In addition to processing bathymetric data, my research on lakes requires sub-bottom profiling, or the collection of low-energy seismic data collected similarly to bathymetric data. The system we use is the SyQwest HydroBox, which, due to many hardware and software limitations, has been particularly frustrating. Still, when it does work properly, the HydroBox data has been invaluable to our lakes research. Unfortunately, the software supplied with the HydroBox is not particularly good at visualizing the collected data after collection.
A few months ago I needed to create detailed bathymetric maps for a water utility with which I was doing research. Our usual approach is to use an old Windows XP laptop with a serial port to log data from our Garmin GPS/Depth sounder unit. On the laptop, we used sofware called UnderSee Explorer (formerly Contour3D) which as far as I can tell, is now completely out of business. Usually the lakes we do research on are quite small, and the battery life of our laptops (around a few hours) is not an issue.
Inspired by the ability to create scientific tech gadgets, I have two new toys on my desk these days: a Raspberry Pi and an Arduino. The simplicity of the Arduino is quite nice, but the ability to code the Raspberry Pi in Python (not to mention built-in Wi-Fi capability) drew me to do some experimenting with the infrared sensor. The problem I ran into almost immediately is that the out-of-the box solution to infrared remotes is lirc, which requires compilation and dedicating a pin specifically for the IR sensor for all time.
The data frame was a concept I first came across in R, where it is a fundemental component of data analysis. Never having done much data analysis in Python, I came across a situation where I needed a data frame but didn’t know about the pandas implementation, so I went about writing my own DataFrame class. When I realized the pandas version existed, I immediately switched all my code over only to find that for my application, the pandas DataFrame was over 2 times slower than my lightweight DataFrame.
One thing that has always bothered me about Python is that there is no RStudio that exists as an IDE for scientific analysis in Python. There are attempts, such as Spyder and Canopy (among others), but the closest you can get is Jupyter, the latest iteration of IPython Notebook. I quite like Jupyter as a venue for analysis using Python, but I don’t like that it’s in a web browser. This is picky, I know, but my internet is my internet and my work is not, and I don’t like them in the same place.