System temperatures in KVN experiment k11352b

PIMA implements a rather sophisticated algorithm for decomposing measured system temperature at a product of elevation dependence and time dependence of Tsys in zenith direction: Tsys = Tsys_zen(t)*m(e). Both Tsys_zen(t) and m(e) are modeled as a linear B-spline. This algorithm does not impose specific elevation dependence and determines it from observations. Since k11352b was a survey style experiment and the sources were observed at elevations from 10 to 88 degrees, decomposition on time dependence and elevation dependence is robust.

Here is result of determination of Tsys_zen(t) as a function of time. Green line is the model, blue dots are Tsys(t)/m(e), where m(e) is an elevation dependence of Tsys, normalized to be 1 at the zenith direction.

We see sudden jumps. It is improbable these are sudden changes in the atmosphere opacity, because that would have been accompanied with winds of hurricane scale that had not been observed.

KVN station measured tipping curves every hour. Using the estimates of the atmospheric opacity from these measurements, we can evaluate receiver temperature as Trec = Tsys - Tatm*(1.0 - e-p*m(e)), where p is opacity, and m(e) is the so-called wet mapping function that is the ratio of the path delay at a given elevation to the zenith direction, and Tatm is the effective atmosphere temperature.

TAMNA Tsys(raw) Opacity Trec
ULSAN Tsys(raw) Opacity Trec
YONSEI Tsys(raw) Opacity Trec

Analyzing these table, we see that

  1. Trec has jumps.

  2. These jumps have the same epochs as epochs of tipping curve measurements. Does anybody have ideas how to explain this?

  3. There was a jump in opacity at 40% at Yonsei station at the beginning of the experiment. This jump is spurious: Tsys better fits the model if to consider exclude that measurement of opacity and interpolate opacity between the prior and the next epochs.

Back to Leonid Petrov's discussion page.

Last update: 2012.07.08_23:41:53