Seasonal Variability and Monthly Trends in Upper Tropospheric Humidity for the Period 1979–2020 ^†

Abstract

The monthly variability of Upper Tropospheric Humidity with respect to ice (UTHi) is examined during the forty-year period 1979–2020. Trends per decade are calculated for every month separately and for 10° latitude bands. Statistical significance is estimated with the Mann–Kendall test. Results show significant positive UTHi trends in northern and southern midlatitude regions. In the northern midlatitudes, the strongest trends are observed in November, December and February and in the southern midlatitudes in April, May and June. These trends range between 0.5 and 1.2% per decade. In contrast, tropical regions display weak or negative trends. These patterns highlight regional and seasonal differences in upper tropospheric moisture changes, which are critical for understanding water vapour feedback and its role in climate variability and long-term atmospheric moisture behaviour.

1. Introduction

Water vapour in the troposphere plays a crucial role in Earth’s climate. It strongly affects planetary radiative balance directly by acting as a greenhouse gas on its own and indirectly by influencing cloud formation, aerosols and atmospheric chemistry [1]. In the upper troposphere where the temperatures reach well below −40 °C it is useful to express water vapour levels as relative humidity with respect to ice. Upper Tropospheric Humidity with respect to ice (UTHi) near the 200–500 hPa level critically influences the appearance of Ice Supersaturated regions (ISSRs), where the UTHi may exceed 100% and cirrus clouds are potentially formed. ISSRs and the associated cirrus clouds can affect Earth’s radiative balance by modifying the solar and infrared radiation within the atmosphere, depending on the cirrus clouds’ physical properties and producing either cooling or warming effects [2]. Since temperature is an important component of saturation vapour pressure [3], climate change can affect the frequency and degree of high UTHi values resulting in changes in the extent of ISSRs and consequently, the formation and coverage of cirrus clouds.

Given the influence of cirrus clouds on radiative transfer and the close relationship between their formation and UTHi, examining the spatiotemporal variability and trends in UTHi provides valuable insights into changes in the upper-tropospheric moisture budget and related climate feedback. This work presents an assessment of monthly UTHi variability and its decadal trends nearly over the globe (60° S to 60° N), highlighting regional and seasonal patterns relevant to cirrus cloud formation zones such as the Intertropical Convergence Zone (ITCZ), and the subtropical subsidence zones.

2. Materials and Methods

This study utilizes data collected from the HIRS (High-Resolution Infrared Radiation Sounder) instrument onboard the National Oceanic and Atmospheric Administration (NOAA 06, 07, 08, 09, 10, 11, 12, 14, 15, 16, 17) and Meteorological Operational Satellite Programme (METOP 02, 01) satellites. The NOAA 06-14 carried the HIRS2 version of the instrument, whereas the NOAA 15-17 and METOP 02, 01 carried the HIRS3/4 versions. From the HIRS data, brightness temperatures at channels 12 (T₁₂) and 6 (T₆) were extracted. The change in the HIRS versions from HIRS2 to HIRS3/4 as well as other changes, e.g., wavelength change from 6.7 μm in HIRS2 to 6.5 μm in HIRS3/4, filter functions, the instantaneous field of view etc., caused inhomogeneity between the measurements coming from different satellites. To ensure data consistency and an account for the jump in the time series around 1999 due to the change from HIRS2 to HIRS3/4, intercalibration methods were applied to the data [4]. The retrieval of UTHi from the brightness temperature data were performed with a second order retrieval, based on Equation (1), using different sets of fit coefficients as mentioned below.

$$\mathrm{U}\mathrm{T}\mathrm{H}\mathrm{\%}={\displaystyle \frac{\exp \left(\mathrm{a}+\mathrm{b}\times {\mathrm{T}}_{12}+\mathrm{c}\times {\mathrm{T}}_{12}^2\right)}{{\mathrm{a}}^{\mathrm{\prime }}+{\mathrm{b}}^{\mathrm{\prime }}\times {\mathrm{T}}_6}}$$

(1)

Gierens and Eleftheratos [4] applied different sets of fit coefficients in the T₁₂ data (a, b and c), each for every HIRS version. Coefficients are provided in

Table 1. Wavelength-dependent constants used for UTHi/w computation in the second-order retrieval equation (adapted from Gierens and Eleftheratos [4]).

	λ (μm)	a	b (1/K)	c (1/K2)
UTHi	6.7	47.69	−0.2846	3.522 × 10−4
	6.5	50.05	−0.3109	4.063 × 10−4
UTHw	6.7	43.36	−0.2619	3.266 × 10−4
	6.5	45.50	−0.2868	4.063 × 10−4

. The a’ and b’ are constants (a′ =10.236, b′ = −0.036).

In addition to the aforementioned intercalibrations, bias corrections were also implemented to counter additional instrumentation issues and eliminate remaining minor inconsistencies in the T₁₂ gap. More specifically, regarding the T₆ data, the continuous increase of CO₂ concentrations from about 330 to 410 ppmv between 1979 and 2020, resulted in a recorded decrease of approximately 2 K [5]. To remove this false trend, a correction algorithm was applied to the data adjusting them proportionally to the month-to-month increase of the CO₂ concentrations. Minor inconsistencies in the T₁₂ data were identified with a timeseries analysis method known as breakpoint analysis, a method commonly used to identify changes in trends in timeseries [6,7]. To remove such inconsistencies, a bias correction applied in the numerator of the UTHi retrieval algorithm was applied.

Detailed information on the application of the intercalibration techniques and bias correction methods can be found in Benetatos et al. [8]. After all the bias corrections, daily, monthly and decadal UTHi grids with a spatial resolution of 2.5° × 2.5° between 60° N and 60° S were produced. We present the UTHi monthly climatology as well as monthly trends of UTHi for the period 1979 to 2020 per 10-degree latitude zones. Statistical significance is estimated with the Mann–Kendall test for monotonic trend.

3. Results and Discussion

The analysis reveals a clear seasonal cycle in UTHi, with pronounced variability in the tropics, the subtropics and the subarctic regions (Figure 1). Hemispheric contrasts are notable throughout the year. It can be clearly seen that areas of increased and decreased humidity follow the Hadley circulation. Tropics are dominated by upward motions that concentrate humidity in the upper tropical troposphere whereas subtropical regions are dominated by downward motions which create clear weather conditions and decreased moisture.

During boreal winter months (December–February), the tropical belt exhibits enhanced humidity near the equator, particularly over the Maritime Continent, central Africa as well as central South America. As the year progresses, a northward shift in high UTHi regions is evident, peaking during boreal summer (June–August), with increased moisture observed over Southeast Asia, central America, northern South America, and the Sahel. This shift aligns well with the northern migration of the ITCZ during the boreal summer and monsoonal activity. The southern migration of the ITCZ can be observed better in northern South America and central Africa during the austral summer months where the high humidity areas are located slightly below the equator. The driest conditions are consistently observed over the subtropical oceanic subsidence zones. In summary, UTHi in the tropics consistently reaches values above 70% with fluctuations in shape, size and geographical extent of high UTHi areas whereas extratropical regions exhibit lower UTHi values (<40%) and moderate fluctuations within the year.

Figure 2 illustrates the decadal trends in 10° zonally averaged UTHi for each month, spanning from 60° S to 60° N. Green triangles signify statistical significance at 95% confidence level.

Table 2. Trend values for all months and per 10° latitude zones. Statistically significant trends (95% confidence level) are shown in bold.

Zone	Jan.	Feb.	Mar.	Apr.	May	Jun.	Jul.	Aug.	Sep.	Oct.	Nov.	Dec.
50–60° N	0.64	1.53	1.03	0.37	0.05	−0.21	−0.58	0.13	0.41	0.52	1.03	1.16
40–50° N	0.48	0.72	0.12	0.47	0.30	0.03	−0.19	0.14	0.51	0.62	1.10	0.86
30–40° N	0.41	−0.13	0.17	0.81	0.39	0.03	0.11	0.16	0.36	0.72	1.02	0.80
20–30° N	0.66	−0.13	−0.13	0.26	−0.22	−0.10	−0.44	−0.52	−0.30	0.14	0.88	0.14
10–20° N	0.67	0.18	0.30	0.11	0.12	−0.53	−0.64	−0.94	−0.52	−0.08	0.07	−0.19
0–10° N	0.27	0.30	0.36	−0.28	−0.15	0.25	0.14	0.15	0.42	−0.12	−0.35	−0.32
0–10° S	−0.48	−0.90	−0.54	−0.39	−0.17	0.24	0.11	0.26	0.32	0.02	−0.18	−0.39
10–20° S	−0.09	−0.76	−0.27	−0.34	−0.20	0.25	−0.13	0.14	−0.38	−0.11	−0.17	−0.24
20–30° S	0.07	−0.29	−0.25	0.23	0.17	0.18	−0.22	−0.18	−0.02	−0.08	0.23	0.06
30–40° S	0.12	0.46	0.33	0.78	0.53	−0.12	−0.57	−0.49	−0.46	0.12	0.45	0.43
40–50° S	−0.16	0.24	0.60	0.85	1.01	0.69	0.14	0.07	−0.39	0.04	0.18	0.23
50–60° S	−0.47	0.07	0.68	0.58	0.89	1.25	0.99	1.22	0.69	0.39	−0.01	−0.43

contains all the trend values per 10° latitude zone and per month which are displayed in Figure 2. Positive trends are observed in the midlatitudes of the northern hemisphere (30–60° N) in February (40–50° N and 50–60° N), March (only 50–60° N), April, September, October, November and December. The majority of these trends are statistically significant. Southern midlatitudes exhibit positive values consistently in March, April, May, June (40–50° S and 50–60° S). July and August exhibit a positive trend only in the 50–60° S zone. As before, the majority of these trends are statistically significant. The tropics exhibit a rather consistent behaviour across the year with either small positive or negative trends.

Moreover, a clear pattern of seasonal trends emerges. January exhibits positive trends in the northern hemisphere and negative or no trends in the southern hemisphere. In February and March, the northern hemisphere shows small positive or no trends in the tropics and subtropics, respectively, and high positive trends in the midlatitudes. Regarding the southern hemisphere, we observed high negative trends in the tropics, and a trend reversal as latitude increases. In April, trends are small and negative in the tropics, shifting to positive in the subtropics and intensifying further in the midlatitudes. In May, the southern hemisphere exhibits highly significant positive trends in midlatitudes. The trends decrease in the tropics and remain rather low in the northern hemisphere. June, July and August exhibit a quite similar trend pattern, with positive trends in 50–60° S and 40–50° S. The trend drops and reverses in the 30–40° S zone (in June the trend is close to zero) and increase again to close-to-zero values in the tropics. Trends drop strongly below zero again in the 10–20° N and then increase again to close-to-zero values in the northern midlatitudes (July trend in the 50–60° N drops to a negative value). September exhibits positive trends in the tropics and small negative trends as we move towards the subtropics of both hemispheres. Trends are positive in the northern hemisphere midlatitudes whereas in the southern hemisphere trends remain negative until 40–50° S and then become strongly positive in the 50–60° S. October, November and December exhibit a broadly similar pattern of trends. Trends in the northern midlatitudes are always strongly positive and significant. Trends decrease in the northern tropics, reverse slightly to negative values, increase in the southern low midlatitudes and fall back close to zero for November and below zero for December. In October, trends remain very close to zero as we move away from the northern midlatitudes to the southern midlatitudes. According to the figure, it appears that the strongest positive and significant trends are observed in the northern midlatitudes in November and December as well as February (just before and during the boreal winter) and in the southern midlatitudes in April, May and June (just before and during the austral winter). July and August also exhibit strong positive trends in the 50–60° S.

4. Conclusions

This study presented the monthly variability of UTHi as well as the monthly trends of the UTHi anomalies over the period 1979–2020 using HIRS satellite measurements. The findings highlight distinct regional and seasonal patterns in UTHi, closely linked to atmospheric circulation systems such as the Hadley cell and the ITCZ. Statistically significant positive trends in the northern midlatitudes have been found just before the start of and during the boreal winter. Accordingly, in the southern midlatitudes, high positive trends have been found just before and during the austral winter. In contrast, tropical regions exhibit relatively stable or slightly negative trends, with some exceptions during boreal summer and winter months.