Extended Abstract (324K)J11.7
The Validity of the Keetch/Byram Drought Index in the Hawaiian Islands
Klaus P. Dolling, University of Hawaii at Manoa, Honolulu, HI; and P. S. Chu and F. M. Fujioka
The validity of using the Keetch/Byram Drought Index (KBDI) in the Hawaiian Islands is explored using a number of statistical and exploratory data analysis methods. Incorporating daily maximum temperatures and daily rainfall amounts, the Keetch and Byram Drought Index describes the amount of soil moisture available. Measurements of soil moisture are essential for fighting wildfires.
The KBDI is compared to the standardized precipitation index (SPI). The SPI is presently employed by the National Weather Service Honolulu Office to monitor drought conditions. The SPI can be calculated as a total for different numbers of months. The Honolulu NWS Office uses values of the SPI for one, two, three, Six, twelve, eighteen, and twenty-four month periods. The three, six, twelve, eighteen, and twenty-four month values are claimed to be useful for wildfire threat. Significant correlations are found with the KBDI and the three and six month SPI.
The KBDI is also tested separately on each island against TAB (total acres burned). A composite of KBDI is constructed for each island. Using a Spearman Rank correlation, the islands of Maui and Hawaii are shown to have the highest significance levels. Further exploration of the data leads to an alternate approach for testing the index. Four reference stations are chosen (one for each island). These stations have the highest mean KBDI and are therefore representative of the driest areas on each island. Two batches of TAB are formed for each reference station (one for months above and one for months below the median value of monthly KBDI). A Wilcoxon rank sum test is than used to test the null hypothesis that the two sample populations are identical. The null hypothesis is rejected for each reference station at the .01 alpha level. Conditional probabilities are also investigated for intervals of KBDI and the corresponding TAB. It is found that as KBDI increases, so does the conditional probability of finding TAB values above the median value.
Joint Session 11, Fire and Drought Indices
Wednesday, 19 November 2003, 1:30 PM-4:30 PM
Previous paper Next paper
Recorded presentation