2019-2022

Time evolution of meteorological conditions 

[File = Met_IntervalData_270619-060422.csv, n = 1.946.777 observations]

The time series data presented here are generated by the Vaisala weather station WXT536 from the HydroMet Buoy. They cover the entire deployment period of the HydroMet buoy, from 06/29/19 to 06/04/22, for 7 parameters: Air temperature (AirTemp_ºC) | Air humidity (RhPct_%) | Atmospheric pressure (BaroPress_mm_Hg) | Rainfall (RainAmount_mm) | Hail (HailAmount_hitPercm2) | Incident solar radiation (SolarRad_W/m2) | Wind direction (WindDir) Wind speed (WindSpdAvg_m/s). 

The LoggerNet program generates two data files in the Campbell datalogger of the Buoy:

(a) Data at every 5 min : 278,111 lines x 7 parameters = 1,946,777 data. WARNING: Manipulating so much data with Excel with 8 MB of RAM is very heavy and the computer may hang up. It becomes indispensable to increase the memory to 16 MB and/or to use the R programming language (for statistical calculation and graphics, www.r-project.org , Foundation for Statistical Computing), open source and multiplatform.

b) Daily data (24 hours): 965 lines x 7 parameters = 6,755 data.

NOTE: The Vaisala WXT536 weather station operated without failure from 06/27/19 (date of deployment) until 01/01/22 (except for a micro outage from 01/10/19 to 02/10/19). Beginning January 2022, it started to exhibit intermittent outages for a few successive days (Table 1). We found that it resulted from a combination of circumstances: a) The Buoy battery charge started to decrease only after almost 3 years; and b) Starting in January, with the rainy period and increased cloudiness, the number of sunny hours per day was not sufficient to recharge the battery to its optimal level with the solar panels. On 22/03/22 we reinstalled the battery after > 36 hours of recharging on the grid, it regained its optimum level and the operation of the Vaisala station was restored without interruption. 

1.- Air Temperature (ºC)

Note: The Vaisala WXT 536 weather station was damaged on 16/08/22 and didn't function since then. It was only replaced by a new identical weather station on 07/11/23.  This resulted in a long data gap.  

The evolution of air temperature (Fig. 1a), with a frequency of 5 min, fluctuates with a large amplitude of 8-12 ºC between day and night, and follows a sinusoidal oscillation with an annual amplitude, minima in dry seasons (June-July, -1ºC, 0ºC; also in September-October in 2019) and maxima in wet seasons (November-December, ≤ 18 ºC). This oscillation is related to the seasonal variation of the sun's height, notorious with the level of incident solar radiation (see Figs. 6a-6b). The occurrence of negative temperatures is very rare over the water mirror (June 2021). More frequently the minimums are > 0 ºC. This is the consequence of the tropical location of Lake Titicaca and the mild climate produced by this large body of water that stores and retains heat much longer than the land. 

The evolution of air temperature (Fig. 1b), with daily frequency, of course follows the same pattern. It is noted that the daily amplitude is higher during dry periods and lower during wet periods. The annual average reaches ~9.6 ºC. A trend line fitted to the data suggests a slight temperature increase of the order of ≤ 0.2 ºC in 3 years (i.e. +0.67 ºC / decade, which may generate a trend* towards +5,36 ºC on 2100 if unpredicted changes would not occur until then**) . Although we did not test its statistical significance, this increase demonstrates that global warming is already underway in Lake Titicaca and the Altiplano.

*At least a decade (10 years) of data is required to estimate a true 'trend'. Here we count only on less than 4 years of data.

**A warming of ≥ 5 ºC is considerable as compared to the ~2 ºC threshold that should not be exceeded according to the Paris Climate Agreement (CCNUCC, 2015) at the risk of exceeding the point of no return. Yet tropical regions at high altitude, which is the case of the Altiplano and Lake Titicaca, are expected to suffer the highest climate warming (Bradley et al., 2004. Geophysical Research Letters, Vol. 31, L16210). Socio-political scenarios where global warming would reach 4ºC in Bolivia have already been analyzed by Hoffmann & Requena (2012, Bolivia en un mundo 4 grados más caliente. Escenarios sociopolíticos ante el cambio climático para los años 2030 y 2060 en el altiplano norte. Instituto Boliviano de la Montaña; Fundación PIEB; La Paz; 168 p.). 

2.- Air Humidity (RhPct, %)

Air humidity varies widely during the day from ≥ 10 to ≤ 90 % during dry seasons (May-September) and to a lesser extent from ≥ 20-40 to ≤ 95 % during rainy seasons (December-April). It is noted both with a frequency of 5 min (Fig. 2a) or daily (Fig. 2b). Apparently, the first three months of the years (January-April) are drier (minimum 30 %, maximum 90 %).

3.- Atmospheric pressure (barometric, mm Hg)

4.- Pluviometry (mm/5 min)

With high-frequency measurements (5 min; Fig. 4a), the durations of the most intense rainfall periods are observed to decrease over the three-year study: from 12 Sept 2019 to 09 May 2020; from 06 Sept 2020 to 04 May 2021; from 31 Oct 2021 to 29 April 2022. Also, maximum rainfall intensities drop from 11 mm/5 min in 2019-2020, 9 mm/5 min in 2020-2021, to ≤ 6 mm/5 min in 2021-2022, which apparently contradicts the hypothesis of increasing event intensities and frequencies; however, our study period is too short to infer meaningful conclusions regarding significant future trends. Nevertheless, these are indications of a marked increase in aridity conditions. In other words, climate change is already occurring. 

With the accumulation of the data on a daily frequency (Fig. 4b), the trend of reduction of maximum intensities is less noticeable: ≤ 40 mm/day in 2019-2020, ≤ 25 mm/day in 2020-2021 with a peak at 47 mm/day, ≤ 20 mm/day in 2021-20222 with a peak at 30 mm/day. However, the concentration of rainfall in shorter periods if confirmed: September-May in 2019-2020, September-April in 2020-2021, and November-March in 2021-2022.

The monthly accumulation of rainfall (Fig. 4c) confirms the trends in: a) the reduction of intensity over the three years: maximums of 288 mm/month in 2019-2020, 162 mm in 2020-2021, and 135 mm in 2021-2022; and b) the concentration of rainfall in shorter periods: July-May in 2019-2020, September-April in 2020-2021, and November-March in 2021-2022. 

Table 2 illustrates the data in Fig. 4c, where the reduction in rainfall is noted both when comparing rainy seasons and when comparing years. Fig. 4 graphically illustrates this reduction between rainy seasons.

5.- Hailstorm (hits per cm2)

At high frequency (5 min), the hailstorm periods were quite similar between 2019-2020 (December-April) and 2020-2021 (December-April, with a point occurrence in October), however very different in 2021-2022 with a single point occurrence at the beginning of February (Fig. 5a). Event intensities were very different: 0.1-0.7 hits/cm2 in 2019-2020, 0.1-0.9 hits/cm2 in 2020-2021, and 0.2 hits/cm2  in 2022.

The daily frequency accumulation (Fig. 5b) reveals the same pattern with maximums in January-March, except in 2022. 

6.- Solar radiation (W/m2)

At high frequency (5 min), incident solar radiation follows a sinusoidal oscillation (Fig. 6a), with maxima during rainy periods (up to 550 W/m2, December-March), when cloudiness is greater and more frequent. In fact, according to seasonality, it corresponds to summer in Bolivia, which is the time when the greatest solar radiation reaches the earth's surface. On the other hand, the minimums are found during the dry periods (≤ 280 W/m2, June-July) of the winter.

In daily frequency accumulation (Fig. 6b), it is noticeable that the average is quite stable at ~90 W/m2, with a sharp fluctuation of the maxima as described above.

7.- Wind direction (degrees)

8.- Average wind speed (m/s)

At high frequency (5 min, Fig. 8a), the wind speed oscillates with a frequency apparently less than a year, with maxima during the period of the winds (August-September), as is known, up to 17 m/s, or > 61 km/h. 

In daily frequency (Fig. 8b), averages are observed from ~4.7 m/s or ~17 km/h, to ~6 m/s or ~22 km/h during the wind period. Minimums range from 0-1 m/s and maximums from 10-17 m/s or 36-61 km/h.

These preliminary graphical analyses only present patterns and amplitudes of raw values. Time series analyses will allow better visualization and quantification of the phenomena.

9.- To remember

This preliminary analysis shows that all relevant meteorological indicators suggest that climate change, such as warming and seasonality alteration, is already accelerating. This should raise everyone's environmental awareness. Despite the short period of our study (34 months), awaiting the results of the statistical analysis of the time series, the following are particularly noteworthy:

(a) The slight gradual gradual increase in air temperature, around < 0.2 °C in 3 years (Fig. 1b), i.e. 0.67 °C per decade.

b) The slightly higher average (70 %) and lower amplitude range of air humidity in January-March 2022 (40-90 %), compared to previous years (Figs. 2a, 2b).

c) The most notorious trend in the reduction of seasonal and annual rainfall (around -200 mm/year; Figs. 4c, 4d), as well as the reduction of the length of the rainy period, progressively shorter: July-May in 2019-2020, September-April in 2020-2021, and November-March in 2021-2022.

d) Progressively more intense hailstorm events (2 isolated events of 0.9 hits/cm2 in February and April 2021) and then scarcer (a single event of 0.2 hits/cm2 in February 2022) (Fg. 5b).

e) A slight trend towards an increase in the level of incident solar radiation during the rainy season (summer, December-January) in relation to the previous two years (Fig. 6b).

f) More frequent wind gust events up to maximum speeds above 16 m/s (> 58 km/h), as in 2021 compared to the previous two years (Fig. 8b).

This context is quite worrisome and requires further attention. This fully justifies the sustainability of the OLT observatory in the long term, to check trends and attempt to anticipate extreme events. The OLT also generates two important benefits for: 1) optimizing the forecasts of regional and global climate models, and 2) refining the water balance of the lake, taking advantage of measurements above the water mirror (in addition to SENAMHI data in the basin).