Development of Rapid Prototyping Capability to Evaluate Potential

Development of Rapid Prototyping Capability to Evaluate Potential Uses of NASA Research Products and Technologies to Estimate Distribution of Mold Spore Levels over Space and Time University of Mississippi Medical Center Science Systems and Applications, Inc. (SSAI) Mississippi State University

Data Relationship Diagram for Mold Spore Distribution Estimation Mold Spore Count NASA / NOAA Meteorological data Ground Meteorological data

Ground Monitoring Equipment and Accessories for Collecting Meteorological Data Logger • • • Basic weather station Silicon pyranometer sensor for solar radiation Soil moisture sensor Soil temperature sensor Leaf wetness sensor

Ground Monitoring Equipment and Accessories for Collecting Mold spores Mold Spore Traps (Burkard Manufacturing Co Ltd. , UK) • 7 -Day Recording Volumetric Spore Traps 12 v (solar charger operated) • 7 -Day Recording Volumetric Spore Traps 110 VAC (electricity operated) Collection Drum Air Intake

Types of Mold We Should Know About Cladosporium (1) The most commonly identified outdoor fungus, but it can easily enter into the house through the HVAC and other airflow entryways. Cladosporium also has an indoor species that grows on textiles, wood and other porous, damp areas. Both indoor and outdoor species are triggers for hay fever and asthma symptoms. Alternaria (2) A large spore mold that can deposit in the nose, mouth and upper respiratory tract causing an allergic response. Indoors, it is often found in carpets, textiles, house dust and potentially damp areas like window frames and showers. It can also be found in plant soil. Stachybotrys (3) Pronounced (stack-ee-BOT-ris), this is an especially toxic black mold that produces airborne toxins (mycotoxins) that can cause serious breathing difficulties, dizziness, flu-like symptoms and bleeding in the lungs. Stachybotrys requires excessive moisture to thrive (usually running water) and is a slimy black mold. Fortunately, stachybotrys is not found in homes as often as the other molds listed above. Aspergillus (4) Usually found in warmer climates in areas of water damage or extreme dampness. Aspergillus species are also commonly found in house dust. Many species produce mycotoxins which may be associated with disease in humans and some animals. Also found in building materials and in fall leaves and other decomposing matter like compost piles. Penicillium (5) A very common mold known to cause allergies, hay fever and asthma. Species may be found growing on wallpaper, wallpaper glue and decaying fabrics in water-damaged buildings or homes. It is also found in carpet and in interior fiberglass duct insulation. Some species can produce mycotoxins. Number in the parenthesis indicate the relative order of predominance of mold spores so far identified

End Users • Mississippi Asthma Coalition • Mississippi Poison Control Center • Mississippi State Department of Health

Field Data Collection

Ground Monitoring Station Locations

Metrological Data Logger Configuration Data is logged every 15 minutes (96 readings/day) Base units collect temperature, humidity, dew point, rainfall, wind direction and speed (average and max gust). One or more additional sensors used on all.

Metrological Data Logger

Metrological Data Logger In addition, data are obtained from NOAA/NWS Jackson International Airport site for comparison. Their barometric pressure used for standard pressure correction of daily collected air volume. Data are reduced to hourly and daily levels/avg. Hourly temp data used to match to NWS bp and integrate STP volumes for daily corrected totals.

Typical Station Installation

Problems Encountered and Solutions Originally informed power requirements for DC powered spore trap was 12 v at 150 m. A and systems configured around those spec’s. Energy analysis suggested 15 W (1 Amp) panel would supply needed power with 2 X safety factor in the worst of weather. 1 st problem. Since solar powered with deep cycle battery, voltage goes up and down with sun and hence the air pump motor and resulting flow rate and volume varied.

Problems Encountered and Solutions Solution to varying voltage was addition of regulator. Worked extremely well with tight regulation. Down side was extra current draw. With regulation, current draw jumped to 250 ma.

Problems Encountered and Solutions

Problems Encountered and Solutions When the short and cold winter days appeared, batteries started dropping and occasionally lost a cell. An extra battery was obtained and sites were lasting about 7 -10 days between rotating changes. We have learned deep cycle batteries are inadequate in regards to charging efficiency at about 70%, that is, they need almost 1. 5 amp hours charge in for every one out. Capacity also dropped at least 20% with cold weather.

Problems Encountered and Solutions: Higher efficiency dc/dc regulators obtained. Extra 5 w (1/3 amp) panel added to one stand. Longer days and warmer weather biggest help. Batteries now lasting at least a month and appear to be leveling off.

Problems Encountered and Solutions: Since loggers are recording battery and motor voltage on spare channels, flow at suboptimal voltage can be determined. We have obtained a good quality flow meter and the regulators are variable. Flow at different voltages can now be obtained and volumes corrected according.

Problems Encountered and Solutions “STP” Learned from Burkard that their supplied custom flowmeters are calibrated at 20 o. C and 760 mm Hg

Problems Encountered and Solutions “STP” Samplers were set to 10 l/min at 20 o. C and ~ 760 mm Hg Results in 14. 4 m 3 of air per day

Problems Encountered and Solutions “STP” With hourly temperature from loggers and barometric pressure from NWS, the volume of air is being corrected.

Problems Encountered and Solutions

Mold Spore Sampling, Count and Identification

• As of April 25, 2008, we have collected 1055 daily samples from 6 sites. • Since the tape attached to the rotating drum inside the Burkard air sampler has to be cut vertically for removal during the weekly collection, one of the days in a week is not a complete 24 hours. • We have elected not to count the spores on incomplete days (Wednesday). • Each 24 -hour section of tape is cut using a standardized grid and mounted on a glass slide for coverslip staining as specified in a standard protocol by the National Allergy Bureau (NAB). • With the exclusion of Wednesday from the daily samples, we have collected 908 samples for mold spore enumeration. • We have conducted identification of weekly predominant species including quantification of Alternaria species and qualitative assessment of the other frequently observed species. • To date, we have counted 188 samples from 6 sites. • There are 12 weeks of mold spore species identification data including at least one week per site.

We have already found all seven of the clinically relevant molds identified by the AAAAI Immunotherapy Committee • Cladosporium cladosporioides • Cladosporium herbarum • Alternaria alternata • Epicoccum nigrum • Helminthosporium • Aspergillus • Penicillium

Examples of Identified Mold Spores

Bipolaris, 1000 X

Cladosporium, 1000 X

Helicomyces, 1000 X

Arthrinium, 1000 x

Stachybotrys, 1000 X

Alternaria, 1000 X

Preliminary Analysis and Results

Regression Analyses • Regression analyses are being performed to investigate the strength of the relationship between measurements of spores/m 3, NDMI (Normalized Difference Moisture Index), and various weather-related variables. • The goals are to identify which variables play the largest role in predicting spores/m 3 and to develop a model for estimation. April 2008 35

MODIS NDMI Values • NDMI (Normalized Difference Moisture Index) – Similar to NDVI (Normalized Difference Vegetation Index) except SWIR and NIR reflectances are used instead of Red band NIR reflectances – NDMI = (NIR–SWIR) / (NIR+SWIR) • Time series of daily NDMI values – Generated using TSPT (Time Series Product Tool) and 500 m MODIS MOD 09 data (8 -day reflectance composites) – Extracted for the monitoring sites and used in the regression analysis MODIS: Moderate Resolution Imaging Spectroradiometer SWIR: shortwave infrared NIR: near-infrared April 2008 36

Remote Sensing TSPT (Time Series Product Tool) • • TSPT software custom-designed for NASA at Stennis Space Center Developed in MATLAB® Purpose: To rapidly create and display various MODIS or simulated VIIRS (Visible/Infrared Imager/Radiometer Suite) products as single-bandcombination time series, such as NDVI or NDMI images, for wide-area crop surveillance and other time-critical applications. Typical MODIS input datasets: – – • MOD 02 Planetary Reflectance (Swath) MOD 09 Surface Reflectance (Tile) MOD 13 Vegetation Indices MOD 43 Nadir BRDF-Adjusted Reflectances Output/display options: – Single time frame and multitemporal change images – Daily time series plots at a selected location – Temporally processed image videos • Features: – Noise removal and temporal processing techniques • • MODIS metadata is used to find and optionally to remove bad, cloudy, and suspect data. TSPT also filters out variance due to atmosphere, sensor geometry, etc. – Capability of fusing data from the MODIS instruments onboard the Aqua and Terra satellites, which nearly doubles the effective temporal resolution April 2008 37

TSPT – Data Flow TSPT Data Flow Inputs • Science Datasets • Sensor Zenith Angle • Cloud & Quality Data Inputs… • may be obtained from DAACs or other sources • may be subsetted and/or reprojected with tools such as the MODIS Reprojection Tool (http: //edcdaac. usgs. gov/landdaac/tools/ modis/index. asp) • may be simulated with Application Research Toolbox (Ross et al. , 2006) April 2008 Multiple, temporally processed images can be created from single sensors or from fused Aqua and Terra data. TSPT Internal Processing Create Daily Fused Product Compute Ideal NDVI, NDMI, or other product Process Temporally (Median Filter, Savitzky-Golay, etc. ) Create Geographic Gridded Data from MODIS Swath or Tile (GUI Only) Generate Visualization Products • 1 -D time series plots given a latitude, longitude, and date range • 2 -D images given a region and a date 38 • 3 -D time series videos given a region and a date range

TSPT – Output Example • Example time series: – MODIS NDVI time series for Mobile Bay area with filtering and cloud removal applied using the TSPT. April 2008 39

Methodology for Preliminary Regression Analysis • The preliminary regression analysis involved the following: – Weather data, mold spore counts, and NDMI values from 4 of the 6 sites: DWFP, Terry, UMC, and Flora – Timeframe: 11/3/2007– 12/4/2007, because of the availability of mold spore count data for those days – “Global” analysis as opposed to site-specific analysis – data from all 4 sites were included • Steps: – Compile dataset for analyses (including data for 4 of the 6 sites) • Weather data – Compute daily average, maximum, and minimum for each variable, and compute other values, such as: » 2 -day average max temperature (identified in literature review) » 7 -day cumulative rainfall (identified in literature review) » Hours of relative humidity >= 80% – Use only those variables that are common to all 4 sites • Mold spore count data (available for most days during the 11/3/07– 12/4/07 timeframe) • NDMI from MODIS data time series April 2008 – Remove data from days for which there was no mold count data (once every ~7 days within the 11/3/07– 12/4/07 timeframe) – Generate Pearson’s correlation coefficient, r, to show strength of relationship between spores/m 3 and each of the weather variables and NDMI – Select the top 4 weather variables based on “r” values to include with spores/m 3 and NDMI in the regression analysis 40 – Perform preliminary regression analyses (using Microsoft Excel)

Spores/m 3 vs. variable: r rh_gte 80 0. 4443 Avg. HMD 0. 4313 Min. HMD 0. 4139 NDMI -0. 3638 Max. HMD 0. 2744 Min. DEW 0. 2677 Max. DEW 0. 2596 rh_lt 50 -0. 2582 Avg. DEW 0. 2358 7 -day cumulative RNF 0. 2253 Min. TMP 0. 1952 Max. WND -0. 1545 Avg. RNF 0. 1533 Sum. RNF 0. 1533 Min. WNG 0. 1177 Min. WND 0. 1014 Avg. WND -0. 0732 Avg 2 -day max TMP -0. 0664 Max. RNF 0. 0585 Max. TMP -0. 0571 Avg. WNS 0. 0497 wrun_km 0. 0497 Min. WNS 0. 0456 Avg. TMP 0. 0333 Max. WNG -0. 0244 Avg. WNG 0. 0140 Max. WNS 0. 0098 Strength of relationship between Spores/m 3 and Variables – Pearson’s “r” April 2008 • • The table (left) shows Pearson correlation coefficients (r values) between daily values of spores/m 3 and each weather variable (for all 4 sites combined). Variables are sorted by r values. The following variables are listed: – Average (Avg), maximum (Max), and minimum (Min), and other daily values for: • • HMD = relative humidity (%), including – rh_gte 80 (Hours of relative humidity >= 80%) – rh_lte 50 (Hours of relative humidity < 50%) DEW = dew point (degrees C) RNF = rainfall (mm), including – Daily sum of values (Sum. RNF) – 7 -day cumulative rainfall TMP = air temperature (degrees C), including – Average 2 -day maximum temperature WND = wind direction (degrees) WNG = wind gust (km/hr) WNS = wind speed (km/hr) – wrun_km = wind run (km/day) – NDMI from MODIS data time series • The top 4 variables, one per measurement type/category (i. e. , humidity, temperature, etc. ), with the strongest/highest “r” values were chosen for the regression analysis. These variables are in bold and 41 are highlighted blue in the table.

Spores/m 3 vs. variable: Pearson’s r R 2 F-test overall p-value rh_gte 80 0. 4443 0. 1974 24. 5970 0. 000003 Avg. HMD 0. 4313 0. 1860 22. 8532 0. 000006 Min. HMD 0. 4139 0. 1714 20. 6786 0. 000015 NDMI -0. 3638 0. 1323 15. 2495 0. 000171 Max. HMD 0. 2744 0. 0753 8. 1403 0. 005262 Min. DEW 0. 2677 0. 0717 7. 7215 0. 006519 Max. DEW 0. 2596 0. 0674 7. 2281 0. 00841 rh_lt 50 -0. 2582 0. 0666 7. 1408 0. 008799 Avg. DEW 0. 2358 0. 0556 5. 8859 0. 017055 7 -day cumulative RNF 0. 2253 0. 0508 5. 3478 0. 022798 Min. TMP 0. 1952 0. 0381 3. 9600 0. 049324 Max. WND -0. 1545 0. 0239 2. 4446 0. 121089 Avg. RNF 0. 1533 0. 0235 2. 4070 0. 123961 Sum. RNF 0. 1533 0. 0235 2. 4070 0. 123961 Min. WNG 0. 1177 0. 0138 1. 4040 0. 23888 Min. WND 0. 1014 0. 0103 1. 0382 0. 310699 Avg. WND -0. 0732 0. 0054 0. 5383 0. 464855 Avg 2 -day max TMP -0. 0664 0. 0044 0. 4433 0. 507067 Max. RNF 0. 0585 0. 0034 0. 3430 0. 559422 Max. TMP -0. 0571 0. 0033 0. 3272 0. 568596 Avg. WNS 0. 0497 0. 0025 0. 2471 0. 620216 wrun_km 0. 0497 0. 0025 0. 2471 0. 620216 Min. WNS 0. 0456 0. 0021 0. 2083 0. 649093 Avg. TMP 0. 0333 0. 0011 0. 1111 0. 739706 Max. WNG -0. 0244 0. 0006 0. 0595 0. 807788 Avg. WNG 0. 0140 0. 0002 0. 0197 0. 888661 Max. WNS 0. 0098 0. 0001 0. 0096 0. 922145 Strength of relationship between Spores/m 3 and Variables – Pearson’s “r” • The table (left) shows Pearson correlation coefficients (r values), R 2, F-test, and pvalues between daily values of spores/m 3 and each weather variable (for all 4 sites combined). Variables are sorted by rvalues. • The following variables are listed: – Average (Avg), maximum (Max), and minimum (Min), and other daily values for: • HMD = relative humidity (%), including – rh_gte 80 (Hours of relative humidity >= 80%) – rh_lte 50 (Hours of relative humidity < 50%) • DEW = dew point (degrees C) • RNF = rainfall (mm), including – Daily sum of values (Sum. RNF) – 7 -day cumulative rainfall • TMP = air temperature (degrees C), including – Average 2 -day maximum temperature • WND = wind direction (degrees) • WNG = wind gust (km/hr) • WNS = wind speed (km/hr) – wrun_km = wind run (km/day) – NDMI from MODIS data time series • The top 4 variables, one per measurement type/category (i. e. , humidity, temperature, etc. ), with the strongest/highest “r” values were chosen for the regression analysis. These variables are in bold and are highlighted blue in the table.

Site-Specific Comparison of Spores and Predictive Variables • Pearson’s correlation coefficient, r, was computed using data from the 4 combined sites and then using site-specific data for comparison. Spores/m 3 vs. variable: r – 4 sites combined r – UMC r – Terry r – Flora r – DWFP NDMI -0. 3638 -0. 1158 -0. 0914 -0. 1298 -0. 1881 rh_gte 80 0. 4443 0. 7327 0. 3159 0. 2111 0. 5186 Min. TMP 0. 1952 0. 1177 0. 3371 0. 4149 0. 2431 7 -day RNF 0. 2253 0. 5061 -0. 0804 0. 5408 0. 4964 Min. DEW 0. 2677 0. 1793 0. 3200 0. 4974 0. 2755 April 2008 43

Preliminary Results of Regression Analysis • Sorted by R 2 • Number of observations = 102 *S = spores/m 3 NDMI = Normalized Difference Moisture Index RH = hours of relative humidity >= 80% RNF = 7 -day cumulative rainfall D = minimum dew point T = minimum air temperature Overall pvalue p-value p-value R Square F Signif. F Int. NDMI RH RNF min-TMP min. DEW NDMI-RH-RNF-T-D 0. 362 10. 87 2. 6 E-08 0. 88 6. 7 E-06 0. 0010 0. 025 0. 065 0. 23 NDMI-RH-RNF-T 0. 352 13. 16 1. 3 E-08 0. 16 1. 13 E-05 0. 0018 0. 036 0. 10 NDMI-RH-RNF 0. 334 16. 38 1. 0 E-08 0. 018 2. 97 E-05 5. 9 E-05 0. 082 NDMI-RH 0. 313 22. 55 8. 5 E-09 0. 012 9. 1 E-05 1. 62 E-06 4 wx 0. 211 6. 46 0. 00012 0. 53 0. 00085 0. 29 0. 41 0. 61

Plot of Actual vs. Predicted Spores/m 3 for 4 Sites R 2 = 0. 3615 y = -775. 4 + (-83179. 3*NDMI) + (984. 7*rh_gte 80) + (-500. 3*min. DEW) + (251. 0*7 day. RNF) + (1040. 5*min. TMP) Equation/model is from regression that yielded the highest R 2 value that used data from 4 combined sites. Each site is plotted individually here to see if any patterns or differences can be observed.

Observations about Preliminary Regression Results • Using the combined data for 4 sites, NDMI was involved in all of the topranking regressions based on R 2 values. • November (late Fall) is not known to be a month with high mold spore counts, so better results are expected with the Spring data • Literature for correlation studies of mold spore counts with weather variables typically show lower R 2 values (i. e. , ~0. 45 or less for specific spores). Therefore, the low R 2 values in this study (i. e. , R 2 = 0. 36) are not surprising, especially when considering the mold spore count data was from November. • Literature typically identifies temperature and moisture (rainfall, dew point, relative humidity) as variables that are highly correlated with mold spore counts. The month of November was relatively dry (total rainfall ~49 mm or ~1. 9 inches) and cool (average temp ~14 degrees C). 46 April 2008

References • • Bruno, A. A. , L. Pace, B. Tomassetti, E. Coppola, M. Verdecchia, G. Pacioni, and G. Visconti, 2007. Estimation of fungal spore concentrations associated to meteorological variables. Aerobiologia 23: 221 -228. Burge, H. A. , 2002. An update on pollen and fungal spore aerobiology. Current Reviews of Allergy and Clinical Immunology. Journal of Allergy and Clinical Immunology 110(4): 544 -552. Mitakakis, T. Z. , A. Clift, and P. A. Mc. Gee, 2001. The effect of local cropping activities and weather on the airborne concentration of allergenic Alternaria spores in rural Australia. Grana 40(4 -5): 230 -239. O’Hara, C. G. , R. Moorhead, D. Shaw, B. Shrestha, K. W. Ross, D. Prados, J. Russell, R. E. Ryan, 2006. Integrated use of tools and technologies for rapidly prototyping simulated data products of future NASA observing systems for evaluation in application of national importance. Eos Transactions AGU, 87(52), Fall Meeting Supplement, Abstract IN 32 A-05. (presentation, Session IN 32 A, Wednesday, Dec. 13, 11: 20) Prados, D. , R. E. Ryan, and K. W. Ross, 2006. Remote Sensing Time Series Product Tool. Eos Transactions AGU Fall Meeting Supplement, Abstract IN 33 B-1341. (poster) Ross, K. W. , J. Russell, and R. E. Ryan, 2006. Simulating Visible/Infrared Imager Radiometer Suite Normalized Difference Vegetation Index data using Hyperion and MODIS. Eos Transactions AGU Fall Meeting Supplement, Abstract IN 33 B-1340. (poster) Sabariego, S. , C. Diaz de la Guardia, and F. Alba, 2000. The effect of meteorological factors on the daily variation of airborne fungal spores in Granada (southern Spain). International Journal of Biometeorology 44: 1 -5. Troutt, C. , and E. Levetin, 2001. Correlation of spring spore concentrations and meteorological conditions in Tulsa, OK. International Journal of Biometeorology 45: 64 -74. 47 April 2008

Contact Fazlay S. Faruque Director of GIS and Remote Sensing The University of Mississippi Medical Center 2500 North State Street Jackson, MS 39216 -4505 Phone: 601 -984 -4993 E-mail: FFaruque@son. umsmed. edu 48 April 2008