COPIES ARE UNCONTROLLED PACKAGE CREATED BY Paterson, Helen 1/19/2015 2:01 PM SHEET PACKAGE DRAWING REPORT 1of 2 The reports and attachments generated with this sheet and listed below constitute the entire Package Drawing Report for: 2444347 These reports and attachments are and should be arranged in the following manner in order to complete the package: NOTE: Attachment file names in parenthesis are the file names as they are in Agile. PACKAGE CONTENTS Placement # Report/Attachment Name Description Size Dimensions 1 BOM for Drawing REPORT FAILED A 11 X 8.5 (Landscape) 2 {1} 2444347 a.pdf Print (2444347-GMI_RFI_SS15.pdf) 3 EO Report_REPORT ON GMI SPECIAL Engineering Order A 11 X 8.5 (Landscape) STUDY #15_ RADIO FREQUENCY INTERFERENCE_REV A_2444347.pdf THIS SHEET IS FOR REFERENCE ONLY Ball Aerospace P.O BOX 1062 & Technologies Corp. BOULDER, CO 80306 REPORT ON GMI SPECIAL STUDY #15: RADIO FREQUENCY INTERFERENCE... DWG 2444347 REV A NO. ART 7.7.5 COPIES ARE UNCONTROLLED PACKAGE CREATED BY Paterson, Helen 1/19/2015 2:01 PM SHEET PACKAGE DRAWING REPORT 2of 2 The reports and attachments generated with this sheet and listed below constitute the entire Package Drawing Report for: 2444347 The following files were located in Agile and are associated with the requested drawing, however, they have not been included in the package report. Please verify that these files do not belong in the package. Otherwise use Agile or the print report to manually correct for this. PACKAGE EXCLUSIONS File Name Description Owner gmi_land_mask.mat Ref - Land Mask File Drawing 2444347-GMI_RFI_SS15_RFIDatabase.xlsx Ref - RFI database Drawing 2444347-GMI_RFI_SS15.docx Native Drawing gmi_flag_earth_rfi.m Ref - Matlab Code Drawing earth_rfi_test_file.mat Ref - Test File Drawing gmi_earth_rfi_map.mat Ref - RFI Map Drawing test_gmi_flag_earth_rfi.m Ref - Test Script Drawing THIS SHEET IS FOR REFERENCE ONLY Ball Aerospace P.O BOX 1062 & Technologies Corp. BOULDER, CO 80306 REPORT ON GMI SPECIAL STUDY #15: RADIO FREQUENCY INTERFERENCE... DWG 2444347 REV A NO. ART 7.7.5 [S Printed By: Paterson, Helen 1/19/2015 2:01 PM (ART:7.7.5) Sheet 1 of 32 y s t e m E n Systems Engineering Report 1600 Commerce Street gin Boulder, Colorado 80301 e e r in Page 1 of 32 g R Title Document No. Rev e p o Report on GMI Special Study #15: Radio Frequency Interference 2444347 A r t : 2 Prepared By Date Contract No. CAGE Code 4 4 David Draper 1/16/2015 NNG05HY12C 13993 4 3 4 7 P r 1 SCOPE o g r a This report contains the results of GMI special study #15. An analysis is conducted to identify sources of radio m : frequency interference (RFI) to the Global Precipitation Measurement (GPM) Microwave Imager (GMI). The G RFI impacts the 10 GHz and 18 GHz channels at both polarities. The sources of RFI are identified for the M I following conditions: over the water (including major inland water bodies) in the earth view, and over land in the earth view, and in the cold sky view. A best effort is made to identify RFI sources in coastal regions, with noted degradation of flagging performance due to the highly variable earth scene over coastal regions. A database is developed of such sources, including latitude, longitude, country and city of earth emitters, and position in geosynchronous orbit for space emitters. A description of the recommended approach for identifying the sources and locations of RFI in the GMI M channels is given in this paper. An algorithm to flag RFI contaminated pixels which can be incorporated into P the GMI Level 1Base/1B algorithms is defined, which includes Matlab code to perform the necessary flagging 2 0 of RFI. A Matlab version of the code is delivered with this distribution. 4: 3 : 2 This special study report includes the following files: 1 5 2444347-GMI_RFI_SS15.docx: This document 1 0 2444347-GMI_RFI_SS15_RFIDatabase.docx: Spreadsheet containing significant land RFI locations 2 / gmi_flag_earth_rfi.m: Matlab code to flag RFI over the earth 9 1 / gmi_land_mask.mat: Land mask used by the algorithm 1 : gmi_earth_rfi_map.mat: Matlab save file containing the earth RFI maps shown in this paper. e t a test_gmi_flag_earth_rfi.m: Test script to run gmi_flag_earth_rfi.m D earth_rfi_test_file.mat: Matlab save file used in test_gmi_flag_earth_rfi.m e s a e In association with this report, the special study also includes the companion document and associated files: el R 2434007A-GMIColdRFI.docx: GMI Cold View RFI Flagging Algorithm v gmi_flag_cold_rfi.m: Matlab code to flag RFI in the cold view. e R test_rfi_flag.m: Test script to run the cold sky flagging algorithm d RFI_data_all: Matlab save file used in test_rfi_flag.m e s RFI_data_subset: Matlab save file used in test_rfi_flag.m a e el R n o ti c u d Export or re-export of information contained herein may be subject to restrictions and requirements of U.S. export laws and regulations, o and may require advance authorization from the U.S. Government r P Revision History, Application, and Approval are controlled by a separate database. e: cl TPL0000018 Rev B 8/09/2010 Systems Engineering Page 1 of 32 y c e Lif Agile Revision: A (C446818) [S Printed By: Paterson, Helen 1/19/2015 2:01 PM (ART:7.7.5) Sheet 2 of 32 y s t e m E n Systems Engineering Report 1600 Commerce Street gin Boulder, Colorado 80301 e e r in Document No. 2444347 Rev. A Page 2 of 32 g R e p o 2 INTRODUCTION r t : 2 The Global Precipitation Measurement (GPM) Microwave Imager (GMI) has been operating nearly 4 4 continuously for nearly a year. The instrument is a conically scanning radiometer consisting of 13 narrow-band 4 3 microwave radiometric channels with frequencies centered at 10.65 GHz, 18.7 GHz, 23.8 GHz, 36.64 GHz, 89 4 7 GHz, 166.0 GHz and 183.31 GHz. The lower frequency channel bands are maintained within the National P Telecommunications & Information Administration (NTIA) allocations for “Earth Exploration-Satellite r o (passive).” The 10.65 GHz channel has a bandwidth of 100 MHz spanning 10.6 to 10.7 GHz. The 18.7 GHz g r a channel has a 200 MHz bandwidth spanning 18.6 to 18.8 GHz. The NTIA-allocated bands used by GMI share m utility with a number of other applications; ref. Figure 1. For example, the 100 MHz band allocated for the : G 10.65 GHz channel is also used for fixed ground transmissions, passive space research, and radio astronomy. It M I neighbors a band allocated for fixed (i.e. geosynchronous) satellite space-to-earth transmissions. Likewise, the 200 MHz 18.7 GHz band shares its allocation with fixed satellite space-to-earth transmitters. The transmitters near and around the GMI bands provide radio-frequency interference (RFI) that corrupts the GMI radiometric measurements. This paper describes a methodology for detecting RFI over ocean and land for GMI. It reports the magnitude and location of RFI for land-based transmitters and the position in geosynchronous orbit for space-based interferers. M P 2 0 : 4 3 : 2 1 5 1 0 2 / 9 1 / 1 : e t a D e s a e el R v e R Figure 1. NTIA Spectrum allocations for the 10 and 18 GHz bands d e s a e el R n o ti c u d o r Export or re-export of information contained herein may be subject to restrictions and requirements of U.S. export laws and regulations, P and may require advance authorization from the U.S. Government : e cl TPL0000018 Rev B 8/09/2010 Systems Engineering Page 2 of 32 y c e Lif Agile Revision: A (C446818) [S Printed By: Paterson, Helen 1/19/2015 2:01 PM (ART:7.7.5) Sheet 3 of 32 y s t e m E n Systems Engineering Report 1600 Commerce Street gin Boulder, Colorado 80301 e e r in Document No. 2444347 Rev. A Page 3 of 32 g R e p o 3 ANALYSIS METHODOLOGY r t : 2 RFI in earth radiometric measurements produces artificially high measurements that are generally inconsistent 4 4 with the natural spectral variability of the earth. A robust method of detecting RFI over land is the “spectral 4 3 difference” method, where a channel of interest is compared to a neighboring channel of the same polarization. 4 7 Large differences between the two channels indicate man-made transmissions. Over the ocean, RFI may be P detected by comparing the ocean measurements to a model fit using a radiative transfer model. This “model r o difference” method may be approximated by replacing the radiative transfer model with a linear combination of g r a other uncorrupted radiometric channels. The model difference method using other channels as proxy for the m model has been successfully used on WindSat (see Li et. al, TGRS, 2006). : G M I The spectral difference and model difference method both rely on the assumption that the other radiometric channels are sufficiently correlated with the channel of interest such that other channels may be successfully used to represent an uncorrupted version of that channel. This assumption may be used to produce a generalized RFI detection method for both land and ocean. The generalized method expands the idea of the model difference method to land scenes. For ocean and land separately, a set of coefficients is determined for the GMI 10.65 through 89 GHz channels such that the magnitude of RFI is represented as the difference between the channel of interest and a linear combination of all channels of different frequencies and their M squares, P 2 0 ∆Tb[i]=Tb[i]−a [i]+ ∑(a [i]Tb[j]+b [i]Tb2[j]) (1) 34:  0 j j  2: j∈{fc(j)≠fc(i)} 1 5 1 where i represents the channel index to the channel of interest and j is the channel index to all other channels 0 2 with different center frequency. Eq. (1) may be written in an even simpler form, as a linear combination of all 9/ 1 channels, / 1 : e ∆Tb[i]= a′[i]+∑(a′[i]Tb[j]+b′[i]Tb2[j]) (2) at o j j D j e s a where j is summed over all channels, a′j[i]=1 for the channel of interest, a′j[i]=0 for channels with the same center ele R frequency of the channels of interest, a′[i]=−a [i], and a′[i]=−a [i] for all channels (j) with a different o o j j v e center frequency than the channel of interest. Likewise, b′[i] = 0 for the channel of interest and all channels R j d with the same center frequency, and b′[i]=−b [i] for all other channels. e j j s a e A separate set of coefficients is determined for land, ocean, and sea ice. el R n o This methodology works well for most earth conditions. It, however, can produce false alarms along coast ti c lines, near lakes and islands, over certain odd terrains, mis-flagged sea ice, or heavy snow. It can also be highly u d o r Export or re-export of information contained herein may be subject to restrictions and requirements of U.S. export laws and regulations, P and may require advance authorization from the U.S. Government : e cl TPL0000018 Rev B 8/09/2010 Systems Engineering Page 3 of 32 y c e Lif Agile Revision: A (C446818) [S Printed By: Paterson, Helen 1/19/2015 2:01 PM (ART:7.7.5) Sheet 4 of 32 y s t e m E n Systems Engineering Report 1600 Commerce Street gin Boulder, Colorado 80301 e e r in Document No. 2444347 Rev. A Page 4 of 32 g R e p o dependent upon geometry, and may produce anomalous results if the attitude or altitude of the spacecraft r t: changes. 2 4 4 4 4 ALGORIGTHM DESCRIPTION 3 4 7 The RFI identification algorithm starts with identifying the type of surface viewed by the instrument. The P classification relies upon a look-up table of percentage water as a function of latitude and longitude. The look- r o up table for the percentage land is derived from data found at g ra http://www.shadedrelief.com/natural3/pages/extra.html, reduced in resolution to be more consistent with the m GMI resolution with additional modification for water features not well represented in the original mask. A : G figure showing the water percentage look-up table is shown in Figure 2. M I M P 2 0 : 4 3 : 2 1 5 1 0 2 / 9 1 / 1 : e t a D e s a e el R v e R d e s a e el R n o ti c u Figure 2. Percentage water for the earth. This mask is used to identify land versus ocean/lake measurements. d o r Export or re-export of information contained herein may be subject to restrictions and requirements of U.S. export laws and regulations, P and may require advance authorization from the U.S. Government : e cl TPL0000018 Rev B 8/09/2010 Systems Engineering Page 4 of 32 y c e Lif Agile Revision: A (C446818) [S Printed By: Paterson, Helen 1/19/2015 2:01 PM (ART:7.7.5) Sheet 5 of 32 y s t e m E n Systems Engineering Report 1600 Commerce Street gin Boulder, Colorado 80301 e e r in Document No. 2444347 Rev. A Page 5 of 32 g R e p o r t: A measurement is classified according to one of 6 surfaces if the following criteria are met: 2 4 4 4 1. Ocean/Lake: Water fraction from Figure 2 is equal to 100% (sea ice and stormy data are later removed) 3 4 2. Land: Water fraction from Figure 2 is less than 5%. 7 3. Coast: Water fraction is between 5% and 100% P r 4. Sea Ice: Measurement is flagged as Ocean/Lake AND o g Latitude is greater than 40° or less than -50° AND r a m the 10 GHz H-pol channel Tb is greater than 125K. : 5. Sea Ice Edge: Sea Ice is dilated by 7 cells in both along and cross-scan. G M 6. Stormy Seas: Measurement is flagged as Ocean/Lake AND I 36Ghz H-pol channel Tb is greater than 200K AND Measurement is not flagged as Sea Ice. The stormy areas are dilated by +/-3 cells. Any data representing stormy seas, sea ice, sea ice edge are removed from the ocean/lake classification. After identifying the surface type, the generalized RFI index ∆Tb[i] is computed as follows: 1. Ocean/Lake: Equation (2) is applied using the coefficients a′[i] and b′[i] computed for ocean surfaces. M j j P 2. Land: Equation (2) is applied using the coefficients a′[i] and b′[i] computed for land surfaces. 2 j j 0 : 3. Coast: Both ocean and land coefficients are applied, and the algorithm chooses the minimum of the two. 4 3 If the result is less than the coast threshold, then the algorithm sets the index equal to zero. 2: 1 4. Sea Ice: Equation (2) is applied using the coefficients a′[i] and b′[i] computed for sea ice surfaces. 5 j j 1 If the result is less than the sea ice threshold, then the algorithm sets the index equal to zero. 0 2 5. Sea Ice Edge: The index is set to zero. 9/ 1 6. Stormy Seas: The index is set to zero. / 1 : e t The coefficients for each surface type and channel are given in appendix A. a D e For real-time flagging of RFI, the following ∆Tb[i] thresholds are suggested: s a e el R Table 1. Recommended thresholds for flagging RFI in a real-time algorithm v e R 10 GHz V 10 GHz H 18 GHz V 18 GHz H Ocean/Lake 15K 15K 10K 10K d e s Land 20K / 50K* 20K / 50K* 10K / 25K* 10K / 25K* a e Coast 50K 50K 30K 30K el Sea Ice 50K 50K 30K 30K R n Sea Ice Edge N/A N/A N/A N/A o ti Stormy Seas N/A N/A N/A N/A c u *Threshold for latitudes above 59 degrees and below -59 degrees latitude where heavy snow persistently exists d o r Export or re-export of information contained herein may be subject to restrictions and requirements of U.S. export laws and regulations, P and may require advance authorization from the U.S. Government : e cl TPL0000018 Rev B 8/09/2010 Systems Engineering Page 5 of 32 y c e Lif Agile Revision: A (C446818) [S Printed By: Paterson, Helen 1/19/2015 2:01 PM (ART:7.7.5) Sheet 6 of 32 y s t e m E n Systems Engineering Report 1600 Commerce Street gin Boulder, Colorado 80301 e e r in Document No. 2444347 Rev. A Page 6 of 32 g R e p o r t: We note that this detection method is subject to false alarms for odd or spatially varying surface characteristics. 2 4 Significant surface snow will cause false alarms in both 10 and 18 GHz channels. Areas such as salt flats, 4 4 widespread deserts, or mountainous terrain may also cause false alarms. 3 4 7 5 ANALYSIS RESULTS P r o The generalized RF detection method is used to evaluate RFI over first 6 months of GMI operations. Three g ra types of RFI are found: land-based RFI detected through the GMI main beam, RFI from geosynchronous m satellites reflecting off the ocean surface detected through the GMI main beam, and RFI from geosynchronous : G satellites interfering with the GMI cold swath. M I 5.1 RFI from Ground-based Transmitters The ∆Tb[i] values are averaged over 6 months of GMI data from March 2015 through September 2015. The maximum daily average values are also computed. The average RFI index and maximum daily average RFI index plots show somewhat different characteristics of the RFI. The average RFI index tends to reduce geophysical variation and shows RFI events that are persistent over time. The maximum daily average RFI index illustrates RFI that is not terribly persistent at the cost of amplifying geophysical variation. M P 2 Figure 3 through Figure 16 show the RFI index in and around each continent except Antarctica. 0 : 4 3 The 10 GHz channels are corrupted by ground-based fixed transmitters centered in various metropolitan areas 2: 1 of the world. The 10 GHz RFI is especially prevalent in the Eastern Hemisphere. In the United Kingdom, Italy, 5 Turkey and Egypt, very high levels of widespread 10 GHz RFI are detected. The 6-month average RFI levels 1 0 for these areas range from 5K up to over 100K. 10 GHz RFI is also widespread in China and Japan. In the 2 / 9 western hemisphere, the 10 GHz RFI is lower magnitude and more scattered. The highest RFI incidents in the 1 / 1 western hemisphere occur in Mexico around cities such as Mexico City, Monterrey and Guadalajara, as well as : e near Sao Paulo Brazil. Please note that the “stripes” over the ocean for the 10 GHz channel are due to sun t a glint. D e s a For the 18 GHz channels, RFI is detected from ground-based transmitters in several specific countries. Areas e with the most 18 GHz RFI include Belarus, Libya, the Sudan, and Chile. These RFI events are generally el R smaller in magnitude than the 10 GHz RFI, with 6-month averages in affected areas ranging from 2K to 20K. v e R Appendix B provides a list of earth source RFI locations and the average RFI index over the first 6 months of d e GMI operation. The accompanying file gmi_earth_rfi_map.mat contains the data corresponding to the figures s a in this section. e el R n o ti c u d o r Export or re-export of information contained herein may be subject to restrictions and requirements of U.S. export laws and regulations, P and may require advance authorization from the U.S. Government : e cl TPL0000018 Rev B 8/09/2010 Systems Engineering Page 6 of 32 y c e Lif Agile Revision: A (C446818) [S Printed By: Paterson, Helen 1/19/2015 2:01 PM (ART:7.7.5) Sheet 7 of 32 y s t e m E n Systems Engineering Report 1600 Commerce Street gin Boulder, Colorado 80301 e e r in Document No. 2444347 Rev. A Page 7 of 32 g R e p o r t : 2 4 4 4 3 4 7 P r o g r a m : G M I M P 2 0 : 4 3 : 2 1 5 1 0 2 / 9 1 / 1 : e t a D e s Figure 3. World-wide 10 GHz RFI index over 6 months of data from March 2014 through September 2014. Top left: average a e v-pol RFI index. Top right: average h-pol RFI index. Bottom left: maximum daily average v-pol RFI index. Bottom right: el maximum daily average h-pol RFI index. Note that the stripes in the 10V and 10H plots over the ocean are due to sun glint. R v e R d e s a e el R n o ti c u d o r Export or re-export of information contained herein may be subject to restrictions and requirements of U.S. export laws and regulations, P and may require advance authorization from the U.S. Government : e cl TPL0000018 Rev B 8/09/2010 Systems Engineering Page 7 of 32 y c e Lif Agile Revision: A (C446818) [S Printed By: Paterson, Helen 1/19/2015 2:01 PM (ART:7.7.5) Sheet 8 of 32 y s t e m E n Systems Engineering Report 1600 Commerce Street gin Boulder, Colorado 80301 e e r in Document No. 2444347 Rev. A Page 8 of 32 g R e p o r t : 2 4 4 4 3 4 7 P r o g r a m : G M I M P 2 0 : 4 3 : 2 1 5 1 0 2 / 9 1 / 1 : e t a D e s a e el R Figure 4. North America 10 GHz RFI index over 6 months of data from March 2014 through September 2014. Top left: v average v-pol RFI index. Top right: average h-pol RFI index. Bottom left: maximum daily average daily average v-pol RFI e R index. Bottom right: maximum daily average h-pol RFI index. RFI is exhibited in many major cities in the US, especially in California, the Midwest and New England. Mexico City and the surrounding areas show very high levels of RFI. d e s a e el R n o ti c u d o r Export or re-export of information contained herein may be subject to restrictions and requirements of U.S. export laws and regulations, P and may require advance authorization from the U.S. Government : e cl TPL0000018 Rev B 8/09/2010 Systems Engineering Page 8 of 32 y c e Lif Agile Revision: A (C446818)