From: Michael Reynolds Date: 15/08/2006 09:39:30 GMT-04:00 To: Andrew Vogelmann , Mark Miller Cc: Michael Reynolds Subject: Very-narrow-slow-shadowband-radiometer All, I have been thinking about the requirements for the new type of radiometer we have discussed. And here are some initial thoughts. The FRSR (Fast-rotating shadowband radiometer) we have is a very simple device; very simple and very reliable. Basically, it attaches a shadowband to a constantly rotating DC motor and watches for the shadow to pass over the MFR head. We use high-speed ADC (analog-to-digital) conversion to collect a time series of about 250 readings (0.36 deg/measurement) while the arm crosses the upward hemisphere. The 250-point time series is then analyzed to locate the shadow dip, decide if it is a cleanly defined shadow, and save the sweep for post-analysis for AOD. So, the good news is that it would appear that all of the key software is in place to do the new job. We could work with you on the sweep processing to output the correct numbers. But, the down side is the 2 deg requirement. The MFR accepts photons through a diffuser "puck" and the puck diameter is 8.1mm. Our FRSR shadowband has a width of 19.2 mm at apex is 72.9 mm above the diffuser surface. The current FRSR shadowband rotation period is about 6.2 sec. The shadowband width, radius, and rotation period were selected as a best compromise for (1) intended shipboard use, (2) the electronic response time of the MFR head, and (3) minimum shadow at small solar zenith angles, and (4) ability to withstand severe winds, and (5) the time required to process a sweep and transmit the data to the data acquisition PC. But, this design has a huge ablation angle, as far as the new design. The ablation angle is about 15 deg for Ze=0 deg. The new design parameters are (correct me if I'm wrong): 1. Ablation angle about 2 deg. 2. Angles of interest are +/-15 deg around the sun's beam (shadow dip). 3. Resolution of 2 deg, but we might oversample and improve. 4. Operation on land. 5. Some human support is reasonable. We might be able to achieve our goals by the following plan: 1. A thinner shadowband with a larger radius. 2. A slower rotation period. 3. Post processing: Is there some clever correlation/spectral technique to de-convolve a time series with oversampling yet a somewhat larger shadowband? A note on shadowband dimensions: First, the width of the shadowband cannot be larger than the diameter of the diffuser puck. That's 8.1 mm. Come to think of it, what exactly is the ablation 'angle' when we are dealing with a collection surface of finite size and parallel rays from the sun. My coordinate system origin will be at the center of the diffuser surface. So, the absolute minimum radius of the shadowband will be R ~= width / theta = 9 * 180 / 2deg / pi = 258 mm (10.1 in.). Actually, we will want some play here, so if we aim for shadowband width = 18 mm radius = 500 mm (19.7 in.) ablation angle = 2.06 deg rotation speed = 18 sec (allows more processing time) The rotation speed can be the same because the time the shadow is over the head increases with increasing Ze. So, all-in-all, at first blush, it seems the FRSR can be modified in the following way. 1. Build a new mounting bracket that will hold the head 24 inches above the plate. 2. Design a double shadowband with one band as before and the second, with the same width and 180 deg to the original, with a radius of 500 mm (19.7 in.). 3. Re-write the TT8 software to accommodate the new configuration. 4. Re-write the Data Acquisition PC software (PRPRX.C).