My thinking about it for five minutes response, from growing up in areas that experienced historic floods like the Rapid City flash flood ( https://rapidcityjournal.com/news/local/images-recall-the-horror-of-1972-flood/collection_15fc7120-eb34-5800-b351-eb95c515be7f.html ), numerous floods in North Dakota on the Sioux River and Red River, ( https://www.ndstudies.gov/gr8/content/unit-iii-waves-development-1861-1920/lesson-1-changing-landscapes/topic-5-blizzards-floods-and-drought/section-3-floods ) and the Missouri River Floods ( V ).

Specific to the United States, anecdotally, the flood effect are 'whole body' events except for major metropolitan areas that have a considerable area outside the flood zone, and even then, the downtown cores are historically located next to rivers - they are usually better protected, but even that fails like in New Orleans. incapacitated Folks don't live in census block 'bubbles', even if their house survives, their relatives and friends one's don't, and their work place may be submerged, the grocery store and transportation links broken.

In some zones, they finally stopped issuing insurance based on individual properties, and just required the whole town be moved. https://www.cnn.com/2019/07/17/us/valmeyer-flooding-climate-crisis-midwest/index.html

My bet is tha if you took a gridded population map and a gridded hydro layer of the flood plain, only in the very fringes would you see any distinctions by block or even tract ( again, in the US ).

Your approach to this issue is admirable. To assess the vulnerability of an area to flooding or any other natural hazard, the presence of a hazard in that area is essential. The necessary condition for causing damage in any area is the existence of hazard, and vulnerability alone could not cause damage. If there is a safe area around the area that is flooded, it means that there is a risk of flood in the surrounding safe area, but the flood hazard does not threaten the adjacent area, and in fact the most important cause of damage (flood=hazard) in that area does not have, but as you mentioned, other factors indirectly affect that region, and these factors can even affect the national level. Therefore, depending on our goal, the type of research will also vary.

Good luck

GWR is a good method to take into account heterogeneity at the local level. I have used GWR quite a bit since it provides local regression coefficients. Try Geoda program also developed by Prof Anselin.

Hi all,

Thanks for your comments and suggestions.

Yes, I already used GWR and GWLR to find local regression coefficients through Stata, GeoDa, and GWR4.0 software and found consistent estimates.

I found GWR4 software is very user friendly and provide estimates quicker than other software and the results recorded nicely to make them useful for mapping out estimates later. But, I had to create spatial weights using GeoDa first and then created a csv format dataset to work within the GWR4 software environment. You can download this software for free from the following website:

https://gwrtools.github.io/gwr4-downloads.html

This software saved my life neither ArcGIS nor GeoDa/Stata could run GWLR models effectively and efficiently. I strongly recommend GWR4 software to anyone who plans to incorporate Generalized Linear Models ( ie, logistic or poisson regression) in spatial analysis or geographical weighted regression analysis.

Gaurav Tripathi