Aerosol optical depth as observed by the Mars Science Laboratory REMS UV photodiodes

Accepted Manuscript

Aerosol optical depth as observed by the Mars Science Laboratory REMS UV photodiodes Michael D. Smith , Mar´ıa-Paz Zorzano , Mark Lemmon , Javier Mart´ın-Torres , Teresa Mendaza de Cal PII: DOI: Reference:

S0019-1035(16)30415-8 10.1016/j.icarus.2016.07.012 YICAR 12128

To appear in:

Icarus

Received date: Revised date: Accepted date:

24 March 2016 14 July 2016 21 July 2016

Please cite this article as: Michael D. Smith , Mar´ıa-Paz Zorzano , Mark Lemmon , Javier Mart´ın-Torres , Teresa Mendaza de Cal , Aerosol optical depth as observed by the Mars Science Laboratory REMS UV photodiodes, Icarus (2016), doi: 10.1016/j.icarus.2016.07.012

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

AN US

CR IP T

Highlights  Systematic REMS observations enable retrieval of daily aerosol optical depth.  Numerous short-lived increases in optical depth punctuate the background variation.  Cleaning of dust from the REMS photodiodes was observed around Ls=270°.  Aerosol particle sizes are larger on average when optical depth is highest.

ACCEPTED MANUSCRIPT

Aerosol optical depth as observed by the Mars Science Laboratory REMS UV photodiodes

a

CR IP T

Michael D. Smitha,*, María-Paz Zorzanob,c, Mark Lemmond, Javier MartínTorresb,e, Teresa Mendaza de Calb

NASA Goddard Space Flight Center, Greenbelt, MD 20771, United States.

b

Luleå University of Technology, Department of Computer Science, Electrical, and Space Engineering, Kiruna, Sweden. Centro de Astrobiología (CSIC-INTA), Torrejón de Ardoz, Madrid, Spain.

d

Texas A&M University, College Station, TX 77843, United States.

e

Instituto Andaluz de Ciencias de la Tierra (CSIC-UGR), Granada, Spain.

*

To whom correspondence should be addressed:

AN US

c

AC

CE

PT

ED

M

Michael Smith Code 693 NASA Goddard Space Flight Center Greenbelt, MD 20771 (301) 286-7495 [email protected]

ACCEPTED MANUSCRIPT

Abstract

AN US

CR IP T

Systematic observations taken by the REMS UV photodiodes on a daily basis throughout the landed Mars Science Laboratory mission provide a highly useful tool for characterizing aerosols above Gale Crater. Radiative transfer modeling is used to model the approximately 1.75 Mars Years of observations taken to date taking into account multiple scattering from aerosols and the extended field of view of the REMS UV photodiodes. The retrievals show in detail the annual cycle of aerosol optical depth, which is punctuated with numerous short timescale events of increased optical depth. Dust deposition onto the photodiodes is accounted for by comparison with aerosol optical depth derived from direct imaging of the Sun by Mastcam. The effect of dust on the photodiodes is noticeable, but does not dominate the signal. Cleaning of dust from the photodiodes was observed in the season around Ls=270°, but not during other seasons. Systematic deviations in the residuals from the retrieval fit are indicative of changes in aerosol effective particle size, with larger particles present during periods of increased optical depth. This seasonal dependence of aerosol particle size is expected as dust activity injects larger particles into the air, while larger aerosols settle out of the atmosphere more quickly leading to a smaller average particle size over time. Keywords: Mars, atmosphere; Atmospheres, composition; Radiative transfer

M

1. Introduction

AC

CE

PT

ED

Mars atmospheric aerosols have a major influence on the solar and thermal radiation field at the surface of Mars. As a consequence, the abundance or optical depth of these aerosols is a key piece of information needed in constructing accurate models of Mars climate behavior (e.g., Gierasch and Goody, 1972; Kahn et al. 1992; Madeleine et al. 2011) and in correcting spectroscopic observations of Mars surface composition (e.g., Smith et al. 2000; Arvidson et al., 2015). With respect to lander observations of surface composition, studies using the Pathfinder IMP camera and the Mars Exploration Rover (MER) Pancam and Navcam cameras have reinforced the requirement for incorporating aerosol effects in visible and near-infrared retrievals (e.g. Thomas et al. 1999, 2000; Lemmon et al. 2004; Johnson et al., 2006a,b; Soderblom et al. 2008). Similarly, investigation of the complex dynamics that influence dust lifting and transport of aerosols, ranging from dust devils to cloud formation to global dust storms, requires the same detailed record of aerosol optical depth on timescales rapid enough to capture short-lived events (e.g., Newman et al., 2002; Kahre et al., 2006; Rafkin et al., 2016). Orbital spacecraft observations over the past two decades have produced significant progress towards understanding the global climatology of dust and water ice aerosols (e.g. Smith et al. 2004; Smith 2009; McCleese et al., 2010; Montabone et al. 2015, and references therein). From landed platforms, characterization of aerosol optical depth has

ACCEPTED MANUSCRIPT

CR IP T

been performed using a number of techniques including direct imaging of the Sun (Smith and Lemmon, 1999; Lemmon et al., 2004, 2015), upward-looking thermal-infrared spectroscopy (Smith et al., 2006), and LIDAR (Whiteway et al., 2009). Here we build upon those previous results by retrieving daily measurements of aerosol optical depth in the column above the Mars Science Laboratory “Curiosity” rover. These new retrievals are important for understanding the local meteorology of the Gale Crater landing site, which is not necessarily well represented by the global climatology (Moore et al., 2016; Rafkin et al., 2016).

M

AN US

To retrieve aerosol optical depth, we use the systematic measurements of ultraviolet (UV) radiation from the set of upward-looking photodiodes that are part of the Rover Environmental Monitoring Station (REMS) suite of instruments (Gómez-Elvira et al., 2012), which are well suited for this purpose. The REMS instrument suite has been used to investigate a number of atmospheric phenomena including variations in surface pressure (Harri et al., 2014; Haberle et al., 2014), atmospheric tides (Guzewich et al., 2016), frost events (Martínez et al., 2016), and dust devils (Kahanpää et al., 2016), as well as for the study of possible liquid brines and water activity in the regolith (MartínTorres et al., 2015). Measurements of the atmospheric state by REMS, including air and ground temperature, relative humidity, and UV radiation, have also been used to put into context the observations from other instruments on Curiosity (e.g., Webster et al., 2015; Kloos et al., 2016).

PT

ED

In this paper we present the results of retrievals of aerosol optical depth using observations from the REMS UV photodiodes. We describe the REMS UV photodiodes and the set of available observations in Section 2. In Section 3 we detail the retrieval algorithm and uncertainties. The results of the retrieval are described in Section 4, and are discussed in greater detail in Section 5. Section 6 provides a summary of our findings.

2. Data Set

AC

CE

The Mars Science Laboratory (MSL) Curiosity rover has been operating on the surface of Mars in Gale Crater (4.59° S latitude, 137.44° E longitude) since August 2012 (Grotzinger et al., 2012). Here we use observations of UV radiative flux from the REMS instrument taken systematically each sol since the beginning of MSL landed operations, which cover approximately 1.75 Martian years, from sol 9 (Mars Year 31, Ls=155°) to sol 1159 (Mars Year 33, Ls=66°). The “Mars Year” convention is from Clancy et al. (2000), where Mars Year (MY) 1 begins on 11 April 1955. The current Mars Year is 33, which began 18 June 2015. 2.1. REMS UV Instrument One component of the REMS instrument suite (Gómez-Elvira et al., 2012) is a set of six photodiodes designed to measure UV flux (Zorzano et al., 2011). The photodiodes are

ACCEPTED MANUSCRIPT

mounted on the MSL rover deck facing directly upward to the sky. Each photodiode nominally views a cone centered straight upward from the rover deck with a 30° halfwidth from zenith. Five of the photodiodes cover narrow spectral bands (FWHM of 30-40 nm) with central wavelengths between 250 and 335 nm. The sixth photodiode is much wider covering the entire spectral range between 230 and 350 nm. Figure 1 shows the relative spectral response of the six REMS UV photodiodes.

AN US

CR IP T

REMS UV observations are obtained systematically throughout each sol as part of a standard background observation set. The UV flux for each of the six photodiodes is recorded every second for the first five minutes of each Martian “hour,” which is defined as one-twenty-fourth of a Martian sol. In addition, several additional hour-long blocks of continuous data are obtained each sol, with the timing of the blocks rotated though the sol to cover all local times over a time period of several sols. In general, in addition to the five-minute observations at the beginning of each hour, each sol has had at least three or four of the hour-long blocks during daylight hours when the UV data are useful. 2.2. Typical REMS UV Data and its Relation to Aerosol Optical Depth

PT

ED

M

The REMS UV raw measurements of photodiode output current from each of the six UV photodiodes are binned every three minutes to form the set of observations used in this study. The viewing geometry, in the form of the zenith angle and azimuth of the Sun in rover coordinates, is known for each observation. For this study, all geometry is always converted into the rover frame of reference so that rover tilt and attitude are automatically accounted for. Figure 2 shows the observed signal from the UVA (330 nm) photodiode for sols 62 and 91 (MY 31, Ls = 185° and 203°). Both the hourly observations and hourlong blocks are apparent. We choose to use the raw output current from the photodiodes instead of calibrated flux. The conversion from current to flux is a linear function, and we prefer to include this conversion empirically into a spatial response function for a more robust and self-consistent retrieval (see Section 3.2).

AC

CE

The observed flux in the UV comes from both the direct radiation from the Sun and the diffuse radiation that is scattered into the field-of-view (Zorzano and Córdoba-Jabonero, 2007). Therefore, there is useable signal whenever the Sun is above the horizon, even when the Sun is not directly within the UV photodiode field of view. The observed UVA signal in Figure 2 is zero before the Sun rises, increases as the Sun climbs higher in the sky and then decreases as the Sun sets. In this particular case, the Sun is within the nominal field of view between 10:00 and 14:00 local true solar time (LTST, which we will use for all local times in this paper). Both the direct solar beam and the diffuse scattered radiation are sensitive to aerosol optical depth, which enables the retrieval of aerosol optical depth given the observed UV flux, even when the Sun is not in the nominal field of view (Zorzano et al., 2009). Figure

ACCEPTED MANUSCRIPT

2.3. Obscured Fields of View and Reflected Light

CR IP T

3 shows a model of the sky brightness and solar disk within the field of view of a single REMS UV photodiode. The observed signal is this brightness integrated over the photodiode convolved with a spatial response function. The bottom panel of Fig. 3 shows the sensitivity of the observed flux to aerosol optical depth for this particular case with the Sun in the field of view. The direct solar beam is attenuated with increasing aerosol, while the amount of diffuse radiation increases. The contribution of the solar beam dominates leading to a net decrease in observed UV signal.

ED

M

AN US

The REMS UV photodiodes generally have an unobstructed view of the sky in their nominal field of view cone extending 30° from zenith in rover coordinates. However, when the Sun is at particular zenith and azimuth angles in the rover coordinate system, the rover mast and other hardware can cast shadows that partially obscure the photodiode fields of view. This can be seen in Fig. 2 for sol 62, where the red points indicate observations that are partially obscured by shadows. This dramatically complicates the analysis because the portion of the field of view that is obscured can have a very irregular shape blocking both the direct solar beam and the surrounding diffuse sky brightness, and the contribution from the entire field of view is important for the retrieval of aerosol optical depth. It is potentially possible to model these obscurations, but instead we choose to simply exclude those observations from our analysis. The total fraction of observations that are obscured by shadows is relatively small (typically ~10%), and on nearly any given sol there are more than enough observations without shadows to perform a reliable retrieval.

AC

CE

PT

Another complication is that the design of the UV photodiodes allows some sensitivity to incident radiation outside the nominal field of view (Zorzano et al., 2011). Modeling shows that the observed signal when the Sun is at greater than 30° zenith angle (i.e., earlier than 10:00 and later than 14:00 in Fig. 2) is larger than expected from the diffuse scattered light of sky brightness alone. The amount of light reaching the photodiodes is a well-defined function of Sun location (i.e., zenith angle and azimuth) with respect to the rover and an appropriate spatial response function can be derived that takes this into account (see Section 3.2 below).

3. Retrieval Algorithm We use radiative transfer modeling of the observed UV signal in each photodiode to retrieve aerosol optical depth. For each photodiode, aerosol optical depth is retrieved by minimizing the least-squares difference between the observed and computed signals. Although it is possible to perform the retrieval on individual observations, we choose to use all the observations without shadows for a given sol with solar zenith angle less than 60° to retrieve a single optical depth for the sol. This provides for a robust retrieval with

ACCEPTED MANUSCRIPT

minimum uncertainty. Analysis of the residuals, or difference between observed and modeled signal, shows little systematic variation with local time caused by changes in aerosol optical depth within a sol. 3.1. Radiative Transfer and Assumptions

CR IP T

The UV radiation observed by REMS comes from a combination of the direct solar beam (when the Sun is within the field of view) and the diffuse field of light scattered into the field of view by atmospheric aerosols (e.g., Zorzano and Córdoba-Jabonero, 2007; Zorzano et al., 2009). At these wavelengths, thermal radiation is negligible, and the observed signal does not depend on atmospheric temperature, surface temperature, or other surface properties.

ED

M

AN US

Because aerosol scattering is significant, full multiple scattering must be included in the radiative transfer modeling. The forward radiative transfer model uses the discrete ordinates method to treat multiple scattering (e.g., Goody and Yung, 1989; Thomas and Stamnes, 1999). The atmosphere is divided into vertical layers, and as many radiation “streams” can be used as is necessary to accurately model the angular dependence of scattering. For this upward-looking geometry, we use 100 vertical layers with decreasing layer thickness near the surface, and 32 streams to model the diffuse radiation field. The effect of spherical geometry near the horizon is treated using the “pseudo-spherical approximation” (e.g., Spurr, 2002; Thomas and Stamnes, 1999) where the diffuse radiation field is computed using a standard plane-parallel geometry using the discrete ordinates approach, but then the source function is integrated along equivalent curved paths through the layers (Smith et al., 2013).

AC

CE

PT

With the limited wavelength range covered by the set of REMS UV photodiodes, we cannot easily distinguish between dust and water ice aerosols. Orbital observations indicate that dust optical depth usually dominates that of water ice (e.g., Smith, 2004, 2008), so we make the simplifying assumption that the aerosol is purely dust. Numerical experiments show that the REMS UV photodiode observations are not sensitive to variations in the vertical distribution of dust, so for simplicity we assume that the dust is well mixed vertically. We further assume that the dust has an effective particle size of 1.5 µm, which is consistent with orbital observations (Wolff and Clancy, 2003; Wolff et al., 2006, 2009). Dust aerosol scattering properties are taken from detailed modeling of MARCI images (Wolff et al., 2010). If the aerosol has a different particle size than we assume, or if it is a mixture of dust and water ice, then the difference in scattering properties implied by that will be accounted for in the empirical spatial response function that we derive (see Section 3.2). If the aerosol physical properties change as a function of time, then that will show in the residuals in our fits to the data (see Section 5.2).

ACCEPTED MANUSCRIPT

CR IP T

To properly model the extended field of view and wavelength coverage of the REMS UV photodiodes, we compute the upward-looking radiance as viewed by the rover for a grid of zenith angle and azimuth angles in rover coordinates for each 5 nm wavelength between 200 and 400 nm. The spatial grid covers zenith angles from 0° to 70° in 1° increments, and azimuths from 0° to 359° in 1° increments. These are convolved with both the spectral response (Fig. 1) and the spatial response (see Section 3.2) for each filter and integrated to find the total observed flux. The photodiode spectral response exhibits a weak dependence on instrument temperature with peak responses shifting to longer wavelengths at higher temperatures (Gómez-Elvira et al., 2012). This known temperature dependence is taken account of in the computation using the measured instrument temperature recorded by REMS.

AN US

Because of the computational expense required for this computation, we use an interpolated look-up table approach where the above angular and spectral grids were computed for dust optical depths from 0.1 to 2.0 (spaced every 0.1 optical depth units) and for solar zenith angles from 0° to 81° (spaced every 3°). The resulting look-up table has more than 290 million points. 3.2. Spatial Response

ED

M

For a more robust and self-consistent retrieval we empirically derive a spatial response function for each photodiode, which includes both the conversion from raw signal (photodiode output current) to physical parameter (radiant flux), as well as the inherent spatial variation in photodiode response and any reflections of light onto the photodiode.

CE

PT

We use all observations taken between sols 60 and 80 (MY 31, Ls=172°–184°) when the Sun had zenith angle less than 60° in rover coordinates to derive the spatial response. Observations at this time are early enough in the mission that dust accumulation on the photodiodes is minimal (see Sections 3.3 and 5.3), while also having available direct measurement of aerosol optical depth from Mastcam images of the Sun. For now, we assume that variation in zenith angle dominates that in azimuth and assume a constant response in azimuth angle.

AC

For each photodiode, the spatial response function as a function of zenith angle is found that minimizes the least-squares difference between the observations and the modeled flux. Figure 4 shows the results normalized to unity at a zenith angle of zero degrees, which are broadly consistent with the angular calibration results obtained before launch (Gómez-Elvira et al., 2012). Although a large majority of the signal comes from zenith angles less than 30° (the nominal field of view), the amount of light that reaches the photodiodes from larger zenith angles is significant and must be taken into account for an accurate retrieval.

ACCEPTED MANUSCRIPT

CR IP T

It is possible that aerosol optical depth can vary during a sol, which could then alias itself as a variation in zenith angle. However, this variation of optical depth would have to be symmetric about local noon and be the same over the 21 sols used here, so we believe the effect of such variation in optical depth on the derived spatial response functions is negligible. The time period used for this derivation contained no local dust storms and sky imaging showed that water ice clouds were not prominent during this time (Moores et al., 2015; Kloos et al., 2016).

AN US

Sampling the azimuthal dependence of the spatial response requires a much larger dataset to include instances of the rover being at many different headings. Using a spatial response that is only a function of zenith angle, the retrieval for aerosol optical depth was performed over all the data considered in this study (sols 9–1159). The residuals from the retrieval showed significant and systematic variations as a function of solar zenith angle and the azimuth in rover coordinates (with zero azimuth defined as the rover forward direction). These residuals were used to compute an additional correction to the computed fluxes that is a function of solar zenith angle and azimuth angle as shown in Fig. 5. This multiplicative correction is small (less than a few percent) for solar zenith angles less than 30°, but can be tens of percent at higher solar zenith angles. This azimuthal dependence is caused by reflections of light and the obscuration of the diffuse field (sky brightness) by various components of the rover hardware.

M

3.3. Accumulation of Dust on the UV Photodiodes

CE

PT

ED

Over time, dust settled onto the REMS UV photodiodes partially blocking their fields of view. Images taken by the MAHLI camera shown in Fig. 6 document the accumulation of dust on the photodiodes. We model this using a multiplicative factor that is a function of time (sol number). This dust fallout calibration factor is determined using the set of derivations of aerosol optical depth obtained from direct imaging of the Sun by the Mastcam camera (e.g., Lemmon et al., 2015). These Mastcam measurements are nominally taken every three to seven sols and are of very high quality because of the direct nature of the observation.

AC

Other methods to determine a dust fallout calibration are possible, including comparing the observed REMS UV flux when shadows cover the field-of-view to non-shadowed observations (Vicente-Retortillo et al., 2015) and fitting the observed flux as a function of solar airmass (Mason et al., 2015), but both of those approaches have limited applicability and larger uncertainties than using the direct observations of the Sun by Mastcam. The disadvantage of using the Mastcam observations is that the REMS UV retrievals are no longer independent of those from Mastcam and so (for example) the wavelength dependence of optical depth between Mastcam (440 and 880 nm) and the REMS UV photodiodes can no longer be determined.

ACCEPTED MANUSCRIPT

CR IP T

For each sol that contains a Mastcam observation of the Sun, the aerosol optical depth derived by Mastcam is used as a given quantity, and instead of using the retrieval to fit for aerosol optical depth we instead fit for the dust fallout calibration factor that the computed flux must be multiplied by to match the observed signal. For each photodiode, these values smoothed by a triangle function with effective width of 100 sols are used to form the dust fallout calibration factor for any given sol. The individual dust calibration factors for each Mastcam measurement and the 100-sol smoothed version used in the retrieval for the UVA photodiode are shown in Fig. 7. As expected, the calibration factor begins at unity and generally decreases as dust accumulates on the photodiodes. The trends of the dust fallout calibration factor for all six photodiodes are similar and are discussed further in Section 5.3. 3.4. Retrieval Process

ED

M

AN US

Once the spatial response and dust fallout calibration factor have been defined as described above, the retrieval of aerosol optical depth from the REMS UV observations is straightforward. For each sol, all UV observations taken with solar zenith angle less than 60° in rover coordinates are used. In practice, this corresponds to a Local True Solar Time range of 8:00 to 16:00. Any observations that show evidence of shadows are removed. The presence of an obscured field of view from shadows was determined by a combination of excluding observations in a defined range of solar zenith and azimuth angles where obscuration is known to always exist, as well as observations that show a clear drop of more than 5% in signal compared to subsequent (or preceding) observations taken at higher solar zenith angle.

AC

CE

PT

Recalling that a single REMS UV “observation” as defined here is the three-minute average of the individual REMS UV measurements taken every second, there are typically a few dozen observations to be fit for each sol. Because the nominal REMS daily observation plan calls for five minutes of observations at the beginning of each hour, the coverage of the UV observations as a function of local time (and solar zenith angle) is usually quite complete. Figure 8 shows the distribution of the more than 50,000 REMS UV observations that were used in this study as a function of seasonal date (Ls) and local time. The solar zenith angle and azimuth are known for each observation, and so using the known spectral response, derived spatial response, and derived dust fallout calibration factor, the only remaining unknown quantity needed to compute the expected flux in each UV photodiode is the aerosol optical depth. For each photodiode, the retrieval algorithm varies the single unknown, aerosol optical depth, to obtained the least-squares difference between observed and computed UV flux over the set of observations for a given sol. This aerosol optical depth is referenced to 880 nm, since that wavelength is used for the Mastcam calibration of dust accumulation on the photodiodes. These optical depths along

ACCEPTED MANUSCRIPT

with the residuals from the best fit for each observation within a sol are recorded. The best fits are generally quite good with residuals less than 1%. Figure 9 shows typical fits for sols 62 and 91. 3.5. Uncertainties

AN US

CR IP T

Because the dust fallout calibration factor ties our UV aerosol optical depth retrievals to the Mastcam-derived values (or at least to a time-smoothed version of the Mastcam values), the uncertainty in the absolute value of the retrieved optical depth is simply the uncertainty in the Mastcam opacity, which is typically 0.03 (Lemmon et al., 2015). However, variations in retrieved optical depth from sol to sol are independent of the Mastcam calibration because only the calibration values convolved over 100 sols are used in the retrieval. To estimate uncertainties in sol-to-sol variations of aerosol optical depth, we use the differences in the retrieved optical depth between different REMS UV photodiodes as a guide since they should all give the same value in theory. The three longest wavelength photodiodes (UVA, UVB, and UVE; See Fig. 1) are very highly correlated with a root-mean-squared difference in retrieved aerosol optical depth of about 0.02. The other three photodiodes (UVC, UVD, UVABC) show greater variations with root-mean-squared differences between photodiode pairs of up to 0.05 (see also the discussion in Section 5.4).

M

4. Results

PT

ED

Here we present results for the retrieval of aerosol optical depth from the REMS UV observations. The retrievals presented here cover approximately 1.75 Martian years of observations, from sols 9 to 1159 (MY 31, Ls=155° to MY 33, Ls=66°) including two full dust storm seasons. Aerosol optical depths are referenced to a wavelength of 880 nm, consistent with previously published values derived from direct imaging of the Sun (e.g., Lemmon et al., 2004; 2015).

CE

4.1. Seasonal Cycle of Daily Average Aerosol Optical Depth

AC

Figure 10 shows the retrieved aerosol optical depths as a function of Ls. For clarity, only retrievals from the UVA, UVB, and UVE photodiodes are shown. Retrievals from the other three photodiodes are similar but have somewhat higher noise. The overall pattern as a function of season (Ls) is similar to what has been observed before by both orbiters (e.g., Smith et al., 2004; Smith 2009; McCleese et al., 2010; Kass et al., 2016) and rovers (e.g., Lemmon et al., 2004; 2015; Smith et al., 2006) for years without a global-scale dust storm. Aerosol optical depth gradually decreased between Ls=0° and 60°, finally reaching a minimum of about 0.4, which lasted until Ls=140°. Relatively small, short-lived increases in aerosol optical depth were observed, most notably at MY 32, Ls=87° when optical

ACCEPTED MANUSCRIPT

depth reached nearly 0.6 for a few sols. Several smaller events lasting two or three sols (e.g,, at MY 32, Ls=59° and 78°) are recorded in the retrievals of all the photodiodes and appear to be real events.

CR IP T

At Ls=140°, aerosol optical depth increased rapidly reaching a value of about 0.7 by Ls=150° in both Martian years. Perhaps coincidentally, retrievals from both Mars Years sampled here show a substantial but short-lived increase in optical depth above this new baseline soon afterward. During MY 31, optical depth increased to about 0.9 at Ls=161° before returning to 0.7 three sols later. During MY 32, optical depth increased to about 1.1 at Ls=158° with the elevated opacity lasting about a week. These events were likely caused by dust from regional storms being blown across the rover’s location in Gale Crater (e.g., Wang and Richardson, 2015).

PT

ED

M

AN US

After remaining at a baseline value of 0.7–0.8 during the period from Ls=140°–200°, aerosol optical depth increased in both years after Ls=200° to reach annual maximum values. The increase was much more abrupt in MY 31 with optical depth jumping from 0.7 to 1.2 over the course of four sols at Ls=207°. From there, optical depth varied between 1.0 and 1.35, but with no clear upward or downward trend until Ls=260°, when optical depth began a systematic drop falling to a minimum of 0.7 at Ls=295°. The increase in optical depth during MY 32 was much more gradual and systematic, rising almost monotonically to a peak value of about 1.6 at Ls=229° before once again falling to a minimum of 0.7 at Ls=305°. Notable, however, is that multiple short-lived spikes in optical depth punctuated the decrease in optical depth during MY 32. Although the spikes in optical depth lasted no more than two or three sols in each case, the increase in optical depth during the strongest ones at Ls=263° and 292° was significant, reaching about 1.5 and 1.1, respectively. A similar spike in optical depth was recorded during MY 31, but only at Ls=303° after optical depth had reached its local minimum.

AC

CE

The end of both Martian years featured renewed dust activity with optical depth reaching 1.2 at Ls=315° during MY 31 and multiple discrete peaks of about 1.3 between Ls=335° and 345° during MY 32. There was much more interannual variation during this period and during the time around the annual optical depth maximum than during other times of the year. 4.2. Short Timescale Variations in Daily Average Optical Depth Much of the above overall description of the annual cycle of aerosol optical depth at Gale Crater could have been written using the Mastcam measurements alone. However, the key advantage of the REMS UV dataset is the ability to produce reliable retrievals of aerosol optical depth for very nearly every sol during which MSL has taken science observations. This is because the REMS data are part of a background observation that is run at all times. On the other hand, although the Mastcam measurements are high quality,

ACCEPTED MANUSCRIPT

they are limited to those times when the observation sequence is specifically inserted into the daily plan for the rover. The period between Mastcam measurements can be anywhere from one to ten or more sols, although 3–7 sols is typical.

CR IP T

Figure 11 compares in detail the aerosol optical depth retrieved from the daily REMS UV photodiode observations against the periodic Mastcam images of the Sun. The high correlation between the three REMS UV retrievals shown argues for the validity of many of the small, short-timescale variations seen in the time series. Even when the retrieved optical depths from the three UV photodiodes diverge somewhat (e.g., MY 32, Ls=240°– 250°) the sol-to-sol pattern in the retrievals remains the same.

PT

ED

M

AN US

It is clear that in numerous cases short-lived increases in aerosol optical depth lasting only one to three sols were completely missed by the Mastcam measurements. Prominent examples during MY 31 are at Ls=160°, 304°, 310°, and 335°. Examples during MY 32 are at Ls=87°, 156°, 237°, and 292°. At other times, Mastcam measurements showed increased optical depth, but the REMS UV retrievals “fill in” the gaps providing much greater detail about the evolution of the dust storms, for example at MY 31, Ls=210° and 320°, and throughout the MY 32, Ls=330°–350° period. In one case, an isolated Mastcam measurement that produced what might have been considered an anomalously high value at MY 32, Ls=263° is confirmed by the REMS UV retrievals to have been a real shortlived increase in optical depth. In a few cases, the REMS UV retrievals are able to fill in abnormally long gaps in the Mastcam record, such as MY 31, Ls=293°–314°, MY 32, Ls=12°–27°, and MY 32, Ls=352° to MY 33, Ls=4°. Each of those gaps in the Mastcam record of aerosol optical depth lasted more than three weeks. It is important to note that REMS was able to continue recording its hourly background observations even during the communication blackout period around solar conjunction, which was the cause of the first and last of the long gaps in the Mastcam record noted above.

AC

CE

The frequency of the short-lived increases in aerosol optical depth as a function of season is shown in Fig. 12. An event is counted if the aerosol optical depth is more than 0.05 greater than a running average 5° of Ls wide. Although they can happen at any time of year, these short-timescale events are most common during the perihelion season (Ls=180°–360°) when overall dust optical depth is greatest. Indeed, none of the largest of these events with aerosol optical depth more than 0.10 greater than the running average (shown as the shaded bar graph in Fig. 12) have been observed during Ls=0°–135°. The fact that 23 of these short-lived increases in aerosol optical depth were observed per Mars Year, with many lasting just 2–3 sols, underscores the importance of the daily record of aerosol optical depth enabled by the REMS UV observations.

5. Discussion 5.1. Correspondence to REMS surface pressure observations

ACCEPTED MANUSCRIPT

CR IP T

The correspondence between dust optical depth and the amplitude of the diurnal and semidiurnal surface pressure tide has long been established through theory and observation (e.g., Zurek and Leovy, 1981; Bridger and Murphy, 1998). The specific relation between dust optical depth observed by Mastcam and surface pressure tides observed by REMS was discussed in detail by Guzewich et al. (2016). As reported in Guzewich et al. (2016), both the diurnal and semidiurnal tide amplitudes are very highly correlated with dust optical depth indicating that the dust optical depth at Gale Crater is largely representative of the globally-averaged value. This is not surprising given that orbital observations show that Gale Crater and the surrounding region is not a preferred location for large-scale dust lifting that causes regional dust storms (Wang and Richardson, 2015; Guzewich et al., 2015).

AN US

Although the overall seasonal trend of dust optical depth retrieved from REMS UV observations follows the typical patterns of the global dust climatology (e.g., Smith 2004; Lemmon et al., 2015; Kass et al., 2016), many of the short-lived variations in dust optical depth described in Section 4.2 appear to be a more local phenomenon. In general, there is little correlation between the short-lived dust events recorded by the REMS UV and the semidiurnal tide amplitude. This points towards the advection of local dust storm activity over the Gale Crater site as the most likely cause for most of the short-lived dust events.

AC

CE

PT

ED

M

It is worth noting that the daily record of aerosol optical depth enabled by the REMS UV observations provides important new information about one particular case described by Guzewich et al. (2016) at MY 32, Ls=150°–165°, in which the diurnal and semidiurnal tide amplitudes observed by REMS showed different time behavior. Given the available observations of aerosol optical depth from Mastcam images, the tidal behavior was interpreted as evidence that only minimal dust from a large regional dust storm observed by the MARCI camera on Mars Reconnaissance Orbiter (Malin et al., 2014) passed over Gale Crater. However, the REMS UV aerosol optical depth shows a significant shortlived increase at MY32, Ls=156° that was completely missed by the Mastcam observations (see Fig. 11). Therefore, it appears that dust from the regional dust storm did indeed pass over Gale Crater on days when there were no Mastcam observations of the Sun. For the second case discussed by Guzewich et al. (2016) at MY 32, Ls=263°, REMS UV observations provide corroborative evidence of a significant short-lived dust event that the tidal signature identifies as local in origin. 5.2. Analysis of Best-Fit Residuals Figure 13 shows the fractional residual for each individual REMS observation for the UVA, UVB, and UVE photodiodes for all sols. The fits are generally excellent, with fractional residuals (defined here as the observed signal minus the computed signal divided by the computed signal) less than a couple percent (i.e., green color in Fig. 13). However, at times there are systematic deviations in the fits on the order of 10–20%.

ACCEPTED MANUSCRIPT

CR IP T

These residuals are persistent over many sols and occur at consistent local times from sol to sol. There is also a high degree of correlation between the patterns in the residuals for the three photodiodes shown, which argues that the residuals are caused by a real physical effect not modeled by our initial assumptions. It is worth noting that correcting for azimuthal dependence of the photodiode spatial response (Section 3.2) is crucial, and without performing that correction the residuals are larger and there is essentially no correlation in the residuals between photodiodes.

AN US

One possible cause for the residuals is a change in aerosol optical depth during a sol. From Fig. 3, recall that increasing aerosol optical depth increases the diffuse radiation observed by REMS (at least until optical depth becomes very large), while at the same time it decreases the contribution from the direct solar beam. For the most part, the residuals are greatest early and late in the day. At these local times, the Sun is not in the nominal field of view, but computations show that there is still enough direct solar beam signal that reaches the photodiodes (e.g., the contribution shown in Fig. 4 for zenith angle greater than 30°) that the variation with optical depth from the direct solar beam still dominates and the net effect of increased aerosol optical depth is a decrease in observed flux. Therefore, the residuals in Fig. 13 could be interpreted in terms of changes in aerosol optical depth as negative values (blue colors) indicating increased optical depth and positive values (yellow and red) indicating decreased optical depth.

AC

CE

PT

ED

M

Another possible cause for the residuals is that the aerosol physical properties are not constant for all seasons and local times as we have assumed. The change in the aerosol physical properties could be a change in the composition (i.e., the relative mixing of dust vs. water ice aerosol), or in the aerosol particle size. There are strong reasons to conclude that this is the case rather than a change in the aerosol optical depth. There are several systematic trends in Fig. 13 that are common to the three REMS UV photodiodes. There is a clear tendency for the residuals to be symmetric about noon local time, with the residuals most apparent in the earliest (before 9:30) and latest (after 14:30) observations. It is possible that aerosol optical depth could change symmetrically about noon, but the more simple explanation is that this is a dependence on solar zenith angle, not local time, which would point to the cause being from a change in aerosol physical properties. Furthermore, the largest residuals occur during Ls=210°–260° and 320°–350° in both Mars Years, which is exactly the period during which there was increased dust activity and the highest aerosol optical depth (Fig. 10). Therefore, the residuals are most likely related to changes in aerosol physical properties caused by the increased dust activity. Figure 14 shows the same residuals as in Fig. 13, but now as a function of solar zenith angle instead of local time. The pattern of negative residual during the periods of highest aerosol optical depth is now even clearer. The residuals remain very small within the nominal field of view of the photodiodes (solar zenith angle less than 30°), and rapidly grow to about 20% at the highest solar zenith angles. The most straightforward change to

ACCEPTED MANUSCRIPT

CR IP T

aerosol physical properties that causes this type of variation is a change in aerosol particle size. Figure 15 shows the change in computed flux as a function of solar zenith angle for dust particles with effective radii of 1.0 and 2.5 µm compared to the nominal 1.5 µm that has been assumed in this work. This is for a particular case with aerosol optical depth unity and the spatial response function of the UVA photodiode, but the trends for other photodiodes and aerosol optical depths are very similar. A residual of -20% (i.e., a relative flux in Fig. 15 of 0.8) could be caused by dust particles that had effective radius 2.5 µm instead of the assumed 1.5 µm. This amount of variation in aerosol particle size is consistent with previous studies from the Mars Exploration Rovers (Wolff et al., 2006; Lemmon et al., 2015) and from orbiters (Wolff and Clancy, 2003; Guzewich et al., 2014).

ED

M

AN US

It is not possible to uniquely determine the aerosol particle size from these residuals since any mismatch in the baseline or time-averaged particle size from the 1.5 µm that was assumed will be compensated for in the spatial response functions that we derive. However, it can be concluded from these data that the aerosol particle size increased (leading to negative residuals) during the times of dust activity when retrieved aerosol optical depth was highest, and that the increase in effective particle radius was about 60% . There is also a suggestion in Fig. 15 that there are generally positive residuals at high solar zenith angles between Ls=60° and 150°, which would imply smaller effective particle radius during that season. This seasonal dependence of aerosol particle size is expected as dust activity injects larger particles into the air, while larger aerosols settle out of the atmosphere more quickly leading to a smaller average particle size over time (Kahre et al., 2008). 5.3. Time Dependence of the Dust Fallout Calibration Factor

CE

PT

The dependence of the dust fallout calibration factor with time for the UVA, UVB, and UVE photodiodes is shown in Fig. 16. For clarity, only the time-smoothed version that is used in the retrieval is shown. Over time, dust falls onto the REMS UV photodiodes partially blocking their field of view. Modeled fluxes are multiplied by this calibration factor to compare against observations.

AC

The slope of the calibration factor with time is indicative of the rate of dust deposition on the REMS UV photodiode area. Dust deposition begins almost immediately and increases notably with the dust activity at MY 31, Ls=205°. The rate of dust deposition slows for a while around MY 31, Ls=270°, but then increases again when aerosol opacity is high after MY 31, Ls=300°. Dust deposition continues at a slower rate until about MY 32, Ls=240° when the calibration factor has dropped to about 0.7. Somewhat surprisingly, this was followed by an extended period of dust cleaning between MY 32, Ls=240° and 300° with the dust fallout calibration factor climbing by about 0.15 units back to near 0.85. The onset of new dust activity at the end of MY 32 brought new dust deposition and a renewed decrease in the dust fallout calibration factor.

ACCEPTED MANUSCRIPT

AN US

CR IP T

The apparent dust cleaning between MY 32, Ls=240° and 300° (see also Fig. 7) occurred over an extended period of time and similarly for all six photodiodes. Although not as obvious, the Ls=240°–300° season during the previous Mars Year (MY 31) also experienced some cleaning in the form of a slower rate of decrease in the dust fallout calibration factor during that period. No other season shows a notable amount of dust cleaning. The cleaning of dust from the photodiodes appears to be caused by winds that are strong enough to exceed the threshold for lifting dust off the rover deck. This analysis of the REMS UV data is another strong piece of evidence that the season around Ls=270° is special in terms of its ability to raise dust and clean the photodiodes. The modeled dust devil activity at Gale Crater also shows peak activity at Ls=270°, although the number of convective vortices observed by REMS did not peak at this season, probably because at that time the rover was at a location with a micro-environment unfavorable for vortex formation (Kahanpää et al., 2016). Furthermore, both observations of line-of-sight optical depth to the crater wall by Moore et al. (2016) and modeling of the mesoscale circulation around the Gale Crater site by Rafkin et al. (2016) indicate that only during the season around Ls=270° does the air above and outside the crater mix with air inside the crater, with the model predicting the highest wind speeds and surface stresses of the year.

ED

M

Observations of wind speed are available from the REMS instrument suite, but are inconclusive about the season when wind speeds are greatest. This uncertainty comes in part from biases caused by the loss of the slightly rear-facing wind sensor boom on landing (which means that winds coming from anywhere to the rear of the rover are lost) and the seasonally varying fraction of each day when the rover is warm enough to provide reliable wind speed observations.

AC

CE

PT

Modeled as an exponential decay, over the course of the first Martian Year of observations the dust fallout calibration factor decreased with an e-folding timescale of a bit more than three Martian Years. This is far longer than the Cunningham Stokes settling time for particles in the typical size range (1–2 µm) for Mars aerosols (Pruppacher and Klett, 1997; Kahre et al., 2008), and is also obviously far longer than the typical clearing time for even global-scale dust storms (e.g., Smith, 2008 and references therein). Partially this is because there is less than a unit aerosol optical depth available for deposition, but also each UV photodiode has a magnet around it to deflect magnetic dust (Gómez-Elvira et al., 2012), which has proven to be effective. At a dust fallout calibration factor of 0.7 the REMS UV data are still extremely useful, and we anticipate that they would still be highly useful at a calibration factor of 0.5 or even lower. 5.4. Differences Between UV Photodiodes Although the six photodiodes are mounted near each other on the rover deck (Fig. 6), they appear to have experienced somewhat different dust deposition rates. In particular,

ACCEPTED MANUSCRIPT

the UVC photodiode has the lowest dust calibration factor (and therefore the highest dust deposition), while the UVA photodiode has the highest dust calibration factor.

CR IP T

As with sol-to-sol variation, the highest correlation coefficient of residuals between pairs of photodiodes is between the UVA, UVB, and UVE photodiodes, which are at a similar wavelength (Fig. 1). The UVC and UVD photodiodes have higher correlation with each other than with the other photodiodes, while the UVABC photodiode is correlated most closely with UVC and UVE.

AN US

Our method of tying the dust fallout calibration for all photodiodes to the Mastcam measurements minimizes any possible difference between different photodiodes, and in theory they should all give the same result. In practice, because of their higher dust fallout calibration factors, the high correlation in sol-to-sol variation and residual, and their close correspondence in central wavelength, we choose the UVA, UVB, and UVE photodiodes as being the best set for determination of aerosol optical depth. We recommend that the mean of the UVA, UVB, and UVE retrievals can be used as a best estimate of aerosol optical depth, and that the spread and sol-to-sol behavior of the three can be used to estimate uncertainty in optical depth and to evaluate whether variations over short timescales are real events or noise. A table with these data is included as supplementary material to this article.

M

6. Summary

AC

CE

PT

ED

Systematic observations taken by the REMS UV photodiodes on a daily basis throughout the landed MSL mission enable a detailed record of aerosol optical depth to be retrieved. The period presented here covers approximately 1.75 Mars Years including two full dust storm seasons. During each of the two Mars Years observed (MY 31 and 32), aerosol optical depth rises abruptly after Ls=140°, which is followed by a short spike in optical depth. Another abrupt rise in aerosol optical depth occurred each year after Ls=200° with maximum annual background values recorded between Ls=210° and 240°. After falling to a local minimum at around Ls=300°, aerosol optical depth again increased to elevated values between Ls=315° and 350°, which was followed by a long, gradual decline in optical depth until Ls=140° the following year. Importantly, the REMS UV retrievals enable the characterization of numerous short-lived, but significant increases in aerosol optical depth that punctuate the annual background variation of aerosol optical depth. Although the measurements of aerosol optical depth provided by Mastcam are of high quality, they often completely miss these short-lived events, which are important for understanding the dynamics of aerosols in the Gale Crater environment. Dust deposition on the REMS UV photodiodes has proceeded at a generally slow rate. Cleaning of dust from the photodiodes was observed in the season around Ls=270° (with

ACCEPTED MANUSCRIPT

an especially strong and extended cleaning event occurring during MY 32 between Ls=240° and 300°), but is not observed during other seasons. This adds another strong piece of evidence that the season around Ls=270° is special in terms of its ability to raise dust and the connection of the circulation within the crater to the general circulation outside the crater during this season.

CR IP T

Looking in detail at the residuals from the retrieval, there is a clear indication that aerosol effective particle sizes are larger on average during the periods with highest optical depth (Ls=210°–260° and 320°–350°). Conversely, there is no clear signal of significant systematic changes in aerosol optical depth as a function of local time.

AN US

The REMS UV photodiodes continue to collect systematic, daily observations extending the existing record of aerosol optical depth reported here. Although deposition of dust onto the instrument continues, given past trends in the dust fallout calibration factor we anticipate that the REMS UV observations will continue to be highly useful for years to come and will provide important information about the interannual variation of aerosol properties above Gale Crater.

Acknowledgements

AC

CE

PT

ED

M

We would like to thank the MSL and REMS operations teams and the MSL ENV science theme group for their work in collecting and processing the REMS UV dataset. We thank the MAHLI team for periodic imaging of the REMS UV photodiodes. Claire Newman and an anonymous referee provided helpful reviews of this manuscript. Smith acknowledges funding from the MSL project as a Participating Scientist.

ACCEPTED MANUSCRIPT

References Arvidson, R.E., et al., 2014. Mars Reconnaissance Orbiter and Opportunity observations of the Burns formation: Crater hopping at Meridiani Planum. J. Geophys. Res. 120, 429–451. http://dx.doi.org/10.1002/2004JE004686.

CR IP T

Clancy, R.T., et al., 2000. An intercomparison of ground-based millimeter, MGS TES, and Viking atmospheric temperature measurements: Seasonal and interannual variability of temperatures and dust loading in the global Mars Atmosphere. J. Geophys. Res. 105, E4, 9553–9571.

Gierasch, P.J., Goody, R.M., 1972. The effect of dust on the temperature of the martian atmosphere. J. Atmos. Sci. 29, 400–402, http://dx.doi.org/10.1175/15200469(1972)029<0400:TEODOT>2.0.CO;2.

AN US

Gómez-Elvira, J., et al., 2012. REMS: The environmental sensor suite for the Mars Science Laboratory rover. Space Sci. Rev. 170, 583–640. http://dx.doi.org/10.1007/s11214-012-9921-1.

Goody, R.M., Yung, Y.L., 1989. Atmospheric Radiation: Theoretical Basis, Oxford Univ. Press, New York. Grotzinger, J.P., et al, 2012. Mars Science Laboratory mission and science investigation. Space Sci. Rev. 170, 5–56. http://dx.doi.org/10.1007/s11214-012-9892-2.

ED

M

Guzewich, S.D., Smith, M.D., Wolff, M.J., 2014. The vertical distribution of Martian aerosol particle size. J. Geophys. Res. 119, 2694–2708. http://dx.doi.org/10.1002/2014JE004704.

PT

Guzewich, S.D., Toigo, A.D., Kulowski, L., Wang, H., 2015. Mars Orbiter Camera climatology of textured dust storms. Icarus 258, 1–13. http://dx.doi.org/10.1016/j.icarus.2015.06.023. Guzewich, S.D., et al., 2016. Atmospheric tides in Gale Crater, Mars. Icarus 268, 37–49. http://dx.doi.org/10.1016/j.icarus.2015.12.028.

AC

CE

Haberle, R.M., et al., 2014. Preliminary interpretation of the REMS pressure data from the first 100 sols of the MSL mission. J. Geophys. Res. 119, 440–453. http://dx.doi.org/10.1002/2013JE004488. Harri, A.-M., et al., 2014. Pressure observations by the Curiosity rover: Initial results. J. Geophys. Res. 119, 82–92. http://dx.doi.org/10.1002/2013JE004423. Johnson, J.R. et al., 2006a. Spectrophotometric properties of materials observed by Pancam on the Mars Exploration Rovers: 1. Spirit. J. Geophys. Res. 111, E02S14. http://dx.doi.org/10.1029/2005JE002494. Johnson, J.R. et al., 2006b. Spectrophotometric properties of materials observed by Pancam on the Mars Exploration Rovers: 2. Opportunity. J. Geophys. Res. 111, E12S16. http://dx.doi.org/10.1029/2006JE002762.

ACCEPTED MANUSCRIPT

Kahanpää, H., et al., 2016. Convective vortices and dust devils at the MSL landing site: Annual variability. Submitted to J. Geophys. Res. Kahn, R., Martin, T.Z., Zurek, R.W., Lee, S.W., 1992. The Martian dust cycle. In: Mars, Univ. Arizona Press, Tucson.

CR IP T

Kahre, M.A., Murphy, J.R., Haberle, R.M., 2006. Modeling the martian dust cycle and surface dust reservoirs with the NASA Ames general circulation model. J. Geophys. Res. 111, E06008, http://dx.doi.org/10.1029/2005JE002588. Kahre, M.A., Hollingsworth, J.L., Haberle, R.M., Murphy, J.R., 2008. Investigations of the variability of dust particle sizes In the Martian atmosphere using the NASA Ames General Circulation Model. Icarus 195, 576–97, http://dx.doi.org/10.1026/j.icarus. 2008.01.023.

AN US

Kass, D.M., Kleinböhl, A., McCleese, D.J., Schofield, J.T., Smith, M.D., 2016. Interannual similarity in the Martian atmosphere during the dust storm season. Geophys. Res. Lett, in press. Kloos, J.L., et al., 2016. The first Martian year of cloud activity from Mars Science Laboratory (sol 0–800). Adv. Space Res. 57, 1223–1240. http://dx.doi.org/10.1016/j.asr.2015.12.040.

M

Lemmon, M.T. et al., 2004. Atmospheric imaging results from the Mars Exploration Rovers: Spirit and Opportunity. Science 306, 1753–1756. http://dx.doi.org/10.1126/science.1104474.

ED

Lemmon, M.T., et al., 2015. Dust aerosols, clouds, and the atmospheric optical depth record over 5 Mars years of the Mars Exploration Rover mission. Icarus 251, 96– 111. http://dx.doi.org/10.1016/j.icarus.2014.03.029.

PT

Madeleine, J.-B., F. Forget, E. Millour, L. Montabone, M.J. Wolff, 2011. Revisiting the radiative impact of dust on Mars using the LMD Global Climate Model. J. Geophys. Res., 116, E11010, http://dx.doi.org/10.1029/2011JE003855.

CE

Malin, M.C., et al., 2014. MRO MARCI Weather Report for the Week of 7 July 2014–13 July 2014, Malin Space Science Systems Captioned image Release, MSSS-334, http://www.msss.com/msss_images/2014/07/16/.

AC

Martín-Torres, F.J., et al., 2015. Transient liquid water and water activity at Gale crater on Mars. Nature Geoscience 8, 357–361. http://dx.doi.org/10.1038/ngeo2412. Martínez, G.M., et al., 2016. Likely frost events at Gale crater: Analysis from MSL/REMS measurements. Icarus in press. http://dx.doi.org/10.1016/j.icarus.2015.12.004. Mason, E.L., Lemmon, M.T., de la Torre-Juárez, M., Vicente-Retortillo, A., Martínez, G., 2015. Constraining atmospheric particle size in Gale Crater using REMS UV measurements and Mastcam observations at 440 and 880 nm. Abstract P23B-

ACCEPTED MANUSCRIPT

2136 presented at the 2015 Fall Meeting, AGU, San Francisco, Calif., 14–18 December. McCleese, D.J., et al, 2010. Structure and dynamics of the Martian lower and middle atmosphere as observed by the Mars Climate Sounder: Seasonal variations in zonal mean temperature, dust, and water ice aerosols. J. Geophys. Res. 115, E12016. http://dx.doi.org/10.1029/2010JE003677.

CR IP T

Montabone L. et al., 2015. Eight-year climatology of dust on Mars. Icarus 251, 65–95. http://dx.doi.org/10.1016/j.icarus.2014.12.034.

Moore, C.A., et al., 2016. A full martian year of line-of-sight extinction within Gale Crater, Mars as acquired by the MSL Navcam through sol 900. Icarus 264, 102– 108. http://dx.doi.org/10.1016/j.icarus.2015.09.001.

AN US

Moores, J.E. et al., 2015. Atmospheric movies acquired at the Mars Science Laboratory landing site: Cloud morphology, frequency and significance to the Gale Crater water cycle and Phoenix mission results. Adv. Space Res. 55, 2217–2238. Newman, C.E., Lewis, S.R., Read, P.L., Forget, F., 2002. Modeling the martian dust cycle. 2. Multiannual radiatively active dust transport simulations. J. Geophys. Res. 107 (E12), 5124, http://dx.doi.org/10.1029/2002JE001920.

M

Pruppacher, H., Klett, J., 1997. Microphysics of clouds and precipitation. Kluwer Academic Press.

ED

Rafkin, S.C.R., et al., 2016. The meteorology of Gale Crater as determined from Rover Environmental Monitoring Station observations and numerical modeling. Part II: Interpretation. Icarus in press. http://dx.doi.org/10.1016/j.icarus.2016.01.031.

PT

Smith, M.D., 2004. Interannual variability in TES atmospheric observations of Mars during 1999–2003. Icarus 167, 148–165. http://dx.doi.org/10.1016/j.icarus.2003.09.010.

CE

Smith, M.D., 2008. Spacecraft observations of the martian atmosphere. Ann. Rev. Earth Planet. Sci. 36, 191–219. http://dx.doi.org/10.1146/annurev.earth.36.031207.124334.

AC

Smith M.D., 2009. THEMIS observations of Mars aerosol optical depth from 2002–2008. Icarus 202, 444–452. http://dx.doi.org/10.1016/j.icarus.2009.03.027. Smith, P.H., Lemmon, M., 1999. Opacity of the martian atmosphere measured by the Imager for Mars Pathfinder. J. Geophys. Res. 104 (E4), 8975–8985. http://dx.doi.org/10.1029/1998JE900017. Smith, M.D., Bandfield, J.L., Christensen, P.R., 2000. Separation of atmospheric and surface spectral features in Mars Global Surveyor Thermal Emission Spectrometer (TES) spectra. J. Geophys. Res. 105, 9589–9608.

ACCEPTED MANUSCRIPT

Smith, M.D., et al., 2004. First atmospheric science results from the Mars Exploration Rovers Mini-TES. Science 306, 1750–1753. http://dx.doi.org/10.1126/science.1104257. Smith, M.D., et al., 2006. One martian year of atmospheric observations using MER Mini-TES. J. Geophys. Res. 111, E12S13. http://dx.doi.org/10.1029/2006JE02770.

CR IP T

Smith, M.D., Wolff, M.J., Clancy, R.T., Kleinböhl, A., Murchie, S.L., 2013. Vertical distribution of dust and water ice aerosols from CRISM limb-geometry observations. J. Geophys. Res. 118, 1–14. http://dx.doi.org/10.1002/jgre.20047.

Soderblom, J.M., Bell III, J.F., Johnson, J.R., Joseph, J., Wolff, M.J., 2008. Mars Exploration Rover Navigation Camera in-flight calibration. J. Geophys. Res. 113, E06S19, http://dx.doi.org/10.1029/2007JE003003.

AN US

Spurr, R.J.D., 2002. Simultaneous derivation of intensities and weighting functions in a general pseudo-spherical discrete ordinate radiative transfer treatment. J. Quant. Spectrosc. Rad. Transf. 75, 129–175. Thomas, G.E., Stamnes, K., 1999. Radiative transfer in the atmosphere and ocean. Cambridge Univ. Press, Cambridge. Thomas, N., et al., 1999. The color of the Martian sky and its influence on the illumination of the Martian surface. J. Geophys. Res. 104, 8795–8808.

ED

M

Thomas, N., Jorda, L., Markiewicz, W.J., 2000. Effect of diffuse sky brightness on the spectrophotometry of rough Martian surfaces. J. Geophys. Res. 105, 26739– 26744.

PT

Vicente-Retortillo, A., et al., 2015. UV opacity at Gale Crater from MSL/REMS measurements. Abstract P23B-2140 presented at the 2015 Fall Meeting, AGU, San Francisco, Calif., 14–18 December.

CE

Wang, H., Richardson, M.I., 2015.The origin, evolution, and trajectory of large dust storms on Mars during Mars years 24–30 (1999–2011). Icarus 251, 112–127. http://dx.doi.org/10.1016/j.icarus.2013.10.033.

AC

Webster, C.R., et al., 2015. Mars methane detection and variability at Gale crater (supplementary materials). Science 347, 415–417. http://www.sciencemag.org/content/347/6220/415/suppl/DC1. Whiteway, J.A., et al., 2006. Mars water ice clouds and precipitation. Science 325, 68–70. http://dx.doi/org/10.1126/science.1172344. Wolff, M.J., Clancy, R.T., 2003. Constraints on the size of martian aerosols from Thermal Emission Spectrometer observations. J. Geophys. Res. 108, 5097. http://dx.doi.org/10.1029/2003JE002057.

ACCEPTED MANUSCRIPT

Wolff, M.J., et al., 2006. Constraints on dust aerosols from the Mars Exploration Rovers using MGS overflights and Mini-TES. J. Geophys. Res. 111, E12S17, http://dx.doi.org/10.1029/2006JE002786. Wolff, M.J., et al., 2009. Wavelength dependence of dust aerosol single scattering albedo as observed by the Compact Reconnaissance Imaging Spectrometer. J. Geophys. Res. 114, E00D04, http://dx.doi.org/10.1029/2009JE00350.

CR IP T

Wolff, M.J., Clancy, R.T., Goguen, J.D., Malin, M.C., Cantor, B.A., 2010. Ultraviolet dust aerosol parameters as observed by MARCI. Icarus 208, 143–155. http://dx.doi.org/10.1016/j.icarus.2010.01.010.

Zorzano, M.-P., Córdoba-Jabonero, C., 2007. Influence of aerosol multiple scattering of ultraviolet radiation on martian atmospheric scattering. Icarus 190, 495–503. http://dx.doi.org/10.1016/j.icarus.2007.03.029.

AN US

Zorzano, M.-P., Vázquez, L., Jimenez, S., 2009. Retrieval of ultraviolet spectral irradiance from filtered photodiode measurements. Inverse Prob. 25, 115023. http://dx.doi.org/10.1088/0266-5611/25/11/115023.

AC

CE

PT

ED

M

Zorzano, M.-P., Martín-Soler, J., Gómez-Elvira, J., 2011. UV photodiodes response to non-normal, non-collimated and diffusive sources of irradiance, Photodiodes Communications, Bio-Sensings, Measurements and High-Energy Physics, Associate Professor Jin-Wei Shi (Ed.), ISBN: 978-953-307-277-7, InTech, Available from: http://www.intechopen.com/books/photodiodes-communicationsbio-sensings- measurements-and-high-energy-physics/uv-photodiodes-responseto-non-normal-non-colimated-and- diffusive-sources-of-irradiance

CE

PT

ED

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

AC

Figure 1. The spectral response of the six REMS UV photodiodes.

PT

ED

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

AC

CE

Figure 2. The observed signal from the REMS UVA photodiode for sols 62 and 91 binned every three minutes. The signal here is the raw photodiode output current in nanoamperes. The available observations contain both five-minute blocks taken at the beginning of each hour, and longer blocks covering full hours. Only observations taken with solar zenith angle less than 60° in rover coordinates are shown. The red points on sol 62 are cases where the UVA field of view was partially obscured by shadows.

AC

CE

PT

ED

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

Figure 3. (top) The computed flux within the nominal 60° cone field of view for a REMS UV photodiode for a typical case with aerosol optical depth, �, of unity and a solar zenith angle of 10°. The color scale is logarithmic. (bottom) The computed direct solar beam and diffuse flux for the REMS UVA photodiode as a function of aerosol (dust) optical depth.

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

Figure 4. The empirically derived spatial response for the six REMS UV photodiodes.

AC

CE

PT

ED

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

Figure 5. The azimuth correction derived from the retrieval residuals for each UV photodiode.

AN US

CR IP T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

Figure 6. Images of the REMS UV photodiodes taken by the MAHLI camera. The accumulation of dust over time is evident. The circular patterns of dust around each photodiode are caused by magnets that deflect magnetic dust away from the photodiodes. Sol 36 is MY 31, Ls=170°. Sol 154 is MY 31, Ls=243°. Sol 323 is MY 31, Ls=346°. Sol 660 is MY 32, Ls=146°. Sol 881 is MY 32, Ls=281°. Sol 1097 is MY 33, Ls=38°.

CE

PT

ED

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

AC

Figure 7. The multiplicative dust fallout calibration factor derived for each Mastcam image of the Sun (black points), and the smoothed representation (red line) that is used in the retrieval.

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

Figure 8. The coverage in Ls and local time for all available REMS UV observations with solar zenith angle less than 60° and no obscurations from shadows. The REMS observations are commanded in local mean solar time, which accounts for the wavy lines spaced one hour apart.

AN US

CR IP T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

Figure 9. Examples of the retrieval best fit for UVA observations taken on sols 62 and 91. Only observations with no obscurations from shadows are shown. The quality of the fits is excellent with residuals of less than a couple percent.

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

Figure 10. The daily retrieved values of aerosol optical depth for the UVA, UVB, and UVE photodiodes. The UVC, UVD, and UVABC retrievals produce similar (although somewhat more noisy) results, and are not shown for clarity.

AC

CE

PT

ED

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

Figure 11. Comparison of the retrieved aerosol optical depth from the REMS UVA, UVB, and UVE photodiodes (black, red, and blue lines, respectively) with the aerosol optical depth derived from direct imaging of the Sun by Mastcam (green points). The overall seasonal cycle is punctuated by numerous significant increases in optical depth on short timescales that are often not characterized by Mastcam.

AN US

CR IP T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

Figure 12. The frequency of short-lived events per Mars Year in a given bin of 45° of Ls. The upper line shows the number of events with aerosol optical depth greater than 0.05 above the background level. The shaded bars show the number of events with aerosol optical depth greater than 0.10 above the background level.

AC

CE

PT

ED

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

Figure 13. The residuals from the retrieval of aerosol optical depth for the UVA, UVB, and UVE photodiodes as a function of Ls and local time. Note the overall low values of the residuals and the high correlation between the patterns from the three photodiodes.

AC

CE

PT

ED

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

Figure 14. The residuals from the retrieval of aerosol optical depth for the UVA, UVB, and UVE photodiodes as a function of Ls and solar zenith angle. The clear seasonal variation in residuals at high solar zenith angle are associated with the timing of increased aerosol optical depth and are indicative of changes in the aerosol effective particle size.

M

AN US

CR IP T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

Figure 15. The computed change in UVA flux for dust with effective particle sizes of 1.0 µm and 2.5 µm compared to the assumed value of 1.5 µm. The negative residuals apparent in Fig. 14 correspond to periods when the aerosol has larger effective particle size.

AN US

CR IP T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

Figure 16. The smoothed version of the dust fallout calibration factor for the REMS UVA, UVB, and UVE photodiodes that is used in the retrieval. The absolute values of the three curves are different, but the overall time dependence is very similar. Note the apparent strong clearing of dust from the photodiodes between Ls=240° and 300° during the second year (MY 32).