Mapping Titanium Abundance Using Chang'e-1 IIM Data

42nd Lunar and Planetary Science Conference (2011) 1807.pdf

MAPPING TITANIUM ABUNDANCE USING CHANG’E-1 IIM DATA. Zongcheng Ling1,2, Jiang Zhang1,2, Jianzhong Liu1, Wenxi Zhang1, Guangliang Zhang1, Bin Liu1, Xin Ren1, Lingli Mu1, Jianjun Liu1, Chunlai Li1 1National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100012, P.R. China, 2School of Space Science and Physics and Shandong Provincial Key Laboratory of Optical Astronomy & Solar-Terrestrial Environ- ment, Shandong University at Weihai, Weihai, Shandong 264209, P. R. China; ([email protected])

Introduction: Obtaining the information of distribution of lunar titanium across lunar surface is of great importance for understanding the lunar geology and petrogenesis of lunar basalt rocks as well as future lunar resource utili- zations. Mapping titanium abundance from spectral measurements has become a classic issue since the te- lescopic studies of the Moon in the 1970s, when Cha- rette et al. showed that TiO2 abundance in mature basal- tic regolith is correlated with its UV/VIS ratio [1]. This “ Charette Relation” provided the first quantitative comparisons of TiO2 for nearside mare regions, and later was refined by many researchers[e.g. 2, 3]. Re- cently, a series of empirical models were developed to predict the TiO content by using remotely sensed data Fig.1. Spectral ratio(522nm/757nm) versus reflectance (757nm) 2 plot for the sample stations observed by Chang’E-1 IIM. such as Galileo multispectral data and Clementine UV-

VIS images, etc[4-7]. China’s first lunar probe Chang’E-1 has already made many achievements including 4 Terabytes science data from eight payloads as well as various science re- turns, e.g., lunar global CCD image, etc [8]. Imaging Interferometer (IIM), together with Gamma Ray Spec- trometer (GRS), are selected to detect the chemical and mineralogical compositions on the lunar surface [9,10]. We have derived a FeO model and examined its ability for iron mapping using IIM data [10]. Here we will present a new algorithm for mapping TiO2 abundance using IIM data Algorithm to derive TiO2 abundance: IIM onboard Chang’E-1 lunar probe is a Fourier transform Sagnac imaging spectrometer. Detailed para- Fig. 2. TiO2 contents of the Apollo and Luna samples versus the meters can be found in the reference [10]. Due to the spectral Ti-sensitive parameter θTi. limited response of CCD detector, IIM’s former 5 spec- We tried to suppress the maturity effect by using the tral channels (B1-B5) show a poor quality (SNR less Ti-sensitive parameter θTi, which would have a pow- than 10), thus we chose the B6 (522nm) and er-law relationship with TiO2 abundance [7]. θTi meas- B24(757nm) to perform Ti content calculations in com- ures the angle between a line parallel to x axis and a line parisons with Lucey’s 415 and 750nm Clementine between the origin and data point. By maximizing the bands. The method for calculating the TiO2 abundance correlation between remotely-measured θTi and TiO2 is similar to that of Lucey et al. [7]. We tried to correlate contents, we got the origin at (0.061, 0.531), which the laboratory TiO2 contents of typical lunar soils with yields a correlation coefficient of 0.86. the remotely-sensed multispectral images for individual We got the expression of θTi as sample stations. The first step is to acquire spectra for ⎛⎞RR522 /757 − 0.531 38 Apollo and Luna sample stations (except those from θTi = arctan ⎜⎟ Apollo 15 site) covered by Chang’E-1 IIM data. In ad- ⎝⎠R757 − 0.061 dition, from the experience of Lucey et al.[7], Luna 16 The plot of TiO2 content and the spectral parameter and Luna 24 has TiO2 content anomalies, thus we also is shown in Figure 2. The equation of best fit for these discarded these two sites and chose the other 36 sample data points is sites to do the regression. The spectral data for 38 sam- wt.% TiO =× 1.158 (θ )5.364 ple stations extracted from Chang’E-1 images were 2 Ti plotted on a ratio reflectance diagram (Figure 1). The standard deviation for the power-law fit is 1.56 42nd Lunar and Planetary Science Conference (2011) 1807.pdf

wt. %. data are very near and not so distinguishable. By peak Regional Case study: fitting of the histogram, we found the two peaks sit at Although our model is applicable for the global IIM 7.42 wt.% and 9.5 wt.% respectively. In general, our data, here we only do the regional case study of lunar model overestimates the low-Ti region by ~1.5 wt.%, surface for ease of validataion and comparision with while underestimates the high-Ti regions by ~0.8 wt.%. Clementine UVVIS data. Clementine UVVIS-derived This is mainly because IIM data did not cover Apollo 15 landing site, thus our model lacks the regions with rela- TiO2 map based on Lucey’s model [7] is also used for comparisons. A mare region near Central Mare Sereni- tively low-Ti basalt as input parameters. In addition, the tatis (MS2), which has been an optical standard for te- relatively poor quality of IIM data (lower SNR) and lescopic studies [11], is selected for the regional case topographic shading effects would be potential reasons for the discrepancies of TiO mapping between study. We chose this region mainly because it is on the 2 Chang’E-1 and Clementine data. boundary of high and low titanium basalts, i.e., to the north is relatively uniform low-titanium mare basalt of Mare Serenitatis and to the south is a sharp boundary with an older high-titanium mare basalt of Mare Tran- quillitatis (as indicated in Figure 3). Note that, all IIM data have been resampled to 100 m/pixel for ease of comparison with Clementine UVVIS data.

Fig. 4. Data distribution of Chang’E-1 IIM and Clementine UVVIS images for MS2 mare region. Conclusions and future work: The derivation and application of the preliminary algorithm for TiO2 mapping using Chang’E-1 IIM images were presented in this abstract. In comparisons with Clementine UVVIS results, our algorithm also exhibits the ability to extract TiO2 abundance distributions on the Moon, although its limitations due to the lack of Apollo 15 data should not be neglected. We will con- Fig.3. Comparisons of TiO2 map near MS2 mare region between tinue to refine our TiO2 model to improve its prediction Chang’E-1 IIM and Clementine UVVIS images. ability in future. Lunar mare is well known for its enrichment in Ti Acknowledgements: This work was supported by the due to ilmenite, which is the most abundant oxide min- National High-Tech Research and Development Program of eral in lunar rocks. Mare Serenitatis in the mid- China (2008AA12A212,211,213, 2010AA122203), China dle-southern part in just on the boundary between relatively “low-Ti” and “high-Ti” regions, which are Postdoctoral Science Foundation (20090450580) and National referred as lunar “red” and “blue” mare, respectively. In Natural Science Foundation of China (11003012). References: [1] Charette et al. (1974) JGR, 79, 1605-1613. [2] IIM 757nm mosaic(Figure 3), it has distinct albedo dif- Johnson et al., (1977) Proc. Lunar. Sci. Conf. 8th, 1029–1036. ference between the two kind of mare basalts. From the [3] Pieters, (1978) Proc. Lunar Planet. Sci. Conf. 9th, TiO2 map, it’s easy to find the color difference boun- 2825–2849. [4] Pieters et al. (1993) JGR, 98, 17127-17148. [5] dary between “red” and “blue” maria, although not so Blewett et al. (1997) JGR, 102, 16319-16325. [6] Lucey et al. clear as Clementine data. As indicated by the data dis- (1998) JGR, 103, 3679-3699. [7] Lucey et al. (2000) JGR, 105, tribution histogram (Figure 4), the average TiO2 values 20297-20305. [8] Ouyang et al.(2010) Chin. J. Space Sci., between Chang’E-1 IIM and Clementine UVVIS image 2010, 30(5): 392-403. [9] Zheng,et al.(2008) Planet Space Sci, are similar (for IIM is about 7.39 wt.% in comparison 56, 881-886. [10] Ling et al.,(2011) Chin. Sci Bull, 56: 1-5. with Clementine’s 7.25 wt.%). However, Clementine [11] Pieters et al., (2008) Adv Space Res, 42, 248-258. data show a bimodal distribution, and has two peaks at 5.94 wt.% and 10.34 wt.%, in relation with low-Ti and high-Ti regions in the images. The two peaks of IIM