automated through the use of computer software code written and 135

maintained by personnel at the Florida Medical Entomology Laboratory in Vero Beach. Downloaded 136

datasets were formatted for inclusion into a GIS and weekly KBDI averages were calculated for 30 sites 137

in Volusia county (Figure 4). Sites were selected using a 4km buffer around the 12 sentinel sites to 138

ensure that relevant environmental data was selected. 139

Analysis of Doppler and KBDI data has been performed within a Geographical Information 140

System (GIS) developed and maintained at the FMEL. The downloaded datasets were converted to 141

different formats (GRID and Feature Class) to allow for final analysis. Conversion of datasets was 142

automated using the GIS and the output files will be maintained within an Environmental Systems 143

Research Institute (ESRI©) Geodatabase. Each Florida KBDI dataset consists of 11,679 data points with 144

a spatial resolution of 4.0 km2 and a temporal resolution of 24 hours. Each Florida Doppler dataset 145

consist of 14,930 data points with a spatial resolution of 4.0 km2 and a temporal resolution of 24 hours. 146

Mosaic model simulations are produced through the North American Land Data Assimilation 147

Systems (NLDAS) project-2, and are available in hourly time steps at 0.125 degree resolution from 1979 148

through the present (NLDAS 2010, Mitchell et al. 2004). At this resolution (approximately 13 km by 13 149

km), each grid cell resolves hydrologic conditions at a geographic scale matching the upper limit of the 150

flight range reported for Cx. tarsalis in California (Reisen et al. 1992) and represents an area in which 151

the vector mosquito population and WNV transmission dynamics are likely localized. The Mosaic 152

model uses three soil layers with thicknesses from top to bottom of 10, 30, and 160 cm, respectively, and 153

a uniform rooting depth of 40 cm. Water storage in each model column layer is the weighted average of 154

the water storage from the column tiles. During the first year, we have acquired two Mosaic model-155

simulated estimates of land surface wetness: root zone soil moisture (RZSM), which represents water 156

content in the top 40 cm of the soil column, and layer 1 soil moisture (L1SM), which represents water 157

7

content in the top 10 cm of the soil column. For the ongoing analysis, local hourly RZSM and L1SM 158

estimates are each temporally aggregated to weekly averages and compared with the spatiotemporal 159

distribution of arboviral (EEEV, HJV, WNV, and SLEV) seropositive sentinel chickens, human WN 160

cases, and equine EEE cases. The MOSIAC datasets are currently going through processing and 161

formatting in preparation of analysis. 162

During the reporting period, we have met with Volusia County Mosquito Control personnel on to 163

become familiar with the district’s surveillance program and to initiate a data transfer of necessary data. 164

Data received from the district included sentinel chicken flock locations, seroconversion data, mosquito 165

exit trap, and mosquito ovitrap data for the year 2014. Personnel at FMEL have acquired environmental 166

datasets for Volusia County that include land cover/land use, hourly Doppler precipitation, daily Keetch 167

Byram Drought Index (KBDI) data, and hourly MOSIAC modeling data for 2014. In addition to the 168

environmental arboviral drivers discussed above, the biological factors associated with each sentinel 169

chicken transmission event in 2014 have been acquired. Exit trap and Ovitrap date for the three weeks 170

prior to all 33 West Nile sentinel chicken seroconversion events have been acquired and inputted into a 171

spatio-temporal database for analysis. 172

Table 1. 173

SITE WN PERCENTAGE

DAYTONA 2 6.06%

DELTONA 7 5 15.15%

FAIRGROUNDS 5 15.15%

HONTOON 2 1 3.03%

LOUIS 2 6.06%

LPGA 3 9.09%

NEEDLES 3 9.09%

NEW SMYRNA 3 9.09%

OAK HILL GROVE

2 6.06%

ORMOND 2 6.06%

PIERSON 3 9.09%

TATER ROAD 2 6.06%

8

TOTAL 33

174

175

176

Acknowledgments 177

This project was funded by the Florida Legislature. We would like to acknowledge the following 178

organizations and people for these help with this project: Florida Department of Agriculture and 179

Consumer Services, Florida Department of Health, Timothy Hope, and Daniel Forsythe. 180

References Cited 181

Cosgrove, BA. 2003. Real-time and retrospective forcing in the North American Land Data Assimilation 182

System (NLDAS) project. Journal of Geophysical Research 108(D22): 8842, /2002JD003118. 183

Day, J. F. 1990. The use of sentinel chickens for arbovirus surveillance in Florida. J. Fla Anti-Mosquito 184

Assoc. 60 (2): 56-61. 185

Day, J. F. and A. L. Lewis. 1992. An integrated approach to SLE surveillance in Indian River County, 186

Florida. Fla J. Publ. Hlth 4: 12-16.Shaman, J., J.F. Day, and M. Stieglitz. 2003. St. Louis 187

encephalitis virus in wild birds during the 1990 south Florida epidemic: the importance of drought, 188

wetting conditions, and the emergence of Culex nigripalpus to arboviral amplification and 189

transmission. J. Med. Entomol. 40:547-554. 190

Day, J.F. and J. Shaman. 2008. Using hydrologic conditions to track the risk of focal and epidemic 191

arboviral transmission in peninsular Florida. Journal of Medical Entomology 45(3):458-469. 192

Florida Department of Health (FDOH). 2014. Surveillance and Control of Selected Arthropod-borne 193

Diseases in Florida. 194

Keetch, J.J and G.M. Byram. 1988. A drought index for forest fire control. Southeast Forest Experiment 195

Station, U.S.D.A. Forest Service Research Paper Number SE-38, 32 pp. 196

Koster, R. D. and M. J. Suarez. 1996. Energy and water balance calculations in the Mosaic LSM. NASA 197

Technical Memo. 9: 58. 104606. 198

9

Mesinger F., et al. 2006. North American Regional Reanalysis, Bull. Amer. Meteor. Soc., 87(3):343-199

360. 200

Mitchell K.E., et al. 2004. The multi-institution North American Land Data Assimilation System 201

(NLDAS): Utilizing multiple GCIP products and partner in a continental distributed hydrological 202

modeling system, J. Geophys. Res. 109: doi:10.1029/2003JD003823. 203

NLDAS-North American Land Data Assimilation Systems Project. 2010. NASA. Digital Media. 204

http://ldas.gsfc.nasa.gov/. 205

Reisen, W. K., Milby, M. M., and R. P. Meyer. 1992. Population dynamics of adult Culex mosquitoes 206

(Diptera: Culicidae) along the Kern River, Kern County, California, in 1990. J. Med. Entomol. 29: 207

531-543. 208

Seo, D.-J., 1999: Real-time estimation of rainfall fields using radar rainfall and rain gauge data. J. 209

Hydrol., 208, 37-52. 210

Seo, D.-J. and J. Breidenbach, 2002: Real-time correction of spatially nonuniform bias in radar rainfall 211

data using rain gauge measurements. J. Hydrometeor., 3, 93-111. 212

Seo, D.-J., J. Breidenbach, and E. Johnson, 1999: Real-time estimation of mean field bias in radar 213

rainfall data. J. Hydrol., 223, 131-147. 214

Shaman, J. and J.F. Day. 2005. Achieving real-time operational hydrologic monitoring and forecasting 215

of mosquito-borne disease transmission. Emerging Infectious Diseases 11:1343-1350. 216

Shaman, J., J.F. Day, and M. Stieglitz. 2002. Drought-induced amplification of St. Louis encephalitis 217

virus, Florida. Emerging Infectious Diseases 8:575-580. 218

Shaman, J., J.F. Day and M. Stieglitz. 2005. Drought-Induced amplification and epidemic transmission 219

of West Nile Virus in south Florida. Journal of Medical Entomology 42:134-141. 220

Shaman, J., J.F. Day, M. Stieglitz, S. Zebiak and M. Cane. 2004a. Seasonal forecast of St. Louis 221

encephalitis virus transmission, Florida. Emerging Infectious Diseases 10:802-809. 222

10

Shaman, J., J.F. Day and M. Stieglitz. 2004b. The spatial-temporal distribution of drought, wetting, and 223

human cases of St. Louis encephalitis in south-central Florida. American Journal of Tropical 224

Medicine and Hygiene 71:251-261. 225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

Figure 1. 245

246

247

248

11

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

Figure 2. 266

267

268

12

269

270

Figure 3. 271

272

273

274

275

276

277

278

279

Figure 4. 280

281

282

283

284

285

286

287

288

Figure 5. 289

290

291

292

293

294

13

295

296

297

Figure 6. 298

299

300

301

302

303

304

305

306

Figure 7. 307

308

309

310

311

312

313

314

Figure 8. 315

316

317

318

319

320

14

321

322

323

324

325

326

327

328

329

330

331

332

333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

356

Figure 9. 357