WEEK 9

I preferred to use the Inverse Distance Weighting and Kriging methods in comparing this season’s rainfall to normal precipitation amounts. I decided not to use splining, as that would be better in predicting thermal values, such as temperature, more than kinetic values, such as rainfall. However, I felt kriging to be the most superior to all three, as it not only revealed values throughout Los Angeles County, but was a geostatistical meaurement also including accuracy of the measurements, relating to the "hard science" Theissen method used in hyrdologic science.

After converting the degrees, minutes, and seconds to decimal degrees and determining the difference between season normal and current season precipitations in Los Angeles County, I converted the Excel file to a .dbf, then projected the .dbf as a shapefile onto the polygon of Los Angeles County, both in the North American Datum 1983 Geographic Coordinate System. I then used spatial analyst to create kriging and IDW models of the season to date, normal, and differing precipitation amounts of the three models.

Overall, the season normal was larger than the season to date. This is not a very accurate depiction of data, due to the fact that the 2009-2010 season has not concluded, and this season will most likely have higher rainfall records. This weekend’s storm has already raised season to date amounts higher than that of my models presented, which were created the Thursday before the storm. I feel that ordinary kriging provided the best representation because it is a visual depiction of the hydrologically trusted Theissen method, relating the variance of rain gage amounts to the distance between each gage. In both kriging and IDW, I expanded the amount of points from the default (12) to 20, in order to cover the entire county boundary. Through kriging, it is evident that there are increased precipitation differences from the center moving northeast in Los Angeles County. The kriging model depicts the trends more acutely through the geometric polygons, as opposed to the more vague, circular IDW model with increased precipitation differences from the county center moving southwards. I chose the IDW method, assuming that all of the gages posted would provide rainfall recordings. I felt that the amount of gages provided was dense enough, but that is subjective and could be considered too small a network for accurate depictions to another geographer. However, I feel that overall, the kriging method was most accurate and presented the best spatial depiction of the data.


SOURCES:

Earls, Julie. http://proceedings.esri.com/library/userconf/proc07/papers/papers/pap_1451.pdf

Shi, Yunfei. http://spiedl.aip.org/getpdf/servlet/GetPDFServlet?filetype=pdf&id=PSISDG00675300000167531I000001&idtype=cvips&prog=normal