# Random Full Scan

Fri, Nov 10, 2017## 1 Root Location

`/home/chepplew/projects/random_fs/`

## 2 Overview

Analysis of AIRS, CrIS and IASI random full scan observations and TOA calculations, for the purpose of comparing data sets from the three sensors and as part of larger effort to support climate trending and anomaly investigation.

First part is to evaluate observations and calcs over equal areas, including average, statistics and bias characteristics.

Preparation of visualization tools.

## 3 Data Availability

RTP random subsets for CrIS and AIRS for 2016, locations:

`/asl/rtp/rtp_cris_ccast_lowres/random_fs/2016/`

and

`/asl/rtp/rtp_airicrad_v6/2016/random_fs/`

## 4 Methods

### 4.1 RTP generation (over to Steven)

The RTP data are designed to be randomly distributed over the surface of the Earth, so that when viewed as a function of latitude circle, should resemble a cosine distribution over the domain -90 to 90 degrees. Shown in figure 1, this is not quite the case with a shoulder near 65-deg latitude.

Figure 1: 16-day populations as function of latitude

### 4.2 Analysis

#### 4.2.1 Equal area bins

The simplest approach is adopted for now, using a prescribed binning from University of St.Petersburgh, Russia (add ref). Of the options provided the one with the smallest bin size is used, covering the Earth surface with 406 bins of area 101.60 sq. degrees. The variation in area across the earth is less than 5 parts per million, as sown in figure 2

Figure 2: A test figure.

The actual bin population should be close to the total number of observations for the year divided by 406 equal area trapezoids. These are shown in figure 3, and reflect the non-cosine distribution of the original random sampling.

Figure 3: Random sample population per equal area bin. Legend delinieates latitude band center

A data file holds the prescription:

`.../random_fs/simple_equal_areas.txt`

and a script is used to load them into a usable Matlab structure:

`.../random_fs/load_simple_equal_areas.m`

#### 4.2.2 Load

Matlab scripts are provided to load the sensor data into Matlab structures from the
RTP files. Currently
there is one script per instrument, and the full year 2016 is loaded. Currently only
basic gelocation and single channel radiance are loaded.

`.../random_fs/load_{airs,cris,iasi}_random_fs.m`

#### 4.2.3 Binning

A single script is currently used, since the data structures provided by the load scripts
are identical, to bin the radiance observations and compute averages.

For visualization purposes the equal area binned data are interpolated onto a uniform
lat/lon grid (currently 10x10-deg).

Figure 4 shows the actual bin average values, for each latitude band, and the interpolated
values, separated into the north and south hemispheres. Note that any succeeding computations
involving differences etc, are done on the original equal area bins.

Figure 4: 2016 CrIS random averaged BT in each cell, equal area and interpolated regular bins.

Figure 5 shows an example of the mean \(900 cm^{-1}\) channel plotted on the regular grid.

Figure 5: CrIS Random averaged BT in equal area bins, interpolated for plotting.

#### 4.2.4 Statistics

Once both sensor observations are binned into equal areas, then mean differences and other statistics can be evaluated. Shown in figure 6, the bias of AIRS minus CrIS brightness temperature for the \(900 cm^{-1}\) window channel. For the purposs of plotting the results have been interpolated onto a regular grid.

Figure 6: AIRS minus CrIS Random averaged BT in equal area bins, interpolated for plotting.