«DIPLOMA THESIS Linking Climate Change with Food Security in the highlands of Khyber Pakhtunkhwa, Northwest Pakistan Presented by: Martin Kienzler ...»
This implicates that the two datasets suit each other quite good in some points but rather show contradictory results in some respects. Hence it has to be evaluated which data should be given more attention in which aspect. The most drastic diﬀerences between the results of the two datasets occur when regarding precipitation trends. While trends on an annual scale match quite good for Khyber Pakhtunkhwa, it can be claimed that they even state oppositional trends for the respective seasons. For example CRU shows negative trends for the summer monsoon while REMO simulates increasing precipitation for most parts of the country in summer. In winter it is the other way round.
Furthermore for the summer months REMO simulates trend rates which in some areas are several hundred percent of the annual mean precipitation, which is not realistic. One major problem of the REMO model data is that somehow one or two months of extreme precipitation apparently are simulated for each grid of the study area between 1971 and
2000. This means that precipitation trends are strongly overestimated for several grid boxes during these months. Besides statistical signiﬁcance, which is not illustrated, is not given for most of the REMO grid boxes when regarding precipitation trends.
Paeth & Hense (2002) argument that for time periods of 30 years or less radiative forcing only plays a minor role in climate variability, especially concerning precipitation.
So variabilities are rather due to changes in less predictable forces such as atmospheric circulation and radiative forcing becomes important on longer periods.
Another problem is, that this study only provides REMO trend analysis for one special period, namely 1971 to 2000. The solution could be to calculate trends for every 30-year period within the available time span of the data, to validate the respective trends and to eliminate the probably misleadingly simulated extreme values. This was done for eleven 30-year periods between 1961 and 2000 for both climate variables (see Appendix B). The last of the periods in each case is the one used for the analysis before. The same was done for the CRU data.
Regarding temperature, it becomes obvious that the rate of warming or cooling trends respectively diﬀered throughout the last forty years of the twentieth century. For example within the region of the monsoon belt the warming trend was strongest between mid sixties and mid nineties. After this warming became less and even turned into a rather cooling trend in some regions. In terms of precipitation the rather increasing trend
Chapter 7. Discussion
described before only began in the periods extending the middle of the nineties. Before the country was rather marked by a strong decreasing precipitation trend, especially the monsoon belt. The positive development became stronger towards the end of the twentieth century.
The comparison with CRU shows that both datasets agree when postulating the strongest warming trend for the monsoon belt. In the CRU data a north-south spreading belt of a negative temperature trend in central Pakistan is visible, which becomes stronger towards the end of the century but is not reproduced by the REMO simulations.
In general both datasets state that the strongest warming trend occurs until around 1995 and becomes less in the following years. For precipitation CRU shows distinct increasing precipitation for most of the country and most of the 30year periods. Only for the regions north of 35◦ latitude and the last two periods a reduction of precipitation amounts is recorded. This is in contrast to the REMO model output. The trend patterns of the CRU data with strongly opposing trends on a small scale point out that values are interpolated from single stations and hence may be calculated improperly for some regions.
Furthermore CRU postulates far higher trend rates than REMO which seems not to be very realistic.
A problem of the REMO simulations is that the “only real boundary conditions” (Paxian et al., 2011) of the model are the observed atmospheric concentrations of carbon dioxide, which are based on a starting value of 1860. This means that the temporal distribution of the trends must not be correct. According to them the model “is not able to reproduce the observed strong trends” in their analysed region. Therefore they conclude that “radiative forcing cannot be the only driver of temperature and precipitation variability” but make the variability of circulation patterns partly responsible for the trends. The same may be true for the region analysed in this study. This does not necessarily show a “deﬁcient model performance” (Paxian et al., 2011) but points out that other real boundary eﬀects such as aerosols or land cover changes have to be taken into account when predicting variabilities of the climate parameters.
Anyhow the analysis of diﬀerent time periods illustrates the high interdecadal variability of temperature as well as precipitation over the region.
Validation with PMD station data Figure 7.1 illustrates the temperature development at the meteorological station Chitral at an altitude of around 1,500 masl. The data is provided by the Climate Data Processing Centre of the Pakistan Meteorological Department (PMD, 2011). The course of mean, minimum and maximum temperature is shown for the whole year, for the summer months and for winter. It becomes obvious that there is only a slight temperature increase on an annual scale between 1965 and 2009. Minimum temperatures rather tend to decrease. This cooling is somehow compensated by a strong incline of the maximum temperatures. In summer, temperatures generally are dropping down, especially the minimum temperatures. On the other hand there is a strong warming during winter months. Winter temperatures show a distinctly higher variability than temperatures in summer and throughout the year. A warm period can be detected between 1998 and
2005. These ﬁndings are clearly ratiﬁed by both datasets, CRU and REMO (see ﬁgures
6.7 and 6.9). Furthermore both datasets claim the strongest warming to occur around the ﬁrst half of the nineties, which is similar to the observed data of the climate station Chitral.
Figure 7.2 – Precipitation development of Balakot (left, 1971-2009) and Chitral (right, 1965-2009) in mm.
The black line represents annual mean precipitation, the red line summer precipitation (JJA) and the blue line winter precipitation (DJF). Data obtained from PMD (2011)
For comparison of the precipitation with the observed data from the Pakistan Meteorological Department the station of Balakot was used besides the one of Chitral (see ﬁgure 7.2), because it is the nearest existing climate station in the surroundings of the case study Kaghan. According to the course of precipitation rainfall amounts seem to have been reduced all over the year in Balakot, especially in summer. These ﬁndings are approved by both datasets, only for winter CRU claims precipitation to increase instead of decrease. In contrast, precipitation tends to increase in Chitral all over the year, at the highest rate in winter. These positive trends are conﬁrmed by both datasets again.
REMO somehow only suggests a very slight increase, even in winter. The strong incline during the winter months is better reproduced by the CRU data.
Comparison of the climate data results and the interviews The interviews conducted within the two case studies only represent a glimpse inside the overall situation. One major problem turned out to be the awareness of the people concerning the topic of climate change. The perception of climatic changes of most farmers at ﬁrst concentrated on the most recent two or three years. Only after asking questions directly pointing to longer periods they tried to remember. The main criticized issue of this method of qualitative research therefore is that interviews often could be associated with subjectivity and lead to misunderstandings between researcher and informants (Philip, 1998). Of course the language barrier was quite challenging.
Though the results received through questioning overall are in good accordance with the climate data. For example in Kaghan valley farmers reported an overall warming trend, both in summer and in winter. Both climate datasets agree with a strong warming trend until recently, even if summer temperatures tend to decrease in the last few years, which is not conﬁrmed by the informants. Both, the questioned farmers and the results of all climate datasets result in a decreasing tendency of summer precipitation in Kaghan valley. The accordance of the statements of the informants in Chitral with the climate data for this regions is even better. The distinct decline of summer minimum (night) temperatures is conﬁrmed by almost all of the farmers. The same is true for strongly increasing winter temperatures. Only summer maximum temperatures are mostly reported to become higher while climate data claims them to decrease, too, but only slightly. Summer rainfall is reported to be reduced, while for winter precipitation statements diﬀer widely. According to the climate data, precipitation should increase throughout the year. For the period 2000 to 2005 the informants reported the complete absence of snow in the valleys. According to the PMD data (see ﬁgures 7.1 and 7.2) this is the period with the warmest winter temperatures during the last 45 years and relatively few winter precipitation. This example points out the good agreement between the recorded climate data and the perception of the local people.
Accordance with diﬀerent studies While considering the value of the statements made in this study, it is best to compare them with ﬁndings of diﬀerent studies, which already were made in this region. For ex
ample Chaudhry et al. (2009) postulated an increase of annual mean and maximum temperatures and a decrease of annual minimum temperatures for Khyber Pakhtunkhwa.
Furthermore summer temperatures show a rather negative tendency while winter temperatures are supposed to go up in all of Northern Pakistan. The same was found out in this study. This opposing trend of summer and winter temperatures are moreover conﬁrmed by Fowler & Archer (2006), Sheikh et al. (2010) and Yadav et al. (2004). The latter constituted a strong cooling trend of the premonsoon (April to June) minimum temperatures, the main reasons for this being deforestation and soil degradation. The decreasing trend of night temperatures compensates the warming trend of the maximum temperatures. The distinct cooling tendency of the spring months is clearly reproduced by this study. Sheikh et al. (2010) even found a negative temperature trend for the most western part of Khyber Pakhtunkhwa for the whole year. In contrast Bhutiyani et al. (2007) found strong increasing temperature trends throughout the year for the western part of India’s Himalaya, which are remarkably higher than the global average.
Until the end of the 21st century REMO simulates an area-wide warming of 5 to 6◦ C along the mountainous regions of northern Pakistan. Sheikh et al. (2010) predict a warming trend of the same extent with the help of the regional climate model PRECIS and a little lesser warming for the foothills of the Himalaya.
Concerning precipitation in northwestern Pakistan Chaudhry et al. (2009) claimed it to be reduced during summer, especially during the second half of the twentieth century. Winter precipitation shows an increasing tendency all over the century. Precipitation generally undergoes a high interdecadal variability and was highest during the 1950s. These ﬁndings can be supported by this study. Similarly Archer & Fowler (2004) analysed several climate stations throughout northern Pakistan and detected an increasing precipitation trend in winter, especially for the stations of northwestern Pakistan. They related this change to the variability of the North Atlantic Oscillation, which is responsible for the moisture content of the western disturbances.
Monsoon dynamics Concerning the development of the summer precipitation within the monsoon inﬂuenced areas of Pakistan the diﬀerent datasets show some discrepancies. But the overall tendency is a rather negative trend of summer rainfall during the last thirty years. This makes it necessary to analyse what other studies found out. A distinct weakening of summer monsoon rainfall and a delay of monsoon onset dates during the second half of the twentieth century was detected by Ashfaq et al. (2009), using the regional climate model RegCM3. Reduction rates were supposed to be strongest since the early eighties.
The date of the monsoon onset is deﬁned as the ﬁrst pentad (5 day - mean) of the year, during which mean precipitation exceeds 5 mm per day (Wang & LinHo, 2002). For northwestern Pakistan monsoon onset usually occurs at the 39th pentad of each year which approximates the 15th of July. According to them, the reason for these ﬁndings is an enhanced greenhouse forcing. Bhutiyani et al. (2010), who analysed diﬀerent climate stations of northwestern mountainous India showed distinctly decreasing summer precipitation amounts during the last century. In contrast Cherchi et al. (2011)
Chapter 7. Discussion
however showed that increased atmospheric concentrations of carbon dioxide would indirectly lead to a slight increase of monsoonal precipitation over South Asia through a considerable rise of the moisture content, despite a possible weakening of the monsoon circulation. This implicates that greenhouse forcing cannot be the only factor aﬀecting South Asian monsoon precipitation. Instead diﬀerent human-induced forcings such as land-use changes through intensiﬁcation of agriculture (Niyogi et al., 2010) and irrigation (Saeed et al., 2009) have to be taken into account, both of which seem to cause a weakening of the summer monsoon. Ramanathan et al. (2005) relate the distinct reduction of summer rainfall since the 1960s to a “deceleration of the summer monsoonal circulation”, an increase of atmospheric stability and a decrease in evaporation. The reason for this seems to be the dimming eﬀect of aerosols such as black carbon. A distinct increase of drought events was found for the 20th century, too.