DISSERTATION INTRASEASONAL VARIABILITY IN THE TROPICAL DIURNAL CYCLE Submitted by Michael B. Natoli Department of Atmospheric Science In partial fulfillment of the requirements For the Degree of Doctor of Philosophy Colorado State University Fort Collins, Colorado Spring 2022 Doctoral Committee: Advisor: Eric D. Maloney Michael M. Bell David A. Randall Jeffrey D. Niemann Copyright by Michael B. Natoli 2022 All Rights Reserved ABSTRACT INTRASEASONAL VARIABILITY IN THE TROPICAL DIURNAL CYCLE The relationship between large-scale intraseasonal variability in tropical convection and the lo- cal diurnal cycle on tropical islands is explored with observations and an idealized model. In part one, the impact of quasi-biweekly variability in the monsoon southwesterly winds on the precipita- tion diurnal cycle in the Philippines is examined using CMORPH precipitation, ERA5 reanalysis, and outgoing longwave radiation (OLR) fields. Both a case study during the 2018 Propagation of Intraseasonal Tropical Oscillations (PISTON) field campaign and a 23-year composite analysis are used to understand the effect of the QBWO on the diurnal cycle. QBWO events in the west Pa- cific, identified with an extended EOF index, bring increases in moisture, cloudiness, and westerly winds to the Philippines. Such events are associated with significant variability in daily mean pre- cipitation and the diurnal cycle. It is shown that the modulation of the diurnal cycle by the QBWO is remarkably similar to that by the boreal summer intraseasonal oscillation (BSISO). The diurnal cycle reaches a maximum amplitude on the western side of the Philippines on days with average to above average moisture, sufficient insolation, and weakly offshore prevailing wind. This occurs during the transition period from suppressed to active large-scale convection for both the QBWO and BSISO. Westerly monsoon surges associated with QBWO variability generally exhibit active precipitation over the South China Sea (SCS), but a depressed diurnal cycle. These results high- light that modes of large-scale convective variability in the tropics can have a similar impact on the diurnal cycle if they influence the local scale environmental background state similarly. In part two, a specific large-scale mode is neglected, and the impact of variability in the back- ground wind at any timescale on the local diurnal cycle is isolated. Luzon Island in the northern Philippines is used as an observational test case. Composite diurnal cycles of CMORPH precipita- tion are constructed based on an index derived from the first empirical orthogonal function (EOF) ii of ERA5 zonal wind profiles. A strong precipitation diurnal cycle and pronounced offshore prop- agation in the leeward direction tends to occur on days with a weak, offshore prevailing wind. Strong background winds, particularly in the onshore direction, are associated with a suppressed diurnal cycle. Idealized high resolution 2-D Cloud Model 1 (CM1) simulations test the dependence of the diurnal cycle on environmental wind speed and direction by nudging the model base-state toward to composite profiles derived from the reanalysis zonal wind index. These simulations can qualitatively replicate the observed development, strength, and offshore propagation of diurnally generated convection under varying wind regimes. Under strong background winds, the land-sea contrast is reduced, which leads to a substantial reduction in the strength of the sea-breeze circu- lation and precipitation diurnal cycle. Weak offshore prevailing winds favor a strong diurnal cycle and offshore leeward propagation, with the direction of propagation highly sensitive to the back- ground wind in the lower free troposphere. Offshore propagation speed appears consistent with density current theory rather than a direct coupling to a single gravity wave mode, though several gravity wave modes apparent in the model likely contribute to a destabilization of the offshore environment. In part three, the hypotheses developed in parts one and two regarding the mechanisms regu- lating the diurnal cycle response are rigorously tested. A novel probabilistic framework is applied to the Luzon test case to improve the understanding of diurnal cycle variability. High amplitude diurnal cycle days tend to occur with weak to moderate offshore low-level wind and near to above average column moisture in the local environment. The transition from the BSISO suppressed phase to the active phase is most likely to produce the wind and moisture conditions supportive of a substantial diurnal cycle over western Luzon and the South China Sea (SCS). Thus, the impact of the BSISO on the local diurnal cycle can be understood in terms of the change in the probability of favorable environmental conditions. Idealized high-resolution 3-D Cloud Model 1 (CM1) sim- ulations driven only by a base-state derived from BSISO composite profiles are able to reproduce several important features of the observed diurnal cycle variability with BSISO phase, including the strong, land-based diurnal cycle and offshore propagation in the transition phases. Background iii wind appears to be the primary variable controlling the diurnal cycle response, but ambient mois- ture distinctly reduces precipitation strength in the suppressed BSISO phase, and enhances it in the active phase. A land-breeze, lingering deep convection over land after sunset, and strong me- chanical convergence appear to all be required in order to produce offshore propagation in CM1. Simulations in which the diurnal cycle of insolation is removed suggest the potential for a natural timescale for convective regeneration related to the island size. iv ACKNOWLEDGEMENTS I first want to thank my advisor, Prof. Eric Maloney for his support through difficult times and personal challenges, as well as his steady advice, and the professional opportunities he has given me. My committee members, Profs. Michael Bell, Dave Randall, and Jeff Niemann, also provided invaluable feedback through the duration of my time in graduate school. The time they have dedicated to my success is greatly appreciated. I would also like to thank Prof. Susan van den Heever, Dr. Leah Grant, and Prof. Russ Schumacher at Colorado State University for their helpful advice and guidance in designing and running the model experiments. This work was supported by the Office of Naval Research (ONR) under the Propagation of Tropical Intraseasonal Oscillations (PISTON) project N00014-16-1-3087, the NOAA CVP program under grant NA18OAR4310299, NASA CYGNSS grant 80NSSC21K1004, and the Climate and Large Scale Dynamics Program of the National Science Foundation under grant AGS-1735978. v TABLE OF CONTENTS ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Chapter 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Large-Scale Modes of Tropical Convective Variability . . . . . . . . . . . 2 1.2 Diurnal Cycle Variability . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.3 Scope and Significance of This Work . . . . . . . . . . . . . . . . . . . . 5 Chapter 2 The Quasi-Biweekly Oscillation and the Philippines Diurnal Cycle . . . . . . 8 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2 Data and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2.1 Data Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2.3 QBWO Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.3.1 2018 PISTON Case Study . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.3.2 Large Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.3.3 Luzon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.3.4 Mindanao . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 2.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.4.2 QBWO and BSISO Similarities . . . . . . . . . . . . . . . . . . . . . . 40 2.4.3 QBWO and BSISO Differences . . . . . . . . . . . . . . . . . . . . . . 41 2.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Chapter 3 The Tropical Diurnal Cycle Under Varying States of the Monsoonal Back- ground Wind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 3.2 Data and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 3.2.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 3.2.2 Binning Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 3.2.3 CM1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 3.3 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 3.3.1 Daily Mean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 3.3.2 Diurnal Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 3.4 CM1 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 3.4.1 Simulation Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 3.4.2 Land-Sea-Breeze Circulation . . . . . . . . . . . . . . . . . . . . . . . 64 3.4.3 Gravity Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 vi 3.4.4 Direction of Propagation Sensitivity Experiments . . . . . . . . . . . . 74 3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Chapter 4 The Mechanisms Involved in the Modulation of the Tropical Island Diurnal Cycle by the Boreal Summer Intraseasonal Oscillation . . . . . . . . . . . . 83 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 4.2 Data and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 4.2.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 4.2.2 CM1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 4.3 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 4.4 CM1 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 4.4.1 Control Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.4.2 Sensitivity Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . 108 4.4.3 Impacts of Island Size and Diurnal Variation . . . . . . . . . . . . . . . 116 4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 Chapter 5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 vii LIST OF TABLES 4.1 Summary of CM1 simulations examined in this study. . . . . . . . . . . . . . . . . . . 91 viii LIST OF FIGURES 2.1 NOAA ETOPO2 Topography (in meters) over the Philippines, with boxes of spatial averaging and important geographic features noted. The track of the R/V Thomas G. Thompson during the August-October 2018 PISTON field campaign is also shown in purple, with August in the darkest color and October in the lightest. . . . . . . . . . . 12 2.2 (a) Power spectrum of ERA5 850-mb zonal wind averaged inside box L in Fig. 2.1 during June-September (JJAS) 1998-2020 (blue), with theoretical red noise spectrum (red; Gilman et al. 1963), and its 90% confidence bound calculated with an F-test (gray, dotted). (b) As in (a) but for Mindanao, averaged inside box M. . . . . . . . . . 15 2.3 Spatial pattern at -4 (top), -2 (middle), and 0-day (bottom) lags from extended EOFs 1 (left) and 2 (right) of 10-20 day bandpass filtered AVHRR OLR anomalies in physical units (W m−2). The bottom row shows the difference between power spectra of each corresponding principal component time series and the corresponding 90% confidence bound of a theoretical red noise spectrum with the same autocorrelation as the PC time series. Values above zero (dotted red line) can be considered statistically significant at the 90% confidence level. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.4 Number of days in a certain Lee et al. (2013) BSISO phase distributed by active QBWO phase. The BSISO is considered to be inactive when the amplitude of the index is less than one. The darker color in the stacked bar chart indicates days that are classified as the same phase number in both indices. . . . . . . . . . . . . . . . . . . . 18 2.5 Time-height diagram of zonal wind from each sounding taken as part of the PISTON field campaign between 14 August 2018 and 13 October 2018. Soundings were taken every 12 hours from the island of Yap (top) and every 3 hours from the R/V Thomas G. Thompson during operational periods (bottom). . . . . . . . . . . . . . . . . . . . 20 2.6 Hovmöller plot of AVHRR OLR averaged between 0 and 25N at each longitude during 1 June-20 October 2018, bandpass filtered to the 10-20 day timescale using a Lanczos filter with 93 weights in W m−2 (left), and OLR anomalies from the seasonal cycle defined by the average daily climatology smoothed with a 7-day running mean (right). Named tropical cyclone tracks from IBTrACS are superimposed with gray dotted lines when the storm center was inside 0-25N. . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.7 Phase space diagram of the QBWO index activity from 14 August-14 October 2018 (the PISTON field campaign period), with the first principal component on the y-axis and the second principal component on the x-axis. The split between the 8 phases is denoted with black dotted lines, while days with an amplitude less than 1 (inside the center circle) are considered inactive, and not part of any phase. August is shown in the darkest color, with October in the lightest pink. The corresponding numbers indicate the date of each month. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 ix 2.8 (a) CMORPH precipitation rate estimates (mm/hr) averaged from 6N-8N across Min- danao (Box M, Fig. 2.1) from 11 September 2018 to 23 September 2018, during one full cycle of the QBWO index. The average topography in this box from NOAA ETOPO2 is shown on the bottom, with the coastlines drawn as vertical dashed black lines. Note that there are some land points west of the western coastline here, part of the Zamboanga Peninsula. The horizontal dashed black lines correspond to 00 UTC, or 0800 local time. (b) Zonal wind at 850-hPa averaged across both latitude and lon- gitude in Box M (red line) and total column water vapor (blue line) from ERA5, with the JJAS composite diurnal cycle removed at each hour. . . . . . . . . . . . . . . . . . 25 2.9 Composite maps by select QBWO phase over the West Pacific ocean of anomalies of OLR (W m−2) and vector anomalies of 850-mb wind from ERA5 (left column), and anomalies of ERA5 total column water vapor (kg m−2) with total 850-mb vector wind (right column). The total number of days in each composite can be ascertained from Figure 2.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 2.10 Anomalies in daily mean CMORPH precipitation rate composited by QBWO phase, with statistical significance at the 95% confidence level shown as dots. . . . . . . . . . 28 2.11 Anomalies in the amplitude of the CMORPH precipitation rate diurnal cycle by QBWO phase. Anomalies are calculated as the difference in diurnal amplitude between each phase composite, and amplitude of the JJAS composite diurnal cycle. Statistical sig- nificance at the 95% confidence level is shown as dots. . . . . . . . . . . . . . . . . . 29 2.12 Hovmöller diagrams of the composite diurnal cycle of CMORPH precipitation rate (mm/day) for select phases of the QBWO index. Precipitation rates are averaged across latitude in box L (Fig. 2.1), with corresponding longitude noted below. The average elevation of topography from NOAA ETOPO2 inside box L is shown at the bottom for reference. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 2.13 (a-c) Composites of the spatially averaged diurnal cycle of CMORPH precipitation rate (mm/day) in the QBWO (solid) and Lee et al. (2013) BSISO (dotted) phase with the highest diurnal range (orange), the QBWO/BSISO phase with the smallest diurnal range (blue), and the full JJAS composite (dotted black). Spatial averaging is done over ocean points inside box A (a), and land points inside boxes B (b) and C (c). (d-f) The corresponding daily mean precipitation (blue) and diurnal range (red) in mm/day of each phase’s spatially averaged composite diurnal cycle, by QBWO phase (solid) and BSISO phase (dotted). Each box covers a domain near Luzon. . . . . . . . . . . . 33 2.14 (a-c) Composites of the spatially averaged diurnal cycle of CMORPH precipitation rate (mm/day) in the QBWO (solid) and Lee et al. (2013) BSISO (dotted) phase with the highest diurnal range (orange), the QBWO/BSISO phase with the smallest diurnal range (blue), and the full JJAS composite (dotted black). Spatial averaging is done over ocean points inside box D (a), and land points inside boxes E (b) and F (c). (d-f) The corresponding daily mean precipitation (blue) and diurnal range (red) in mm/day of each phase’s spatially averaged composite diurnal cycle, by QBWO phase (solid) and BSISO phase (dotted). As in Fig. 2.13 but for Mindanao. . . . . . . . . . . . . . . 36 x 2.15 Daily mean values of select variables from ERA5 composited by QBWO phase (solid, orange) and BSISO phase (dotted, blue), averaged over box L covering Luzon (left) and box M covering Mindanao (right). Total column water vapor (kg m−2) is shown on top (a, d), downwelling shortwave radiation (W m−2) at the surface in the middle row (b, e), and 850-hPa zonal wind (m s−2) at the bottom (c, f). Corresponding JJAS mean values for each variable on each island are shown as a horizontal dotted black line. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 2.16 Daily mean anomalies from JJAS average of select variables from ERA5 composited by QBWO phase (top) and BSISO phase (bottom) over the Philippines. For each variables, phase 3 of each index is shown at the left, and phase 7 at the right. The grouping of 4 panels at the left shows total column water vapor (kg m−2), the middle grouping shows downwelling shortwave radiation (W m−2) at the surface, and the right grouping shows 850-hPa zonal wind (m s−2). . . . . . . . . . . . . . . . . . . . . . . 39 3.1 (a) NOAA ETOPO2 Topography (in meters) over the northern Philippines, with boxes of spatial averaging and important geographic features noted. (b) Structure of the first EOF of ERA5 zonal wind averaged in JJAS 1979-2020 inside Box A of (a), in m/s, by pressure level (hPa). (c) Normalized power spectrum of the principal component (PC) time series corresponding to the EOF in (b) in blue, with a theoretical red noise spectrum based on a time series with the same autocorrelation as the PC time series shown in dotted red. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 3.2 (a) Composite ERA5 zonal wind profiles (m/s) for all JJAS 1998-2020 days that fall in a certain bin of the PC time series of the EOF in Fig. 3.1b. Values are averaged inside Box A (Fig. 3.1a). Bins include 0.25σ on either side of the value noted. That is, +0.5σ days include any day between 0.25 and 0.75σ. The minimum and maximum bins are unbounded. (b) Anomalous moisture profiles in g/kg for the same bins noted in (a). (c) The number of JJAS 1998-2020 days that fall into each bin is shown as bars, with this number noted on top. The second number after the slash indicates the number of days in which a tropical cyclone center was near Luzon (inside 10-22N, 115-127E). Color coding is based on zonal wind, with easterly low-level wind bins shown in blues, and westerly bins shown in reds. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 3.3 Daily mean precipitation anomaly (mm/day) from CMORPH (JJAS, 1998-2020) aver- aged by bins of zonal wind EOF index. Anomalies are from the average precipitation rate on all JJAS days. Increasing zonal wind rotates clockwise around the figure. The +/- 2σ bins are not shown due to heavy tropical cyclone influence. . . . . . . . . . . . 56 3.4 Anomaly in the diurnal cycle amplitude (defined by the first harmonic of the composite diurnal cycle) composited by bin of the zonal wind EOF index for JJAS 1998-2020. Anomalies are from the amplitude of the full JJAS composite diurnal cycle. Increasing zonal wind rotates clockwise around the figure. The +/- 2σ bins are not shown due to heavy tropical cyclone influence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 xi 3.5 Hovmöller diagrams of CMORPH composite precipitation rate (mm/day) on days binned by zonal wind EOF index averaged across latitude inside Box L (Fig. 3.1a). Time starts at 08:00 local time in each panel, and increases downward. Dotted lines are estimates for a line of best fit between 16:00 and 01:00 of the longitude with the maximum precipitation rate at each time. This calculation only includes longitudes on the side of the island (east or west) that contains the maximum precipitation rate at 16:00. The estimated speed of propagation following this line of best fit is noted in the legend for each panel. This is not shown for the +1.5σ bin since little offshore prop- agation can be discerned. Increasing zonal wind rotates clockwise around the figure, with the composite 850-hPa zonal wind shown as a vector in each panel. The +/- 2σ bins are not shown due to heavy tropical cyclone influence. . . . . . . . . . . . . . . . 59 3.6 Precipitation rate (mm/hr) for the full 14-day simulation of select Cloud Model 1 (CM1) experiments. The x-axis in each is in km, with the coastlines marked as vertical dotted black lines. 05:00 on each day is noted as a horizontal dotted black line. The base-state 850-hPa zonal wind is shown as a labelled vector in each panel. . . . . . . . 62 3.7 Daily composite of the 14-day CM1 simulations showing precipitation rate by lon- gitude at 1-km resolution. Each simulation varies the base state with the wind and moisture profile bins shown in Fig. 3.2 (as well as surface variables and thermal pro- files, which are not shown). Dotted gray lines note the coastlines, and the dotted black line follows a line of best fit connecting the longitude of maximum precipitation rate at each time between 20:00 and 08:00. This is calculated based on precipitation rate smoothed to 8-km resolution. Increasing zonal wind in the base state rotates clockwise around the figure, with the base-state 850-hPa zonal wind shown as a labelled vector in each panel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 3.8 Lowest model level zonal wind (in m/s) Hovmöller diagrams for select experiments. The total zonal wind is shown on the top, and the perturbation from the base-state is shown on the bottom. The line of best fit for the maximum precipitation rate shown in Fig. 3.7 for the corresponding experiment is shown as a dotted black line. Coastlines are denoted with dotted gray lines. The base-state 850-hPa zonal wind is shown as a labelled vector in each panel in the top row. . . . . . . . . . . . . . . . . . . . . . . . 66 3.9 (a) Diagnosed 2-m temperature (C) for each experiment averaged for each time across the western half of the island. (b) As in (a), except averaged over the eastern half of the island. (c) Onshore (i.e. westerly positive) perturbation zonal wind (m/s) at the lowest model level (25m) for each experiment averaged for each time between the western coast and 25-km offshore. (d) As in (c), except with easterly winds defined as positive, averaged between the eastern coast and 25-km offshore. . . . . . . . . . . . . . . . . . 67 3.10 14-Day composite of perturbation potential temperature in colors, and total cloud mix- ing ratio (liquid+ice) contoured starting at 0.005 g/kg, then 0.02 g/kg and every 0.02 g/kg thereafter to 0.1 g/kg. Each variable is averaged over (a) the western half of the island for the -1.0σ experiment, (b) the eastern half of the island for the +0.0σ experi- ment, and (c) the eastern half of the island for the -1.0σ experiment. Local time starts at 08:00 and increases to the right, with height in kilometers on the y-axis. . . . . . . . 69 xii 3.11 Perturbation potential temperature (K) for the second day of the 0.0σ experiment in color at (a) 2.0-km, (b) 4.0-km, (c) 6.4-km, and (d) 9.1-km. The black contour indi- cates 2 mm/hr precipitation rate on Day 2 and is the same in all panels. The base-state 850-hPa zonal wind is shown as a labelled vector in (a). . . . . . . . . . . . . . . . . 70 3.12 14-Day composites from the -1.0σ experiment (left), the 0.0σ experiment (middle), and the +1.0σ experiment(right) showing Hovmöller diagrams of convective available potential energy (CAPE, J/kg; top), convective inhibition (CIN, J/kg; middle), and the level of free convection (LFC, m; bottom), with local time starting at 08:00 increasing downwards. The line of best fit for the maximum precipitation rate shown in Fig. 3.7 for the corresponding experiment is shown as a dotted white line. The base-state 850-hPa zonal wind is shown as a labelled vector in each panel in the top row. . . . . . 72 3.13 14-day composite CAPE (a; J/kg), CIN (b; J/kg), and LFC (c; m) averaged across the nearest 100-km of coastal waters on the eastern side of the simulated island in red, with composite precipiation rate (mm/hr) averaged in the same region in blue. . . . . 73 3.14 (a) Idealized base-state zonal wind profiles in the lower troposphere for sensitivity experiments set 1, based on linear interpolation between the -0.5σ and +0.0σ experi- ments shown in Fig. 3.2a. (b) As in (a) but for sensitivity experiments set 2, which are taken from set 1, but forced to a line (in pressure-wind coordinates) connecting a wind of -1 m/s at 1000-hPa and 0 m/s at 900-hPa if the set 1 profile is more westerly than the ideal line profile at a given height. (c) As in (a) but for sensitivity experiments set 3, which are interpolated between the -1.0σ and 0.0σ experiments, with the shear profile between 800-hPa and 850-hPa extended to the surface. Profiles are color coded by the propagation velocity of the smoothed (to 5-km spacing) maximum precipita- tion rate between 20:00 and 08:00 in each experiment, with red indicating eastward propagation, and blue indicating westward propagation. The gray profiles are chosen subjectively as experiments with weak or inconsistent offshore propagation in which the objective algorithm to calculate propagation speed failed. . . . . . . . . . . . . . . 75 3.15 (a) Zonal wind profiles averaged from 17:00-20:00 within 25-km of the average loca- tion of maximum precipitation between 17:00 and 20:00 in the composite, for sensi- tivity experiments set 1. (b) As in (a) but for sensitivity experiments set 2. (c) As in (a) but for sensitivity experiments set 3. Profiles are color coded by the propagation veloc- ity of the smoothed (to 5-km spacing) maximum precipitation rate between 20:00 and 08:00 in each experiment, with red indicating eastward propagation, and blue indicat- ing westward propagation. The gray profiles are chosen subjectively as experiments with weak or inconsistent offshore propagation in which the objective algorithm to calculate propagation speed failed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 3.16 The average longitudinal position of the maximum precipitation rate (after smoothing to 5-km spacing) between 17:00 and 20:00 is shown on the x-axis (in kilometers), with the base-state zonal wind (in m/s) at the 0.68-km above the surface on the y-axis. Dotted black lines indicate the center of the island (vertical) and 0 m/s (horizontal). The dots are color-coded by experiment set, with Set 1 in red, Set 2 in black, and Set 3 in blue. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 4.1 (a) NOAA ETOPO2 Topography (in meters) over the northern Philippines, overlaid with boxes of spatial averaging used in this study. . . . . . . . . . . . . . . . . . . . . 86 xiii 4.2 (a) Composite ERA5 zonal wind profiles (m/s) for each BSISO Phase in JJAS 1998- 2020. Values are averaged inside Boxes A-F together (Fig. 4.1a). (b) Anomalous moisture profiles in g/kg by BSISO Phase. (c) The number of JJAS 1998-2020 days that fall into each bin is shown as bars, with this number noted on top. . . . . . . . . . 88 4.3 Hovmöller diagrams of CMORPH composite precipitation rate (mm/day) on JJAS days by Lee et al. (2013) BSISO index. Results are averaged across latitude inside Boxes A-F (Fig. 4.1a), but additional longitude to the west of Box A and the east of Box B is also included. Time starts at 08:00 local time in each panel, and increases downward. Vertical dotted lines show the position of each coastline. The average to- pography from NOAA ETOPO2 inside boxes A-F is shown at the bottom for reference. 93 4.4 The composite diurnal cycle of CMORPH precipitation (mm/day) averaged inside each box on all JJAS 1998-2020 days in Fig. 4.1 (blue line). The composite diur- nal cycle only from days on which the amplitude of the diurnal cycle (as defined by the difference between daily maximum precipitation and daily minimum precipitation rate) is greater than its 85th percentile (red line). A histogram showing the number of days on which the maximum precipitation rate occurs in each half-hour bin (gray). The first (23:30-00:00 UTC or 07:30-08:00 Local Time) and last (00:00-00:30 UTC or 08:00-08:30 Local Time) bins are excluded since those bins capture days on which precipitation is either increasing or decreasing through the whole day due to longer time-scale features. The top row shows Boxes C-E, which include mainly land points, while the bottom row shows Boxes A, B, and F, which include mainly oceanic points. . 94 4.5 The relative change in the probability of an 85th percentile diurnal cycle amplitude day occurring in each box from Fig. 4.1 given a specific BSISO phase (y-axis). For example, if 30% of BSISO Phase X days showed a 85th percentile diurnal cycle am- plitude in Box Y, it would display as a 100% increase, since 15% of all days can be expected to exceed the 85th percentile. . . . . . . . . . . . . . . . . . . . . . . . . . . 96 4.6 In gray, histograms of the daily spatial average value inside Boxes A-F (Fig. 4.1) of ERA5 06:00-12:00 Local Time surface shortwave radiation (W/m2; top row), total column water vapor (kg/m2; middle row), and 850-hPa Zonal Wind (m/s; bottom row), for . The histogram in color is the histogram of the same variables, but only for days in which the diurnal cycle amplitude exceeded its 85th percentile over the coastal South China Sea (Box B in Fig. 4.1; left column), and the east coast of Luzon (Box E in Fig. 4.1; right column). Binned ERA5 variables still come from Boxes A-F together, but the diurnal cycle inside each sub-box determines the days from which to derive the distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 xiv 4.7 In contours, the 2-D histogram by bins of ERA5 850-hPa Zonal Wind averaged in Boxes A-F together (Fig. 4.1 on the x-axis (m/s), and ERA5 total column water vapor (kg/m2) averaged in Boxes A-F on the y-axis, and is the same in each panel. There are 20 total bins in both the zonal wind and total column water vapor distributed evenly between their respective maximum and minimum values. Contours are located every 12 days, with the innermost contour indicating 84 days. 2806 total days are used from JJAS 1998-2020 in this analysis. Colors indicate the anomalies in the histogram (of values for Boxes A-F together) given that an 85th percentile diurnal cycle amplitude day occurred in one of the 6 boxes A-F, and varies by panel. Values are scaled before calculating anomalies such that they represent the expected number of days in each 2- D bin if 2806 85th percentile days were to occur (i.e. the histogram is divided by 0.15, and then the anomaly is calculated). The top row’s histograms comes from average precipitation rate Boxes C-E, which include mainly land points, while the bottom row comes from Boxes A, B, and F, which include mainly oceanic points. The vertical dotted black line indicates the JJAS mean 850-hPa Zonal Wind, while the horizontal dotted line indicates the JJAS mean total column water vapor. . . . . . . . . . . . . . . 99 4.8 As in Fig. 4.7 for the contours and vertical lines. Colors indicate the anomalies in the histogram given that the BSISO was active in a certain phase. Values are scaled before calculated anomalies such that they represent the expected number of days in each 2-D bin if 2806 days from each BSISO phase were to occur. For example, to display the Phase 1 anomaly, the 2-D histogram is multiplied by 2806/188 (188 is the number of BSISO Phase 1 active days shown in Fig. 4.2c in JJAS 1998-2020), and then the anomaly is calculated. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 4.9 As in Fig. 4.3, but with the composite restricted to BSISO active days in a certain phase on which the first EOF of the Luzon vertical profile of zonal wind described in Chapter 3 was less than or equal to −0.25σ, indicating anomalous easterly winds. The number in parentheses next to the title indicates the number of days in each composite. 102 4.10 As in Fig. 4.3, but with the composite restricted to BSISO active days in a certain phase on which the first EOF of the Luzon vertical profile of zonal wind described in Chapter 3 was greater than or equal to +0.25σ, indicating anomalous westerly winds. The number in parentheses next to the title indicates the number of days in each composite. 103 4.11 Meridionally averaged precipitation rate (mm/hr) for the full 7-day simulation the Cloud Model 1 (CM1) control simulations with a BSISO Phase 3 base-state (a), and a BSISO Phase 7 base-state (b). The x-axis in each is in km, with the coastlines marked as vertical dotted black lines. 05:00 on each day is noted as a horizontal dotted black line. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 4.12 Daily composite of the 7-day control CM1 simulations showing meridionally averaged precipitation rate (mm/hr) by longitude at 1-km resolution. Each simulation varies the base state with ERA5 BSISO composite profiles. The wind and moisture profiles are shown in Fig. 4.2a-b. Dotted gray lines note the coastlines, and the dotted black line follows a line of best fit connecting the longitude of maximum precipitation rate at each time between 20:00 and 08:00. This is calculated based on precipitation rate smoothed to 5-km resolution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 xv 4.13 As in Figure 4.12, but for the wind-only CM1 simulations. The base-state for these simulations uses the wind profile from each BSISO phase composite (shown in Fig. 4.2a), and the JJAS mean profile for the moisture and temperature. . . . . . . . . . . . 110 4.14 (a) CM1 precipitation rate (mm/day) averaged over land for each of simulations with varying BSISO composite derived base-states. Bars for the Control simulations (Fig. 4.12) are solid colors, while those for the Wind-Only simulations (Fig. 4.13) are hatched. (b) As in (a), but averaged over the ocean grid points on the leeward side of the island (west for Phases 2-5, east for others). . . . . . . . . . . . . . . . . . . . . 111 4.15 Composite precipitation rate (mm/hr) averaged over land for the moisture experiments (left), and solar experiments (right). Simulations in the top row use BSISO Phase 3 profile in the base-state, while the bottom row forces BSISO Phase 7 profile in the base-state. The moisture experiments use the BSISO profile only for the winds, the JJAS mean temperature profile, and prescribe the moisture profile. The solar experi- ments use the control BSISO profile in the base-state, but vary the solar constant. . . . 112 4.16 As in Fig. 4.15, but averaged over ocean on the leeward side of the island (west for Phase 3, east for Phase 7). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 4.17 Daily composite of meridionally averaged lowest model level zonal wind (m/s) in color, with the 2 mm/hr contour of composite precipitation rate superimposed. BSISO Phase 3 base-state is used in each simulation in the top row, while BSISO phase 7 base-state is used in the bottom row. The size of the island positioned in the center of the domain is 25-km (a, f), 100-km (b, g), 200-km (c, h), 400-km (d, i), and 800-km (e, j). The 200-km simulation is identical to the control simulation in Fig. 4.11a. The domain size is 800-km for (a-c, f-g), but increases to 1600-km for (d-e, i-j). . . . . . . 117 4.18 Meridionally averaged precipitation rate (mm/hr) for the full 7-day Island Size exper- iments. BSISO Phase 3 base-state is used in each simulation. The size of the island positioned in the center of the domain is 25-km (a), 100-km (b), 200-km (c), 400-km (d), and 800-km). The 200-km simulation is identical to the control simulation in Fig. 4.11a. The domain size is 800-km for (a-c), but increases to 1600-km for (d-e). . . . . 118 4.19 Meridionally averaged precipitation rate (mm/hr) for the first 2 days of the No-Diurnal- Cycle experiments. Simulations from each of the 8 BSISO base-states are shown. The diurnal cycle is removed by fixing the solar constant at the daily mean top-of- atmosphere incident shortwave on 1 August at 17N and fixing the solar zenith angle at 0◦. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 4.20 Meridionally averaged precipitation rate (mm/hr) for the full 7-day Island Size exper- iments with no insolation diurnal cycle. BSISO Phase 2 base-state is used in each simulation. The size of the island positioned in the center of the domain is 25-km (a), 100-km (b), 200-km (c), 400-km (d), and 800-km). The domain size is 800-km for (a-c), but increases to 1600-km for (d-e). The diurnal cycle is removed by fixing the solar constant at the daily mean top-of-atmosphere incident shortwave on 1 August at 17N and fixing the solar zenith angle at 0◦. . . . . . . . . . . . . . . . . . . . . . . . . 122 4.21 2-m Temperature (C) averaged over land points by time for the full 7-day Island Size No Diurnal Cycle Experiments. Phase 2 base-state is used for (a), while Phase 6 base- state is used for (b). Shades of orange have a larger island and domain size compared to the control. Shades of purple have a smaller island but the same domain size. . . . . 123 xvi Chapter 1 Introduction The Maritime Continent (MC) and its surrounding waters are prone to numerous types of at- mospheric phenomena that make it highly vulnerable to climate risks Yusef and Francisco (2009). Variability in the diurnal cycle can be a critical factor in determining total precipitation on the is- lands and in coastal waters (Biasutti et al., 2012; Bergemann et al., 2015; Zhu et al., 2017). The warm sea surface temperatures (SSTs), numerous islands of varying size, and complex topography make understanding the abundant precipitation in this region a challenging problem for numeri- cal models with global ramifications (Ramage, 1968; Neale and Slingo, 2003). The diurnal cycle is also critical for the development of extreme rainfall and the high mean-state rainfall found in coastal oceans (Ruppert and Chen, 2020). While the diurnal cycle has been extensively studied, uncertainty remains regarding its variability and response to large-scale controls. The canonical diurnal cycle behavior over MC islands develops from convergence associated with the sea-breeze or mountain-breeze in the late morning, typically contributing maximum pre- cipitation rates in the late afternoon and evening hours (Dai, 2001; Kikuchi and Wang, 2008). Fre- quently, convection will then propagate offshore during the overnight hours, leading to an overnight or morning maximum in precipitation rates over coastal oceanic regions (Yang and Slingo, 2001; Mori et al., 2004; Sakurai et al., 2005; Natoli and Maloney, 2019). Offshore propagation has been attributed to convergence associated with the land-breeze (e.g. Houze et al. 1981; Ho et al. 2008; Fujita et al. 2011), advection by the mean wind (e.g. Ichikawa and Yasunari 2006, 2008; Yanase et al. 2017), and destabilization of the offshore environment by low-level ascent initiated by gravity waves (e.g. Mapes et al. 2003; Love et al. 2011; Hassim et al. 2016; Yokoi et al. 2017). Diurnal cycle behavior and the tendency for offshore propagation varies widely from one day to the next, motivating continued research. The MC region is also influenced by numerous large-scale modes of variability from features on global, inter-annual time scales like the El Niño Southern Oscillation 1 (ENSO; Rauniyar and Walsh 2013) to equatorial waves on synoptic scales (Ferrett et al., 2020). Any of these can significantly affect the diurnal cycle and local precipitation (Sakaeda et al., 2020). The monsoon is an important driver of precipitation variability that generally follows the sea- sonal cycle. For example, the Philippine archipelago experiences southwesterly monsoon winds during boreal summer (June-September, JJAS), which brings moisture-laden air away from the equator and delivers much of the region’s annual precipitation (Moron et al., 2009; Matsumoto et al., 2020). Drier trade easterlies dominate the rest of the year, which is generally much drier (Lee et al., 2021). While the agricultural sector relies on monsoon moisture, it can also bring dev- astating flooding (Cruz et al., 2013). The southwest monsoon is not always consistent throughout the season. rather, it is subject to numerous active and break cycles (Annamalai and Slingo, 2001; Olaguera et al., 2020). These alternating periods of relatively enhanced activity and quiescence are modulated by several other modes of intraseasonal variability in the tropics. 1.1 Large-Scale Modes of Tropical Convective Variability The largest source of convective variability on intraseasonal timescales in the tropics is con- tributed by the Madden-Julian Oscillation (MJO; Madden and Julian 1971, 1972). The MJO is an eastward-propagating area of enhanced convection in the tropical warm pool with a time-scale of 30-90 days. The active phase is characterized by strong westerly winds and abundant free- tropospheric moisture, while the suppressed phase exhibits easterly winds, a dry free-troposphere, and sunnier skies (Madden and Julian, 1994; Maloney and Hartmann, 1998; Riley et al., 2011). During boreal summer (June-September, JJAS), convection on this timescale tends to propagate northward into the Asian and West Pacific summer monsoon regions, and influence the onset of the monsoon in addition to producing active and break periods in the heart of the season (Wang and Xu, 1997; Annamalai and Slingo, 2001). This mode is often referred to as the boreal sum- mer intraseasonal oscillation (BSISO). The MJO and BSISO can generally be considered to be the same phenomenon that exists in a different seasonal background state (e.g. Jiang et al. 2018; Wang and Sobel 2022). 2 As the BSISO (or MJO) can be manifest as active and break periods in the southwesterly mon- soon over the SCS (Chen and Chen, 1995; Bagtasa, 2020), other modes of variability that similarly modulate the monsoonal flow over the Philippines may also impact the diurnal cycle. Differences in how another mode impacts the monsoon background may be insightful in ascertaining the pri- mary controls on the diurnal cycle itself. Variance in boreal summer outgoing longwave radiation (OLR) shows the global maximum of the quasi-biweekly (10-20 day) time scale occurring in the South China and Philippine Seas (Qian et al., 2019). The importance of this mode in determining monsoon activity has been a subject of research for decades, first identified in the Indian monsoon region (Krishnamurti and Bhalme, 1976; Krishnamurti and Ardanuy, 1980; Chen and Chen, 1993), before being later explored in the west Pacific and east Asian monsoon regions (Chen and Chen, 1995; Chen et al., 2000). This mode has often been described as the quasi-biweekly oscillation (QBWO), consisting of a northwestward propagating region of anomalous moisture, convection, westerly winds, and cyclonic vorticity (Kikuchi and Wang, 2009; Tao et al., 2009; Li et al., 2020). Disturbances tend to emerge in the equatorial western Pacific and propagate through the Philippine Sea, South China Sea, and into Asia, frequently impacting the Philippines (Chen and Sui, 2010; Yan et al., 2019). Many of these studies refer to the QBWO in a statistical rather than physical sense, but there is evidence that multiple phenomena can contribute to quasi-biweekly variability, and thus project onto various QBWO indices. The westward propagating mode ubiquitous in the west Pacific is often traced to a moist, R1 wave (Matsuno, 1966) that is modified by the monsoon background state (Chatterjee and Goswami, 2004). In addition to its modulation of the monsoon onset and persistence (Qian et al., 2019), the QBWO has noteworthy impacts on tropical cyclones (Zhou et al., 2018; Han et al., 2020), extreme rainfall (Liu et al., 2014), and heatwaves in China (Chen et al., 2016; Gao et al., 2018). 3 1.2 Diurnal Cycle Variability The MJO impact on the diurnal cycle has been one of the more widely studied relationships, in part because of the potential for the diurnal cycle to feed back onto MJO propagation across the MC (Oh et al., 2013; Peatman et al., 2014; Hagos et al., 2016). However, a consensus remains out of reach, and the mechanisms involved in explaining this potential relationship are poorly understood. While oceanic precipitation generally follows the enhanced moisture of the MJO active phase, several studies have shown a relative minimum in the amplitude of the diurnal cycle and in total precipitation over land masses during the active phase of the MJO (Sui and Lau, 1992; Rauniyar and Walsh, 2011; Oh et al., 2012). Such a signal has also been observed for regions impacted by the BSISO (e.g. Chen and Takahashi 1995; Ho et al. 2008; Xu and Rutledge 2018), although a weaker diurnal cycle is still present over land during the active phase (Chudler et al., 2020). Taking a more precise view, Peatman et al. (2014) demonstrated a peak in the amplitude of the diurnal cycle in the transition from suppressed to active MJO state for several MC islands using satellite observations. Vincent and Lane (2017) identified a double-peak in the diurnal cycle amplitude as a function of MJO phase in a WRF simulation, with a secondary peak at the end of the MJO active state, but noted this was less significant in observations. An understanding of the mechanism regulating this diurnal cycle behavior has not yet been convincingly established. Many of the above studies have attributed the enhanced diurnal cycle during the suppressed phase to the reduced cloudiness, which leads to a stronger thermal differen- tial between the land and sea during daytime, and thus a stronger sea-breeze and stronger diurnal precipitation. This, however, would not explain the specific preference for a diurnal cycle peak near the end of the MJO suppressed period. Peatman et al. (2014) speculated that frictional mois- ture convergence associated with the Kelvin wave east of enhanced MJO convection (Gill, 1980) can explain this difference. Equatorial wave dynamics fall short of explaining why the strongest diurnal cycle occurs during the transition to BSISO active conditions in the northern Philippines, much further from the equator (Natoli and Maloney, 2019). Moreover, budget analyses by Lu et al. (2019) and Chen et al. (2019) found moisture convergence to be an important factor, but attributed 4 it to convergence of MJO-scale moisture by the local land-sea breeze circulation rather than equa- torial wave dynamics. Free tropospheric moisture availability has been shown to be an important control on tropical precipitation and its diurnal cycle (Bretherton et al., 2004; Vincent and Lane, 2017). The wind profile, especially in the lower troposphere, has also been identified as an im- portant player. Specifically, the strong winds in the MJO active phase reduce the land-sea contrast and thus daytime convection over land (Shige et al., 2017; Wang and Sobel, 2017; Wu et al., 2017, 2018; Yokoi et al., 2019). Conversely, light winds tend to favor a substantial diurnal cycle on the leeward side of islands (Virts et al., 2013; Short et al., 2019; Qian, 2020). Since MJO moisture leads the strong westerly winds in phase (Maloney and Hartmann, 1998), the arrival of increased moisture before the strong winds during the transition to active could present a compelling hypoth- esis explaining the preference for strong, offshore propagating diurnal cycles during these MJO phases. 1.3 Scope and Significance of This Work This dissertation aims to address several of the gaps in our knowledge regarding the tropical diurnal cycle and its variability that have not been adequately covered by the body of literature outlined above. The main goal is to understand the mechanisms regulating diurnal cycle behavior as influenced by large-scale modes of tropical convective variability. In particular, each chapter will address a part of the hypothesis stated at the end of the previous section and in Natoli and Maloney (2019). More explicitly, it is proposed that the intraseasonal variability in diurnal cycle behavior can be understood in terms of the free tropospheric moisture and background wind profiles that are favored by a certain phase of a large-scale mode of tropical convective variability, such as the MJO. In each chapter, a different perspective will be taken in the aim of examining this problem. While there will be a heavy focus on the boreal summer MJO (i.e. the BSISO) and the Philippine archipelago as a case study, it is possible that the results and conclusions can be generalized to other tropical islands in the MC. 5 This work is intended to provide the most clarity to date on the regulators of convective precipi- tation on the diurnal timescale. The varied perspectives presented in each chapter will all reinforce certain aspects of the stated hypothesis, provided a holistic view on the intraseasonal variability in tropical island diurnal cycle. An improved understanding of this relationship is important for sev- eral reasons. Most directly, due to the skill in predicting large-scale modes of convective variability such as the MJO on subseasonal timescales, a solid grasp on how this influences the local diurnal cycle on MC islands could lead to improved lead time for hydrometeorologically significant events such as flooding and drought. Additionally, while there is subseasonal prediction skill for the MJO, there is lesser skill in predicting whether the MJO convection will successfully transit the MC and progress into the west Pacific (Kerns and Chen, 2016; Kim et al., 2016). The diurnal cycle over MC islands has been identified as a potential reason for this heightened uncertainty (Hagos et al., 2016; Zhang and Ling, 2017), which suggests that improved comprehension of diurnal cycle variability could lead to improved prediction skill regarding MJO transit across the MC. In the next chapter, the QBWO will be explored in detail to address how other modes of trop- ical convective variability impact the local diurnal cycle. Prior to this work, this relationship has not been studied in prior literature. The results will show that the diurnal cycle in the Philippines behaves very similarly through a QBWO event compared to a BSISO event, and that the same mechanisms proposed by Natoli and Maloney (2019) to explain the BSISO-diurnal cycle relation- ship can explain the QBWO-diurnal cycle relationship. This chapter has been published, with minor changes, in Monthly Weather Review. In Chapter 3, we take a step back from a specific large-scale mode of tropical convective variability, and explore the impact of monsoon winds on any timescale. The impact of the background wind profile on the diurnal cycle will be isolated through analysis of observations and an idealized model to show how several important aspects of diurnal cycle variability can be explained by the background wind on an individual day. This chap- ter, with slight modifications, has been submitted for publication in Journal of the Atmospheric Sciences. Our stated hypothesis will be tested most directly in Chapter 4. Through a probabilis- tic analysis of observational data and sensitivity tests using an idealized model, it will be shown 6 that the a background state of near to above average moisture and weak offshore winds leads to a strong diurnal cycle with offshore propagation into coastal waters. Certain phases of the BSISO make such conditions more likely, which is invoked to explain the observed BSISO-diurnal cycle relationship found in composites. The results of this chapter are in preparation for publication in Journal of Climate. Lastly, some overall conclusions, caveats, and an outline of potential avenues for future research will be discussed in Chapter 5. 7 Chapter 2 The Quasi-Biweekly Oscillation and the Philippines Diurnal Cycle1 2.1 Introduction While the relationship between the boreal winter MJO and the tropical island diurnal cycle has received considerable scrutiny in the recent literature, the boreal summer mode (e.g. the BSISO) and its impact on landmasses in the Asian/West Pacific summer monsoon region has had much less attention. However, research surrounding the 2018 field campaign titled Propagation of Intrasea- sonal Tropical Oscillations (PISTON) showed that diurnal cycle variability over the Philippine archipelago can behave similarly through a BSISO event (Natoli and Maloney, 2019) compared to that over large equatorial islands such as Sumatra, Borneo, and New Guinea through an MJO event (e.g. Peatman et al. 2014; Vincent and Lane 2016). For example, the diurnal cycle in the Philippines appears to reach a maximum during the BSISO suppressed state (Ho et al., 2008; Park et al., 2011; Xu and Rutledge, 2018; Xu et al., 2021), although the typical afternoon maximum is still present in the active state (Chudler et al., 2020). Natoli and Maloney (2019) noted a maximum in the diurnal amplitude over land and coastal waters of the South China Sea (SCS) during the transition from suppressed to active BSISO state when the mid-tropospheric moisture begins to increase, but prior to the arrival of strong westerly monsoon winds. The relationship between local precipitation and other modes of tropical variability has also been recently getting more attention. Ferrett et al. (2020) showed a significant modulation of local precipitation extremes by various types of equatorial waves. Additionally, Sakaeda et al. (2020) took a thorough look at the impact of various equatorial wave modes on the MC diurnal 1This chapter, as well as some text from Chapter 1, has been published in Monthly Weather Review with minor changes under the title “Quasi-Biweekly Extensions of the Monsoon Winds and the Philippines Diurnal Cycle” 8 cycle during boreal winter, noting important differences in behavior between various wave modes and individual islands. These results highlight the importance of a more local-scale approach, in particular, that the mode of variability should be considered based on its modulation of the local environmental background conditions, which then modulates the diurnal cycle. They also distinguished diurnal cycle behavior within an individual island related to the position relative to the wind (leeward vs. windward), and aspect of topography. Specifically, the diurnal cycle was found to be enhanced on the leeward side of MC islands for the MJO and n=1 equatorial Rossby (R1) waves, consistent with Virts et al. (2013) and Qian (2020). Since the monsoon system over the SCS is modulated by variability on the quasi-biweekly timescale (Chen and Chen, 1995), and significant variability on that timescale was observed during PISTON (Sobel et al., 2021), another interesting opportunity to explore diurnal cycle variability is presented. The impact of the QBWO on the diurnal cycle in the Philippines has not been explored, but recent work suggests that the same mechanisms could be at play as those important to the MJO, as well as equatorial waves. This chapter aims to explore this relationship in detail, and determine how well the ideas presented for the MJO/BSISO-diurnal cycle interaction apply to a different mode of tropical convective variability that has received less attention. Specifically, if another large-scale feature impacts the environmental background conditions (e.g. lower tropospheric wind and mid-tropospheric moisture) in a similar way to the BSISO, will the diurnal cycle respond similarly? The first goal is to describe west Pacific variability on the 10-20 day timescale and its impor- tance to the Philippine archipelago. This includes examination of prominent variability on this timescale that occurred during a recent major field program (Sobel et al., 2021). Second, an index for the QBWO will be described that can be used to composite precipitation and other variables. Third, we aim to establish the impact of the quasi-biweekly mode on the diurnal cycle of the Philip- pines and its offshore propagation. The final goal is to compare and contrast the QBWO-diurnal cycle relationship with the MJO-diurnal cycle relationship over the Philippines to help reveal im- portant controls on diurnal convection and the mechanisms involved. 9 The next section will describe the data and methods used, followed by a description of the QBWO index used in this study. In Section 3, results will be discussed, starting with a case study during the 2018 PISTON field campaign, then leading into a composite analysis for the period 1998-2020 from the large scale to the island scale. Section 4 includes a discussion of the mechanisms and a comparison with the BSISO, with a summary and major conclusions for this chapter outlined in section 5. 2.2 Data and Methods 2.2.1 Data Description This chapter employs several datasets to analyze quasi-biweekly variability in the monsoon and the diurnal cycle of precipitation. First, precipitation data comes from version one of the Climate Prediction Center (CPC) Morphing Technique (CMORPH; Joyce et al. 2004; Xie et al. 2017). The data is available as 30-minute total precipitation accumulation estimates at 8-km spatial resolution, covering 60◦S-60◦N. The CMORPH method takes precipitation rate estimates from passive microwave satellite retrievals and then uses cloud-motion vectors derived from infrared satellites to morph and interpolate through space and time to other passive microwave estimates. Thus, infrared information is only used to predict storm motion, and is not directly used to estimate precipitation rates. These initial estimates are bias-corrected against gauge data and the Global Precipitation Climatology Project (Adler et al., 2003) to yield the final product. Other studies have shown that this bias-corrected CMORPH technique removes most of the bias over land in warm climates (as in this study), and performs favorably when compared with the commonly used TRMM 3B42 precipitation dataset (Xie et al., 2017). CMORPH also demonstrates similar skill compared against the IMERG product (Huffman et al., 2015; Sahlu et al., 2016). The same analysis described below was performed for IMERG during the available period of 2000-2020 and the results remain robust. Complementing the precipitation data, the 5th Generation Reanalysis from the European Cen- tre for Medium-Range Weather Forecasting (ERA5; Hersbach et al. 2020; Copernicus Climate 10 Change Service (C3S) 2017) is used for JJAS, 1998-2020. Variables analyzed here include total column water vapor, surface downwelling shortwave radiation, and 850-hPa wind. Each of these fields are considered at 1-hour temporal resolution and 0.25◦ spatial resolution. In this study, the purpose of the ERA5 data is to contextualize the precipitation results and elucidate potential mech- anisms. Additional variables were examined on numerous pressure levels through the troposphere, but these did not add further insight and are not included in this discussion. In addition, interpolated outgoing longwave radiation (OLR) data from the Advanced Very High Resolution Radiometer (AVHRR) is analyzed at daily temporal and 2.5◦ spatial resolution for JJAS, 1979-2020 (Liebmann and Smith, 1996). OLR is used to calculate the QBWO index used in this study, as well as track large-scale convection associated with it. Zonal wind data from balloon soundings in the 2018 PISTON field campaign are also used at 3-hourly resolution from the R/V Thomas G. Thompson and 12-hourly resolution from Yap Island (Sobel et al., 2021). Processing and quality control for sounding data follows Ciesielski et al. (2014). These locations relative to the Philippines are shown in Figure 2.1. Lastly, topographic data from the National Oceanic and Atmospheric Administration’s (NOAA) ETOPO2 dataset are incorporated to provide geographic context for the results (National Geophysical Data Center, 2006). The BSISO index used in this dissertation for comparison to the QBWO results is that by Lee et al. (2013), which we used in Natoli and Maloney (2019). The QBWO index used will be described below. 2.2.2 Methods The compositing method in this dissertation follows that of Natoli and Maloney (2019), in which a single composite diurnal cycle is created for CMORPH precipitation for all days in JJAS in the analysis period, defined here as the boreal summer composite diurnal cycle. In addition, separate composite diurnal cycles are created by averaging measurements from only days in that period in which an index of intraseasonal variability (e.g. QBWO or BSISO) was considered active and in a certain phase (one of eight). An anomaly in this study refers to the difference between the composite of interest and the JJAS mean. Statistical significance of the precipitation results also 11 CB A F E D L M Palau Yap Luzon Mindanao 118°E 120°E 122°E 124°E 126°E 128°E 130°E 132°E 134°E 136°E 138°E 6°N 8°N 10°N 12°N 14°N 16°N 18°N PISTON Ship Track and Philippine Geography Aug Sep Oct 200 400 600 800 1000 1200 1400 1600 1800 2000 meters Figure 2.1: NOAA ETOPO2 Topography (in meters) over the Philippines, with boxes of spatial averaging and important geographic features noted. The track of the R/V Thomas G. Thompson during the August- October 2018 PISTON field campaign is also shown in purple, with August in the darkest color and October in the lightest. 12 follows Natoli and Maloney (2019) using a bootstrapping method, where the composite diurnal cycle in an ISO phase was compared against 1000 composite diurnal cycles taken from random days in the study period, with a Poisson distribution used to account for the fact that ISO active days tend to come in non-independent groups of several days. More details can be found in Natoli and Maloney (2019). This study also calculates power spectra for a few different time series. This is done by cal- culating the spectrum for each season individually (e.g. JJAS 1998, 1999, etc.) after applying a Hanning window to reduce the Gibbs phenomenon. Then, spectra are averaged from all years to increase degrees of freedom, only considering the relevant season (boreal summer). The theoret- ical red noise spectra follow equation 5 of Gilman et al. (1963), which provides an estimate for how a power spectrum of a pure red noise process with the same autocorrelation as the time series of interest would appear. An F-test is employed to determine if the calculated power spectrum is significantly different from its corresponding theoretical red noise spectrum. OLR data is also bandpass filtered to 10-20 days in this study to prepare the data for calculation of the QBWO in- dex, and highlight variability on relevant timescales for analysis of the 2018 PISTON period. This is done by applying a Lanczos filter with 93 weights to detrended OLR data at each grid point (Duchon, 1979). 2.2.3 QBWO Index An index was created to track the QBWO in the west Pacific and facilitate analysis of its rela- tionship to the Philippine diurnal cycle. Many prior studies have created indices for this features, but a consensus has yet to emerge on the best method (Kikuchi and Wang, 2009; Han et al., 2020; Yan et al., 2019; Qian et al., 2019). The timescale studied for the QBWO also differs in the lit- erature, but most include the 10-20 day period, with some extending to 25 or 30 days on the low frequency end, and others extending to 5 or 7 days on the high frequency end. Here, we attempt to exclude both timescales more characteristic of synoptic scale variability (5-10 days), as well as the longer time scales approaching the BSISO mode (20-30 days), and select a band of 10-20 13 days upon which to base our index. This timescale was found to display consistent westward prop- agating activity in the region of interest that also resembles the QBWO behavior documented in previous studies (Chatterjee and Goswami, 2004; Chen and Sui, 2010; Li et al., 2020). Addition- ally, the 10-20 day band well-captures the spectral peak in lower tropospheric wind variability near the Philippines. Figure 2.2a shows the power spectrum calculated from 850-mb ERA5 zonal wind averaged over northern Luzon (box L in Fig. 2.1) during JJAS 1998-2020. A statistically significant spectral peak is identified around 10-15 days. This peak is robust across averaging domains that vary in both size and shape surrounding the Philippine archipelago. Thus, the 10-20 day band encompasses the spectral peak in the region of interest, produces the structure outlined in previous studies, and excludes other time scales that may muddy results (Chen and Sui, 2010; Yan et al., 2019). Fig. 2.2b shows the same for Mindanao over box M, indicating a weaker but noticeable peak in the 10-15 day band that does not reach statistical significance. The architecture of our index is most similar to that of Qian et al. (2019), only differing in temporal and spatial domain, and filtering time scale that improve variance explained by the index. EEOFs are calculated from the 10-20 day OLR anomalies inside the domain of 0-35N, and 115- 165W for JJAS 1979-2020, with information included at lags 0, 2, and 4 days prior. The spatial patterns associated with the two leading modes of variability in 10-20 day OLR are shown in Fig. 2.3, which explain 16.67% and 16.31% of the variance respectively. They are well separated from the other EOFs (not shown) and represent a propagating wave-like signal based on a lag correlation analysis of their unfiltered principal components (described below in more detail) that maximizes at 3-4 days. The coherence squared between the two PCs averaged inside the 10-20 day band is 0.81. The patterns are presented in Figure 2.3 such that time progresses going downward, and the pattern at lag 4 of EEOF 2 is roughly equivalent to the lag 0 pattern of EEOF 1. Thus, the time progression continues through EEOF 1 first, and then through the lags of EEOF 2. The spatial patterns shown here were not highly sensitive to choice of domain, filtering timescale (as long as 10-20 day band was included), lag timescale, and months analyzed. Other periods in addition to 14 571015202530456090 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 N or m al iz ed P ow er Luzon (Box L) u850 Power Red Noise ( =0.83) 90% Confidence Bound 571015202530456090 Period (Days) 0.000 0.025 0.050 0.075 0.100 0.125 0.150 0.175 N or m al iz ed P ow er Mindanao (Box M) u850 Power Red Noise ( =0.86) 90% Confidence Bound Figure 2.2: (a) Power spectrum of ERA5 850-mb zonal wind averaged inside box L in Fig. 2.1 during June-September (JJAS) 1998-2020 (blue), with theoretical red noise spectrum (red; Gilman et al. 1963), and its 90% confidence bound calculated with an F-test (gray, dotted). (b) As in (a) but for Mindanao, averaged inside box M. 15 JJAS were considered, but precipitation patterns over the northern Philippines appear somewhat distinct in May or October (not shown), which motivated the choice for the shorter season. To calculate the principal component (PC) time series, the unfiltered OLR anomalies (with the seasonal cycle removed) are projected back onto the EEOF patterns in Fig. 2.3. Since unfiltered OLR anomalies make up the PCs, it must be assured that they still capture the 10-20 day timescale well, as we do allow for other time scales to project on the index. Spectra for both PC1 and PC2 (Fig. 2.3d, h) show strong, statistically significant peaks in spectral power on 10-20 day timescales. While there is some bleeding to both higher and lower frequencies, no distinct peak can be seen elsewhere in the spectrum, which provides confidence that this index is picking up westward propagating signals that oscillate on roughly 10-20 day time scales. The use of an EEOF index also allows for more direct comparison to MJO or BSISO studies that employ the commonly-used RMM index for the MJO (Wheeler and Hendon, 2004), or the Lee et al. (2013) index for the BSISO. We can split the phase space into 8 phases according to the sign and magnitude of the corresponding PC time series for each day. Since the choices of the sign of each PC and which PC to make the x-axis or y-axis in the phase space are arbitrary, we defined them in this study such that the “active” phases for the Philippines most closely correspond to the “active” phases of the Lee et al. (2013) index for the BSISO. In other words, phases 2-4 generally correspond to suppressed convection and low-level easterly winds over Luzon for both indices, while phases 6-8 generally indicate enhanced convection and strong westerlies. This allows for the direct comparison of the precipitation behavior and background conditions over the Philippines later in this manuscript. It is important to verify that our QBWO index is reasonably independent from the Lee et al. (2013) BSISO index before composites for each are directly compared in the subsequent sections. Fig. 2.4 shows the number of days in a certain Lee et al. (2013) BSISO phase classified by each QBWO phase. The vast majority (between 71 and 80%) of active QBWO days have an inactive BSISO, and there is no preference for a day to be classified as the same numbered phase in each index. This percentage is consistent with BSISO activity across the entire study period, as the index 16 120°E 130°E 140°E 150°E 160°E 10°N 20°N (c) Lag: 0 Days 10°N 20°N (b) Lag: -2 Days 10°N 20°N (a) Lag: -4 Days 8 6 4 2 0 2 4 6 8 W /m 2 120°E 130°E 140°E 150°E 160°E (g) Lag: 0 Days (f) Lag: -2 Days (e) Lag: -4 Days 571015203090 Period (Days) 0.001 0.000 0.001 0.002 0.003 0.004 0.005 No rm al ize d Po we r A no m al y (d) PC 1 Power-Red Noise 90% Conf. PC Power 90% Confidence 571015203090 Period (Days) (h) PC 2 Power-Red Noise 90% Conf. QBWO EEOF 1 QBWO EEOF 2 Figure 2.3: Spatial pattern at -4 (top), -2 (middle), and 0-day (bottom) lags from extended EOFs 1 (left) and 2 (right) of 10-20 day bandpass filtered AVHRR OLR anomalies in physical units (W m−2). The bottom row shows the difference between power spectra of each corresponding principal component time series and the corresponding 90% confidence bound of a theoretical red noise spectrum with the same autocorrelation as the PC time series. Values above zero (dotted red line) can be considered statistically significant at the 90% confidence level. 17 is inactive about 75% of the JJAS days between 1998 and 2020. Anti-correlation between QBWO and BSISO activity has also been found on interannual timescales (Yang et al., 2008). The third and fourth multivariate EOF identified by Lee et al. (2013), which are by definition independent from the first two EOFs which make up the BSISO index, have been shown to capture some QBWO variability (Qian et al., 2019). Thus, the QBWO index appears to be randomly selecting from BSISO activity, and we can assume that they are independent. 1 2 3 4 5 6 7 8 QBWO Phase 0 50 100 150 200 250 300 N um be r of D ay s BSISO Activity by QBWO Phase BSISO P1 BSISO P2 BSISO P3 BSISO P4 BSISO P5 BSISO P6 BSISO P7 BSISO P8 Inactive Figure 2.4: Number of days in a certain Lee et al. (2013) BSISO phase distributed by active QBWO phase. The BSISO is considered to be inactive when the amplitude of the index is less than one. The darker color in the stacked bar chart indicates days that are classified as the same phase number in both indices. 18 2.3 Results 2.3.1 2018 PISTON Case Study The operational period of the 2018 PISTON field campaign (14 August - 14 October 2018) is used as a case study to assess this index and 10-20 day variability for a specific time period, before leading into a more general composite analysis in the next subsection. This time period was selected because prominent 10-20 day variability was apparent in raw data during a major field campaign (Sobel et al., 2021). One of the original goals of the 2018-19 PISTON project was to sample lower frequency intraseasonal oscillations, like the BSISO. However, the 2018 leg of the experiment witnessed minimal BSISO activity during the two month long cruise, only sampling a suppressed phase of an MJO-like disturbance in early October. While exploration of the tropical QBWO was not an original goal, the noteworthy variability observed on this timescale described below presents an opportunity to learn more about this feature (Sobel et al., 2021). Figure 2.5 shows a time-height diagram of zonal wind observations from radiosondes released during PISTON. The top panel shows 12-hourly soundings released from the Yap island, while the bottom shows the 3-hourly soundings released aboard the R/V Thomas G. Thompson, with white space when the ship was in port or in transit (see Fig. 2.1 for locations). Both locations in the west Pacific observed significant variability in zonal wind on 10-20 day timescales. Roughly every two or three weeks, the region experienced surges of fairly strong westerly winds in the low levels, extending through much of the troposphere and lasting about 7-10 days. Westerly winds tapped into deep monsoonal flow bringing increased moisture and increased mesoscale convective system activity (Chudler and Rutledge, 2021; Sobel et al., 2021). Such monsoon surges were often caused by and/or enhanced by tropical cyclones (TCs) passing northeast of the study domain, similar to events described in Cayanan et al. (2011) and Bagtasa (2017). These were interspersed with tranquil periods of weak trade easterlies. The identification of enhanced QBWO activity during the 2018 boreal summer season is consistent with prior work suggesting a preference for such activity during El Niño years (the late summer of 2018 featured a strengthening El Niño event) and during periods of decreased BSISO activity (Yang et al., 2008; Yan et al., 2019). 19 14 Aug 21 Aug 28 Aug 04 Sep 11 Sep 18 Sep 25 Sep 02 Oct 09 Oct 100 200 300 400 500 600 700 800 900 1000 Pr es su re (h Pa ) (a) Yap 14 Aug 21 Aug 28 Aug 04 Sep 11 Sep 18 Sep 25 Sep 02 Oct 09 Oct 100 200 300 400 500 600 700 800 900 1000 Pr es su re (h Pa ) (b) Thomas G. Thompson 10 8 6 4 2 0 2 4 6 8 10 m s* *- 1 PISTON 2018 Soundings of Zonal Wind Figure 2.5: Time-height diagram of zonal wind from each sounding taken as part of the PISTON field campaign between 14 August 2018 and 13 October 2018. Soundings were taken every 12 hours from the island of Yap (top) and every 3 hours from the R/V Thomas G. Thompson during operational periods (bottom). 20 Fig. 2.6 shows both the total OLR anomalies from the seasonal cycle averaged from 0-25N, and the anomalies on the 10-20 day time scale (note the difference in color-scale) during the 2018 west Pacific monsoon season and PISTON period. Superimposed on these anomalies are the longitudi- nal positions of TC storm centers that entered the 0-25N latitude band during the period (Knapp et al., 2018, 2010). It can be seen that the TCs do occasionally project onto this timescale, but the 10-20 day band does include more than just propagating TCs (Ko and Hsu, 2006, 2009). 10- 20 day filtered anomalies during this period are generally westward-propagating, and consistently active throughout the monsoon season. This holds true when other years are selected, but only 2018 is shown here. Thus, the 2018 field campaign observed notable 10-20 variability in lower tropospheric winds (Fig. 2.5), which corresponds to westward propagating signals in OLR when filtered to this band (Fig. 2.6). The evolution of our QBWO index through the field campaign is shown in Figure 2.7. It can be seen that prominent 10-20 day variability consistently projected onto the index during the two month period. QBWO activity generally moved through each of the phases in order, and remained in a single phase for 1-2 days. According to this index, the strongest period of activity that progressed through a complete cycle occurred from roughly 11 September to 23 September, with days at least one day in each phase and an amplitude (a = √ PC12 + PC22) greater than 1.0 throughout the period. CMORPH precipitation estimates averaged across latitude in Box M (Fig. 2.1) over Mindanao are shown for this highlighted 13-day period in Figure 2.8. Fig. 2.8b shows total column water vapor and 850-hPa zonal wind anomalies from the JJAS composite mean diurnal cycle from ERA5 averaged inside box M. Mindanao is shown here rather than Luzon because Typhoon Mangkhut made a direct landfall on 14 September. From 11 Sep to 15 Sep, Mindanao experienced strong westerly winds at 850-hPa with increased moisture. Concurrently, there was relatively little pre- cipitation over the main island, with some heavy precipitation occurring over the Moro Gulf (Box D in Fig. 2.1) to the west. As the QBWO index moved through phases 7-8 on 15-16 September, drier conditions moved over Mindanao, and there was relatively little precipitation anywhere in 21 120 E 130 E 140 E 150 E 160 E 01 Jun 15 Jun 01 Jul 15 Jul 01 Aug 15 Aug 01 Sep 15 Sep 01 Oct 15 Oct (a) 10-20 Day 25 20 15 10 5 0 5 10 15 20 25 W/m2 120 E 130 E 140 E 150 E 160 E 01 Jun 15 Jun 01 Jul 15 Jul 01 Aug 15 Aug 01 Sep 15 Sep 01 Oct 15 Oct MALIKSI GAEMI PRAPIROON MARIA SON-TINH AMPIL WUKONGJONGDARI SHANSHAN YAGI LEEPI RUMBIA SOULIK CIMARON JEBI MANGKHUT BARIJAT TRAMI KONG-REY (b) Total 50 40 30 20 10 0 10 20 30 40 50 W/m2 OLR Anomalies on Various Timescales: PISTON 2018 Figure 2.6: Hovmöller plot of AVHRR OLR averaged between 0 and 25N at each longitude during 1 June- 20 October 2018, bandpass filtered to the 10-20 day timescale using a Lanczos filter with 93 weights in W m−2 (left), and OLR anomalies from the seasonal cycle defined by the average daily climatology smoothed with a 7-day running mean (right). Named tropical cyclone tracks from IBTrACS are superimposed with gray dotted lines when the storm center was inside 0-25N. 22 3 2 1 0 1 2 3 EEOF 2 3 2 1 0 1 2 3 EE O F 1 1516 17 18 19 20 21 22 23 24 2526 27 28 29 30 31 11 2 3 45 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 222324 25 26 27 28 29 30 11 2 3 4 5 6 7 8 910 11 12 13 14 Ph as e 1 Phase 2 Phase 3 Phase 4 Phase 5 Phase 6Phase 7 Ph as e 8 QBWO Index: Aug-Oct 2018 August September October Figure 2.7: Phase space diagram of the QBWO index activity from 14 August-14 October 2018 (the PIS- TON field campaign period), with the first principal component on the y-axis and the second principal component on the x-axis. The split between the 8 phases is denoted with black dotted lines, while days with an amplitude less than 1 (inside the center circle) are considered inactive, and not part of any phase. August is shown in the darkest color, with October in the lightest pink. The corresponding numbers indicate the date of each month. 23 the domain. Then from 17-20 September, the main island exhibited pronounced diurnal precipita- tion over the high topography, with westward propagation into the evening and overnight each day (most prominent on 17 Sep). Moisture was slightly higher than normal during this period, while winds started with easterly anomalies and transitioned to westerly anomalies by the 20th. The end of the cycle from 21-23 September, during phases 4-5 in our index, displayed markedly different diurnal precipitation behavior. With weakly positive moisture anomalies and westerly wind anomalies, the diurnal cycle was relatively inactive over the Moro Gulf and western Min- danao (although there was some nocturnal precipitation on 22 Sep in the Moro Gulf), while the eastern coastline experienced strong evening precipitation each day, with some indication of propa- gation to the east into the Philippine Sea. Even from a short case study, these results are consistent with other studies pointing to high moisture and offshore lower tropospheric wind as environ- mental background conditions favoring a strong diurnal cycle, which here is related to 10-20 day variability (Vincent and Lane, 2017; Natoli and Maloney, 2019; Sakaeda et al., 2020; Peatman et al., 2021). The PISTON period is used in this study to show that 10-20 variability and its impact on the diurnal cycle can show up in raw data during a major field campaign and test our index during a real event. However, a two month period is not sufficient to draw robust conclusions. Thus, we will discuss a composite analysis based on the index described above in the following sections. 2.3.2 Large Scale Variables are composited by each of the 8 phases of this index, with days on which the index amplitude is less than 1 excluded. The total number of days included in each composite can be found in Fig. 2.4. Figure 2.9 shows the large scale structure of the QBWO as captured by this index, with every other phase shown. Daily unfiltered OLR anomalies with ERA5 850-hPa vector wind anomalies superimposed are shown on the left, with ERA5 total column water vapor anomalies and total wind (not anomalies) on the right. The index captures the northwestward propagation of alternating zones of suppressed and enhanced convection, associated with anti-cyclonic and 24 11 Sep 12 Sep 13 Sep 14 Sep 15 Sep 16 Sep 17 Sep 18 Sep 19 Sep 20 Sep 21 Sep 22 Sep 23 Sep (a) Mindanao Precipitation: PISTON 2018 0 10 m s**-1 kg m**-2 (b) u850 TCWV 121 122 123 124 125 126 127 0 500 M et er s Zamboanga Peninsula/ Moro Gulf 0 1 2 3 4 5 6 7 8 m m /h r Figure 2.8: (a) CMORPH precipitation rate estimates (mm/hr) averaged from 6N-8N across Mindanao (Box M, Fig. 2.1) from 11 September 2018 to 23 September 2018, during one full cycle of the QBWO index. The average topography in this box from NOAA ETOPO2 is shown on the bottom, with the coastlines drawn as vertical dashed black lines. Note that there are some land points west of the western coastline here, part of the Zamboanga Peninsula. The horizontal dashed black lines correspond to 00 UTC, or 0800 local time. (b) Zonal wind at 850-hPa averaged across both latitude and longitude in Box M (red line) and total column water vapor (blue line) from ERA5, with the JJAS composite diurnal cycle removed at each hour. 25 cyclonic wind anomalies respectively, consistent with QBWO structure observed in prior studies (Chen and Sui, 2010; Qian et al., 2019; Yan et al., 2019). Suppressed convection dominates much of the tropical western Pacific in phase 1, with anoma- lous easterly winds and dry air pushing across the region. The remnant of a westerly monsoon surge can be seen with southwesterly winds and moist conditions over the northern South China Sea and Taiwan. By phase 3, the suppressed convection and easterly anomalies are maximized over the northern Philippines, along with a significant dry anomaly. In total wind, this phase is characterized by trade easterlies dominating the entire domain outside of the mid-latitude wester- lies on the northern fringes. Some indications of weakly enhanced convection begins to emerge in this phase around 10N, 145E. In phase 5, the enhanced convection becomes much more prominent, with a well defined anomalously cyclonic circulation centered over the Philippine Sea. Monsoon westerly winds start to strengthen over the Philippines and nearby waters, collocated with increas- ing moisture content. Enhanced convection, total column water, and westerly winds are maximized over the northern Philippines in phase 7, with an obvious monsoon surge penetrating deep into the Pacific. Overall, these structures are very similar to QBWO structures depicted in prior work (Chen and Sui, 2010; Qian et al., 2019). Figures 2.10 and 2.11 show the impact of the QBWO on precipitation across the Philippine archipelago. Daily mean precipitation anomalies generally follow the anomalies in column mois- ture shown in Fig. 2.9, consistent with many other studies highlighting the importance of moisture, particularly in the lower to middle free troposphere, for maintaining convection and precipitation (Bretherton et al., 2004; Holloway and Neelin, 2009, 2010; Kuo et al., 2017; Vincent and Lane, 2017). Enhanced precipitation is manifest in a southwest to northeast band that moves to the north- west. The vast majority of these points are statistically significant at the 95% confidence level determined via a bootstrapping method. An interesting exception is Mindanao in the southern Philippines (see Fig. 2.1), which generally does not follow the precipitation pattern of neighboring seas. There is some evidence that surges of the monsoon do not provide as significant a modulation 26 10°N 20°N 30°N (a) Phase 1 10°N 20°N 30°N (b) Phase 3 10°N 20°N 30°N (c) Phase 5 110°E 120°E 130°E 140°E 150°E 160°E 10°N 20°N 30°N (d) Phase 7 W/m^2 40 32 24 16 8 0 8 16 24 32 40 (e) Phase 1 (f) Phase 3 (g) Phase 5 110°E 120°E 130°E 140°E 150°E 160°E (h) Phase 7 kg m**-2 5 4 3 2 1 0 1 2 3 4 5 OLR and 850-mb Wind Anomaly Total Column Water Vapor and Total 850-mb Wind Figure 2.9: Composite maps by select QBWO phase over the West Pacific ocean of anomalies of OLR (W m−2) and vector anomalies of 850-mb wind from ERA5 (left column), and anomalies of ERA5 total column water vapor (kg m−2) with total 850-mb vector wind (right column). The total number of days in each composite can be ascertained from Figure 2.4 27 of oceanic convection near and south of this island when compared to islands further north (Natoli and Maloney, 2019; Xu et al., 2021). 6°N 8°N 10°N 12°N 14°N 16°N 18°N (a) Phase 1 (b) Phase 2 (c) Phase 3 (d) Phase 4 118°E120°E122°E124°E126°E 6°N 8°N 10°N 12°N 14°N 16°N 18°N (e) Phase 5 118°E120°E122°E124°E126°E (f) Phase 6 118°E120°E122°E124°E126°E (g) Phase 7 118°E120°E122°E124°E126°E (h) Phase 8 10 8 6 4 2 0 2 4 6 8 10 m m /d ay Daily Mean Precipitation Rate Anomaly (QBWO) Figure 2.10: Anomalies in daily mean CMORPH precipitation rate composited by QBWO phase, with statistical significance at the 95% confidence level shown as dots. The variability of the amplitude of the diurnal cycle through the QBWO cycle is noted in Fig. 2.11. Diurnal amplitude is defined in this study as the amplitude of the first diurnal harmonic of the composite diurnal cycle. A strong diurnal cycle begins to emerge over Mindanao in phase 2, peaking there in phase 3. This signal is also present in the Moro Gulf, the small body of water to the southwest of Mindanao, likely indicating offshore propagation from land-based convection (Natoli and Maloney, 2019). The central Philippines and Luzon see strong diurnal cycles maximizing in Phases 4 and 5, still about 1/4 cycle ahead of the moisture maximum which occurs around phase 7. As in many prior studies examining the impact of the BSISO on the diurnal cycle in the 8-phase framework, the amplitude of the diurnal cycle over the northern Philippines (Figure 2.11) is not 28 6°N 8°N 10°N 12°N 14°N 16°N 18°N (a) Phase 1 (b) Phase 2 (c) Phase 3 (d) Phase 4 118°E120°E122°E124°E126°E 6°N 8°N 10°N 12°N 14°N 16°N 18°N (e) Phase 5 118°E120°E122°E124°E126°E (f) Phase 6 118°E120°E122°E124°E126°E (g) Phase 7 118°E120°E122°E124°E126°E (h) Phase 8 5 4 3 2 1 0 1 2 3 4 5 m m /d ay Precipitation Diurnal Cycle Amplitude Anomaly (QBWO) Figure 2.11: Anomalies in the amplitude of the CMORPH precipitation rate diurnal cycle by QBWO phase. Anomalies are calculated as the difference in diurnal amplitude between each phase composite, and ampli- tude of the JJAS composite diurnal cycle. Statistical significance at the 95% confidence level is shown as dots. 29 in phase with the daily mean precipitation (Peatman et al., 2014; Xu and Rutledge, 2018; Natoli and Maloney, 2019; Chudler et al., 2020). Despite widespread oceanic convection and abundant moisture in phase 7, the amplitude of the diurnal cycle is strongly suppressed over large islands of the Philippines. The strongest diurnal cycle tends to occur several phases before the maximum in daily mean precipitation and column moisture, when winds are still weakly easterly (Fig. 2.9f,g). Generally, this is consistent with the impact of the BSISO on the diurnal cycle. In subsequent sections, the differences between the diurnal cycle behavior associated with the QBWO and the BSISO are examined in detail with the goal of elucidating the mechanisms important to diurnal cycle regulation. 2.3.3 Luzon Luzon is the largest and most populous island of the Philippines, and presents an excellent case for examining the diurnal cycle due to the north to south orientation of its coastline and mountain ranges (Fig. 2.1). Fig. 2.12 shows Hovmöller plots of composite diurnal cycles from each QBWO phase to better interpret offshore propagation. CMORPH precipitation rate is averaged across lati- tude inside box L (Fig. 2.1), which covers northern Luzon, and shows a strong diurnal cycle over land peaking in the late afternoon for all phases. While the diurnal cycle is present in all, there is variability in its prominence and behavior. Phase 3, for example, has a weaker precipitation maximum and some initial propagation offshore both east and west, but precipitation dissipates rather quickly. In phases 4 and 5 (which have the strongest diurnal cycle amplitude anomalies in Fig. 2.11), precipitation rate maximizes over the highest topography and then persists much later into the night while propagating offshore, with the westward direction favored. Oceanic precipi- tation increases further in phase 6, while phases 7-8 shows a constantly elevated precipitation rate offshore (particularly west of Luzon), with lesser diurnal variation. There still some evidence of a diurnal cycle over the highest elevations of the island. While the diurnal cycle over western Luzon and the South China Sea appears to peak around phase 5, there is a notable asymmetry. The diurnal cycle on the eastern part of the island appears 30 08:00 11:00 14:00 17:00 20:00 23:00 02:00 05:00 Ho ur (P HT ) (a) Phase 1 (b) Phase 2 (c) Phase 3 (d) Phase 4 08:00 11:00 14:00 17:00 20:00 23:00 02:00 05:00 Ho ur (P HT ) (e) Phase 5 (f) Phase 6 (g) Phase 7 (h) Phase 8 118 120 122 124 Longitude ( E) 0 500 1000 M et er s 118 120 122 124 Longitude ( E) 118 120 122 124 Longitude ( E) 118 120 122 124 Longitude ( E) 0 6 12 18 24 30 mm/day Diurnal Propagation by QBWO phase Figure 2.12: Hovmöller diagrams of the composite diurnal cycle of CMORPH precipitation rate (mm/day) for select phases of the QBWO index. Precipitation rates are averaged across latitude in box L (Fig. 2.1), with corresponding longitude noted below. The average elevation of topography from NOAA ETOPO2 inside box L is shown at the bottom for reference. 31 stronger in phase 1, with some weak propagation into the Philippine Sea. This asymmetry has also been noted for the impact of both the BSISO/MJO and some convectively coupled equatorial waves on the diurnal cycle (Ichikawa and Yasunari, 2006, 2008; Sakaeda et al., 2017, 2020; Natoli and Maloney, 2019), and warrants a closer look. Figure 2.13 shows the diurnal cycle over certain subsets of the island, with boxes of spatial averaging shown in Fig. 2.1. Fig. 2.13a-c show the composite diurnal cycles in these boxes for select phases of the QBWO and the BSISO, according to the Lee et al. (2013) index. The orange lines show the phase with the largest diurnal range (difference between daily maximum and daily minimum precipitation rate) in the composite, while the blue lines show the phase with the smallest. These results were also considered for the diurnal amplitude, and the conclusions are similar. The right column shows the progression of the diurnal range and daily mean precipitation rate through each of the 8 phases of both indices. The daily mean precipitation rates track together very closely between the BSISO and QBWO in Fig. 2.13d-f for each region. This indicates that the phase numbers are approximately equivalent in terms of proximity to the peak of the large scale convection associated with the feature of interest. Generally, daily mean precipitation varies slightly more strongly with QBWO phase than with BSISO, but the differences are modest. The diurnal range is also remarkably similar. Over northwest Luzon (Fig. 2.13e) and the coastal South China Sea (Fig. 2.13d), the largest range of the diurnal cycle leads the daily mean precipitation by about 1/4 cycle in both the QBWO index and the BSISO index. The magnitude of the change in diurnal range appears similar for both indices despite the slightly stronger modulation of the daily mean precipitation by the QBWO. The details of the diurnal cycle (Fig. 2.13a-c) look remarkably similar as well. Over land in northwest Luzon (Fig. 2.13b), the highest amplitude phases have a sharply enhanced afternoon peak compared to the JJAS mean, but precipitation is strongly suppressed at all other times of the day. In the smallest diurnal range phases for each index, northwest Luzon sees consistently elevated precipitation rates throughout the day, with a slight bump during the evening peak that doesn’t quite reach the JJAS mean precipitation rate at that time. The behavior over the South 32 12:00 15:00 18:00 21:00 00:00 03:00 06:00 09:00 0 10 20 30 40 Pr ec ip ita tio n Ra te (m m /d ay ) (a) Coastal South China Sea QBWO Phase 5 QBWO Phase 1 BSISO Phase 5 BSISO Phase 1 Jun-Sep Mean 1 2 3 4 5 6 7 8 15 10 5 0 5 10 15 Pr ec ip R at e An om (m m /d ay ) (d) Coastal South China Sea 12:00 15:00 18:00 21:00 00:00 03:00 06:00 09:00 0 10 20 30 40 Pr ec ip ita tio n Ra te (m m /d ay ) (b) Northwest Luzon QBWO Phase 5 QBWO Phase 7 BSISO Phase 3 BSISO Phase 7 Jun-Sep Mean 1 2 3 4 5 6 7 8 15 10 5 0 5 10 15 Pr ec ip R at e An om (m m /d ay ) (e) Northwest Luzon 12:00 15:00 18:00 21:00 00:00 03:00 06:00 09:00 Time (PHT) 0 10 20 30 40 Pr ec ip ita tio n Ra te (m m /d ay ) (c) Northeast Luzon QBWO Phase 2 QBWO Phase 4 BSISO Phase 8 BSISO Phase 3 Jun-Sep Mean 1 2 3 4 5 6 7 8 QBWO or BSISO Phase 15 10 5 0 5 10 15 Pr ec ip R at e An om (m m /d ay ) (f) Northeast Luzon QBWO Diurnal Range QBWO Daily Mean BSISO Diurnal Range BSISO Daily Mean QBWO vs. BSISO Diurnal Cycles Figure 2.13: (a-c) Composites of the spatially averaged diurnal cycle of CMORPH precipitation rate (mm/day) in the QBWO (solid) and Lee et al. (2013) BSISO (dotted) phase with the highest diurnal range (orange), the QBWO/BSISO phase with the smallest diurnal range (blue), and the full JJAS composite (dot- ted black). Spatial averaging is done over ocean points inside box A (a), and land points inside boxes B (b) and C (c). (d-f) The corresponding daily mean precipitation (blue) and diurnal range (red) in mm/day of each phase’s spatially averaged composite diurnal cycle, by QBWO phase (solid) and BSISO phase (dotted). Each box covers a domain near Luzon. 33 China Sea (Fig. 2.13b) is also similar, with phase 5 in each index exhibiting heavier precipitation during the typical peak of around 2100 when westward propagating precipitation arrives. Phase 1 in each index has a fairly constant precipitation rate all day, indicating that little convection that initiates over land is propagating offshore (as also seen in Fig. 2.12a). The two modes also exhibit the same east/west asymmetry, with the largest diurnal ranges coming after the convective maximum for each index in the eastern part of the island. Over land in northeast Luzon (Fig. 2.13f), the strongest diurnal cycle occurs after the peak in daily mean precipitation, in phases 8 and 1 for the BSISO, and phases 1 and 2 for the QBWO. Precipitation rate over this region throughout the day (Fig. 2.13c) exhibits similar behavior at the end of the convective maximum (phases 8, 1, 2) compared with northwest Luzon (Fig. 2.13b) in the phases leading up to the convective maximum (phases 3-5) in both indices. This is consistent with con- clusions drawn by Sakaeda et al. (2020) on diurnal cycle asymmetry through the passage of a large scale disturbance like the MJO or an R1 wave. Overall, the diurnal cycle behavior over Luzon associated with the QBWO index strongly resembles the results previously seen for the BSISO. This motivates the hypothesis that the impact on the diurnal cycle is not unique to either mode, rather, that each mode impacts the background state near Luzon similarly, leading to congruent diurnal cycle behavior. 2.3.4 Mindanao In Mindanao, the diurnal cycle contributes much more to variability in daily mean precipitation than it does over Luzon (Natoli and Maloney, 2019). As such, the disconnect between the diurnal range and daily mean precipitation is not as distinct as for Luzon. Figure 2.14 demonstrates that the amplitude of the diurnal cycle is more closel