Good decision making should be based on good scientific advice. It is clearly the job of the scientific community to support decision makers whenever they feel able to do so. Floods cause tremendous suffering, not only in terms of loss of life but in the immense disruption caused to lives, livelihoods and economies. Most residents of Southern Mexico have been affected in some way by floods in Tapachula (2005) and Villa Hermosa (2007). Fortunately these severe events caused comparatively few lives to be lost directly. However their consequences to the regional economy were extremely grave. In both cases politicians in the lower part of the watersheds cited deforestation as a major contributory factor.
Conservation of the native forests of Chiapas will not be achieved unless the full value of the services they provide is appreciated by society. Yet the linkages between flooding and deforestation are complex. When a study claims to provide new evidence regarding linkages between deforestation and flooding it must be considered with some care.
Comments on the scientific value of the paper I cited in the previous post should be based only on evidence. I therefore checked most of the websites the authors used to derive their information before analyzing the paper. An important source was the world atlas of flood events at Dartmouth College
The data found there can be mapped directly using their web site as shown below.
I also downloaded and reformatted the raw data and moved it into postgis. It can be used as a shapefile through the zipped file here floodevents(change-extension-to-zip).doc
The data is clearly of variable quality. This is par for the course. Prior to 2006 the geographical coordinates were not entered explicitly in this particular data base. All the original raw data from the Dartmouth data (that I downloaded in Excel format) including textual descriptions of the floods can be loaded strait into R using read.table by changing the extension of the following file to txt flood_events.doc. Here are some quick (Qgis) maps derived from the data showing loss of life from the floods that count with coordinates.
There are various points to note before I look at the statistical analysis used in the article in greater detail. Southern and Eastern Asia seems to be particularly flood prone. Most floods on this continent occur around coastlines, on the lowlying floodplains and in the Himalayan watersheds. Africa has comparatively few floods. Flooding in North and Central America tends to be associated with the effects of tropical depressions and hurricanes.
The patterns of flooding at this scale seem to be determined largely by global circulation patterns, but the consequences and probability of being reported will be determined by settlement patterns and economic activity. Floods are rarely reported from the Amazon basin, even though much of the low lying areas of forest (várzea) are seasonally flooded and these floods vary in severity tremendously between years. These sort of details interest forest ecologists trying to trace tree distribution patterns but never make news because they do not affect human lives.