Report on SHM datasets

Report on SHM datasets

Abstract

The paper looks at the simulation archive history and how it provides the weather data that is used to forecast change in the weather patterns. The paper also looks at the relationship between the current, past and future weather patterns. The weather forecasting is majorly done based on past experiences. However, there is a significant difference between the scopes of the archive, which is also another critical area of focus on the paper.

Most importantly, the paper looks at the simulation of the data archived, which is the significant way of data analysis in the process of data simulation. The aspects of weather information and reading are also part of the study. There are various diagrams in the data set and the mat lab report which shows different data records and patters. Therefore, the system has to be analyzed through the analysis principles and frameworks of the simulation.

The paper also looks at the different types of data. For example the sorted data in the simulation archive which is not measured. It then processes to look at the data in the selected area. It finally looks at the grid cell where the weather elements of the chosen field are calculated based on the altitude. The altitude determines the difference in the coordinates and the grid locations. Finally, the paper explores the data sale hence describing the importance of the study.

The chaotic nature limits the atmosphere to limit its short-term predictability. Additionally, little knowledge is available on how the weather forecasting difficulty could be affected by the anthropogenic change in climate (Stone et al., 2019). Despite the increase in complexity and resolution of the systems of weather forecasts, atmospheric forecasts face a high level of limit in its predictability. That roots from the atmospheric attractor’s intrinsic properties which is a high-dimensional geometrical object used by all the possible atmospheric states for settling. It is customarily called the “butterfly effect” or initial conditions’ dependence. The turbulence in the atmosphere injects energy at all temporal and spatial scales thus generating disorder and causing limits in the short-term predictability.

Introduction

Predictability depends mainly on time and space so that changes in interpretation and its detention is very complicated. Predictability is indeed never constant, being strongly affected by the large-scale configuration of the atmosphere from which the forecast is started. For instance, the mid-latitude transitions from the blocked to states of atmospheric zonal have fewer predictions compared with a zonal flow that is large scale. The large-scale zonal flow is a robust eastward jet bearing few meridional oscillations. If the predictable vs unpredictable patterns’ proportion were to experience change with the anthropogenic forcing, the intrinsic predictability of the atmosphere would consequently change. There is, therefore, great difficulty with the task of changes diagnosis in the predictability of the atmosphere and also in the dynamics of the mid-latitude atmosphere and the anthropogenic forcing. Consequently, it presents a significant challenge in achieving this when carrying out focusing on the averaged quantities like climate variability.

The figure below illustrates the evolution of the global monsoon rainfall over land according to the estimation of various reanalyses and GPCP

Body

Historical reanalyses join the observations from the past weather with the simulations of the modern numerical models of weather prediction for purposes of providing a weather history that is consistent and global. The recent times witnessed a new reanalysis. The reanalysis allows the ocean’s observation to have an impact on the atmosphere, and vice versa. History of weather construction needs thoughtful blending of model simulations and observations. To understand the reason why imagine attempting to understand the climate change effects on rainfall in Australia through the representation of the in-depth’s past century of precipitation. In that case, observations only provide a very narrow view that is limited to the variables observed at specific locations with the stations of weather which have lasted for decades. The observations will possess varying accuracy back in time (Stone et al., 2019). On the other hand, unconstrained model simulations might come with the provision of sufficient average precipitation trends but will fail to capture events of high rainfall or droughts.

Reanalyses combine the observations using model forecasts to capture the best view of the whole picture. Some of the variables obtained include; temperature, precipitation, and the pressure of upper-atmospheric winds, soil moisture, velocity of the middle atmosphere. Model forecasts and observations are combined with the use of the similar data assimilation methods that are used by the weather centres around for the generation of an analysis or a guess that is best of the current global weather system’s state for initializing their weather forecasts.

In principle, a weather history could be strung together from the operational analyses. However, since the centres of weather continually update the models of the forecast, observing systems, a long routine analysis series, and assimilation methods have artificial changes. To avoid some of the artefacts, reanalyses serve to fix an assimilation method and a forecast model and move back in time to carry out a reanalysis on particular networks of observation.

The fix on the observation network is crucial when reanalyses have assimilated all the observations available at any time. That is when they would have produced spurious changes in consideration of the system’s significant changes. Reanalyses that are traditional assimilate upper-air satellite data and observations and therefore can only extend as far back as the data are consistently and globally available. That is 1957 for observations of upper-air and 1979 for that of satellite data. For purposes of reconstructing the long records that are required for purposes of climate studies, and extreme events’ investigation like marine winds and surface pressure, most historical reanalyses only utilize observations from conventional surface-levels like marine winds, surface pressure. Thus, it sacrifices a level of consistency and accuracy. The reanalyses include the NOAA-CIRES Twentieth Century Reanalysis project and the first reanalysis of ECMWF, ERA-20C spanning more than a century. They have been utilized for the investigation of changes in intensity and frequency of floods, droughts, and tropical cyclones. That is to gain a better understanding of Southern Oscillation teleconnections/El Nino, and also for the determination of the suitable climate for TseTse fly and the relationship it has to African economic development.

The figure illustrates the evolution of the global monsoon rainfall over land as estimated from various reanalyses and GPCP

Compiling the observations that are needed for a historical analysis still has a heavy reliance on humans. Most of the early observations were recorded using the hand in ship logbooks now residing in private collections, museums, and government archives, including others. Sometimes it is conducted by individuals reading meticulously through yellowed and cracked journals pages in a museum basement’s archives, searching for the weather observations not yet added to historical reanalyses. Those observations need to be made digital with the international, university, and the efforts of the national meteorological services. Many are typed or put into different databases all over the world. Observations must further be processed for purposes of removing unrealistic or duplicate values. The whole process of discovering a historical weather observation source to the point that they are ready for assimilation into reanalysis can take many years. When observation was relatively sparse in the early twentieth century, the observations coming from even a single weather station or the journey of a ship can cause some impact. As there is more rescuing of observations, new historical reanalyses could gain from observation networks that are denser than their predecessors.

Twentieth Century’s First Coupled Reanalysis

The broad set of historical observations continues to increase, thus greatly benefitting CERA-20C. That has registered notable achievement being the use of a model of coupled ocean-atmosphere with a new and robust algorithm of coupled data assimilation. That coupled system is the improvement over both ocean-only reanalysis and atmosphere-only reanalyses. The atmospheric reanalyses utilize temperatures of a prescribed sea surface and assimilate the ocean observations only. For instance, when observations have been absorbed into an ocean reanalysis, there is no permission for the atmosphere to adapt. Thus there are high chances of developing inconsistencies between the atmosphere and the ocean. Therefore, that can lead to spurious trends that are multidecadal in the fluxes of the air-sea. That is because of the inconsistencies with the observations of the ocean. The conditions of atmospheric boundary play some kind of tug-of-war on the state of the ocean. In converse, a system that is coupled can cause assimilation to both the oceanic and atmospheric observations at a go, allowing every state to adapt and react to the other, thus leading to a more significant consistency between the atmosphere and the ocean. CERA-20C posses no spurious trend in the fluxes of air-sea because the two states balance at the interface of the ocean/atmosphere as opposed to battling each other.  The removal of spurious trends does not imply the removal of the real signals of climate change. It is also noteworthy that the improved interactions of the air-sea lead to estimations that are more accurate. There are still few rough edges since that is the first reanalysis to utilize the assimilation of strongly coupled data. Laloyaux et al. (2018) point out the presence of a drift in the global content of ocean heat as the CERA-20C ocean component drifts away from the initialization towards the preferred state of the model. That provides an emphasis on the inconsistency in the models that are coupled and uncoupled. In as much as this could be as a result of model error, it provides direction to focus efforts for the aim of improving the models.

ERA-20C is another important advance of CERA-20C over the previous centennial reanalyses of ECMWF, and its important progress is the uncertainty treatment. Unfortunately, it is impossible to recreate past weather in a reanalysis, accurately. Observations could fail to be exact because of the errors in measurements. The models of weather forecasts have limitations as a result of inaccurate accurate physics representation and finite resolution. For example, there exists inherent uncertainty in the weather systems’ dynamics.

Another example is the historical Australian rainfall. If the time series of the regionally averaged precipitation over Australia from the analysis led to a yield of a slightly negative trend of a period, could that mean that the rainfall of the Australian is decreasing significantly? Without the presence of error bars, the significance of the direction would never be determined. Because the networks of observation are less accurate and sparse, to the level of 1900 than 2010, the error bars that existed in the earlier periods should be higher than the ones in the periods of recent times. For a better understanding of whether trends or the other signals carry significance, it is crucial to quantify the uncertainty of estimates from reanalysis.

CERA-20C provides an ensemble bearing ten different possible realizations for every variable in every time, unlike ERA-20C. Every member of the ensemble utilizes or uses a stochastic noise that is different, within the system of assimilation to represent physical and observational uncertainty. Thus, the ensemble could be used in quantifying risk through the examination of the spread or how the realizations are different from each other. For instance, the ten accomplishments will most probably be similar to each other in the year 2010. That is when the user has confidence in the ensemble average as an estimate that is accurate. The ten realizations could significantly differ in 1900, indicating less confidence. In converse, the single estimate that is from ERA-20C can never be used in the measurement of faith.

Laloyaux et al. (2018) give examples of how the ensemble spread is interpreted as confidence and how the CERA-20C uncertainty demonstrates changes in time. However, they discuss the shortcomings that accrue in providing quantification to difficulty with the use of 10 realizations. In the general sense, the larger ensembles come with more accuracy in the uncertainty estimates. The relatively small group of CERA-20C’s carries more confidence, leading to erroneous conclusions as regards to significance.

In addition to CERA-20C, the future version of 20CR that is generated by CIRES, NOAA, and the U.S Energy Department will add to the small historical sparse-input reanalyses’ historical collection. The following table shows the summary for the products of atmospheric reanalysis

Overview of Atmospheric Reanalysis products

References

Laloyaux, P., de Boisseson, E., Balmaseda, M., Bidlot, J. R., Broennimann, S., Buizza, R., … & Kosaka, Y. (2018). CERA‐20C: A coupled reanalysis of the Twentieth Century. Journal of Advances in Modeling Earth Systems, 10(5), 1172-1195.

Stone, D. A., Christidis, N., Folland, C., Perkins-Kirkpatrick, S., Perlwitz, J., Shiogama, H., … & Murray, D. (2019). Experiment design of the International CLIVAR C20C+ Detection and Attribution project. Weather and Climate Extremes, 24, 100206.