Wednesday, June 16, 2010

Torturing Data and Helicopters - Multivariate L/X band Regression

If you torture the data long enough, it will confess - Roland Cease. I have been messing with some PALSAR and TerraSAR-X backscatter over pastures for a while now and only reasonable relationship between field data and backscatter seems to be in the soil roughness. There are lots of explanatory variables for the strength of the backscatter and 2 response variables in different polarizations in the SAR. The challenge is to relate the impact of the explanatory variables measured at the relatively few field sites to the observations in the radar.

There are some interesting instances of EM38 usage where the readings are high due to proximity to power lines and the readings are high in the SAR due to scattering from the same. These match ups have nothing to do with soil moisture but everything to do with electromagnetics.

The mostly carbon fibre quadcopter has sprouted lots of aluminium extensions and will have a sizeable RCS now. Hopefully it will have sufficient power to fly. Something that definitely does not have enough power to fly is my little contra-rotor with the spy camera and 9V battery attached.

I will have to try flying the heli without the large 9v battery and just with some button cells stacked up to provide the same. Now there are rechargeable button cells from Telong making it feasible to deploy the camera with same recharge cycles as the helicopter main battery.

No comments: