Print Friendly and PDF

Advanced Search Results Detail

Project ID: 11-3-1-12

Year: 2011

Date Started: 08/01/2011

Date Completed: 02/26/2013

Title: Improved Wildfire Prediction Using Remote Sensing Technology on Guinea Grasslands in Hawaii

Project Proposal Abstract: Recurring wildfires in landscapes dominated by nonnative, invasive grasses threaten adjacent ecosystems and developed areas. The invasive grass/wildfire cycle, a vicious cycle of frequent fire and nonnative grass invasion, and increased ignitions from anthropogenic activities have increased fire frequencies throughout the tropics, often with severe consequences for threatened biota and ecosystem goods and services. Wildfires in ecosystems now dominated by the nonnative guinea grass (Urochloa maxima), in particular, have become extremely problematic in the tropics. In order to simultaneously provide for effective fuels management and protection of remnant native species and ecosystems, models that accurately predict fire behavior (i.e. probability of ignition, rate of spread, fire intensity) are urgently needed. The proposed research directly relates to JFSP priority topic 'fuel management effectiveness and effects'. Fuel moisture content is a key driver of fire behavior that is needed to parameterize fire models. The method most commonly used for quantifying fuel moisture is simply to measure the proportion of fresh weight:dry weight of field-collected samples. However, this method is time and labor intensive and provides fuel moisture for only a snapshot in time. The development of methods to quickly and accurately predict fuel moisture is critical for improving fire and fuels management in Hawaii and other Pacific islands. The primary objective of the proposed work is to develop a transferable methodology with remotely sensed Terra-MODIS Enhanced Vegetation Index (EVI) data to accurately predict real-time, site-specific fuel moisture in invasive grasslands on the Island of Oahu. EVI is calculated from the red, blue, and infrared bands of satellite imagery, and gives an index of the "greenness" of vegetation, which is directly related to water content. Because these remotely sensed data are easily accessible and freely available online, it would be an extremely valuable tool for land managers to utilize 16-day EVI composites to predict time and site specific fuel moisture, which could then be used to greatly improve fire models and fire danger ratings for use in Hawaii and throughout the tropical Pacific. In a pilot study, I found a positive linear relationship between EVI 16 day composite images and live (r2= 0.69) and dead (r2=0.89) fuel moisture sampled weekly in situ on Oahu between March and July, 2009. These preliminary results suggest that there is strong potential for using this approach to improve real time fuel moisture prediction. However, further research will be important for developing this methodology. In particular, the proposed research will be critical for: (i) determining if our preliminary relationships hold over longer sampling periods and across multiple sites; and (ii) establishing simple algorithms to make the approach readily accessible and useable by a wide array of land managers. Bi-weekly fuel moisture data (Oct. 2009- Oct. 2010) have been collected in three guinea grass dominated sites on Oahu (Dillingham, Yokohama, Schofield). Terra-MODIS 250m 16-day composite EVI data will be obtained via NASA's WIST Data Download website for all dates corresponding to field sampling. A MODIS reprojection tool, which converts imagery from a raw sinusoidal (SIN) projection to a geographically corrected image, will be used and the reprojected image will be spatially correlated with collected fuel moisture data. In situ fuel moisture and remotely sensed data will then be analyzed with linear and nonlinear regression techniques to identify the most robust relationships between live and dead fuel moisture and EVI data. A simple algorithm will be developed to allow land managers to download EVI data and accurately predict real time fuel moisture. This approach has the potential to greatly improve predictive fire models, allowing real-time and more accurate estimates of fire behavior by land managers.

Principal Investigator: Creighton M. Litton

Agency/Organization: University of Hawaii-Manoa

Branch or Dept: Department of Natural Resources & Environmental Management

Other Project Collaborators




Branch or Dept

Agreements Contact

Yaa-Yin X Fong

University of Hawaii-Manoa

Office of Research Services

Budget Contact

Georgette X Sakumoto

University of Hawaii-Manoa

Office of Research Services

Co-Principal Investigator

Lisa M. Ellsworth

Oregon State University

Department of Fisheries & Wildlife

Project Locations

Fire Science Exchange Network









Other Federal Lands




State Lands

Project Deliverables

Final Report view or print

("Results presented in JFSP Final Reports may not have been peer-reviewed and should be interpreted as tentative until published in a peer-reviewed source.")

  ID Type Title
view or print   92 Ph.D. Dissertation Improved Wildfire Management in Megathyrsus maximus Dominated Ecosystems in Hawai’i (L.M. Ellsworth)
view or print   3369 Journal Article Applied Vegetation Science
view or print   5581 Computer Model/Software/Algorithm Protocol for Using MODIS Data for Fuel Moisture Prediction
view or print   5582 Photo Guinea Grass
view or print   5583 Photo Oahu EVI
view or print   5584 Photo Field Sampling in a Guinea Grass Dominated Landscape
view or print   6959 Poster Changes in Land Cover and Fire Risk Associated With Nonnative Grass Invasion in Hawaii
view or print   6960 Poster Changes in Land Cover and Fire Risk Associated With Nonnative Grass Invasion in Hawaii

Supporting Documents

There are no supporting documents available for this project.

Convert PDF documents to an html document using Adobe's online conversion tool.
Download Adobe Acrobat Reader