Pakistan Flood 2010 Case Study Geography In China

Floods and River Management

Rivers have always flooded and humans have always attempted to manage rivers. The ancient civilisations of Mesopatania and Egypt would build dykes and dams to hold back the waters following inundation. Indigenous skills, worldwide have created ingenious methods of interacting with rivers to increase food supply. Over time humans have modernised their engineering and in places have built vast structures to control and regulate the flow of rivers. We need not look any further than recent developments in China. The shear ambition of China's South-North Water Transfer Project dwarfs the remarkable achievement of the 3-Gorges Dam. However, major multi-purpose dam projects once common place in developed countries are now almost entirely part of the development strategies of developing countries and NICs, with India and China leading the way. In Europe and the US large scale river managment schemes have had mixed success and in many cases river management projects are reverting back to softer approaches in the realisation that the natural river system and catchment has much to offer.River management can be divided into hard and soft engineering. In both cases they attempt to reduce or regulate discharge in the river. Hard engineering is expensive, and tends to have a large impact on the river and the natural ecology and hydrology. Softer approaches tend to be more ecologically sensitive.

The Causes of Floods

Floods are a natural river process in response to changes in drainage basin inputs. They are an essential characteristic of the landscape and are fundamental to the development of floodplains, wetlands and many river features. Floods are therefore overwhelmingly caused by the physical environment. The interaction between atmospheric condition, drainage basin size, shape, geology and vegetation as well as the geometry of the channel varies over time and space. As a result, floods vary in magnitude and frequency. Human interaction, interference and management of the drainage basin and river channel have an influencing role as well. Large scale dams can control discharge and prevent floods. However, like in the failure of the Banqaio Resevoir Dam, in China,1975, where over 171000 people died and 11 million lost their homes, humans can be the direct cause of floods.With increased population and pressure on natural resources humans are having an increasingly important impact on the drainage basin. In many cases human impact is adding to or exacerbating both the frequency and magnitude of floods.

Physical Causes of Floods

The most important physical cause of floods concerns the interaction between precipitation and drainage basin response. The capacity of a river to cope with inputs becomes strained during both extended periods of rainfall (antecedent conditions) and short-term extreme rainfall events. e.g. Seathwaite, Cumbria, 2009, when 495mm of rain, the equivalent of several months of rainfall fell in just a 4 day period.In the case of the former, soils become saturated and the water table rises to the surface. As a result water cannot infiltrate and surface run-off occurs. For the latter, extreme rainfall leads to pooling on the surface and surface run-off. The nature of the drainage basin and its storage capacity is also and essential factor. Steep sided relief and/or impermeable rock and thin dense soils all accelerate surface run-off, which in turns leads to higher discharge and shorter lag times. Vegetation cover has an important role to play. Dense forest vegetation intercepts and transpires over 40% of precipitation inputs. Root networks further absorb water. The forest canopy intercepts rainfall slows inputs as throughfall. As a result surface run-off is minimalised and deep infiltration encouraged. Densely vegetated drainage basins therefore drastically reduce the magnitude and frequency of floods.


Human Causes of Floods

Human causes of flooding are a result of growing population pressure. Humans impact the interaction between precipitation and the drainage basin response through deforestation, as a result of agricultural development, floodplain drainage, urbanisation and channel management. Deforestation reduces the intercpetion and transpiration feedback resulting in increased quantities and rates of surface-run off. As a result more water reaches the river faster. In addition, deforestation exposes the soil to greater rates of erosion and nutrient leaching, which in turn increases the liklihood of further soil erosion and gullying. Soil erosion leads to sedimentation of the channel, which in turn reduces the capacity and hydraulic efficiency of the river, increasing the liklihood of floods. For example, deforestation in Nepal and Tibet is well known to be increasing the frequency  and magnitude of floods in Bangladesh. Floodplain drainage, especially in more developed countries has created space for modern agricultural systems and urban infrastructure. In doing so, the natural storage capacity of the floodplain and the wetlands they support has been lost. During low frequency high magnitude floods, the water simply has no where else to go. The impacts of floods are also exacerbated by the very fact that settlements have been built on the floodplain. Urbanisation, which leads to the expansion of built-up, impermeable surface, such a roads, parking lots and shops mauls further increases the rates of run-off. The very design of settlement infrastructure is to transfer water as quickly possible to the river. This is achieved through road camber, building design, drainage and sub-surface infrastructure. The importance of mobility and the car further expands the reach of impermeable surfaces through the continued loss of front garden in favours paved drives. Due to the growing number of 2/3 car families, an area of vegetated garden equivalent to 21 times the size of Hyde Park was lost in London alone between 1998 and 2006. Read this Guardian article onlost London gardens. In addition, rivers capacity is often reduced in local sections of the river, in urban areas. At bridging points and contained sections, bottlenecks form that without additional spillways can quickly become flooded during high flow.In less developed countries, population pressure leads to increases in agriculture and urbanisation, which further increases the rates of soil erosion and sedimentation.  In some cases, poor drainage can exacerbate flood events and in places where the river has been redirected and during extreme events the river simply takes its own route, regardless of what's its way. Finally, channel management of rivers has an impact on flooding. In most cases flood managment, such as dam construction and channelisation reduces the frequency of floods. However, with the main purpose of flood management aimed to increase capacity and move discharge as quickly and efficiently as possible past a settlement. The inevitable consequence is higher discharge and flood magnitude downstream. This is best illustrated in the Mississippi flood of 1993.


River Flood Case Studies- A fresh perspective

The Boscastle Flash Flood, 2004 - Click on the page to take you to outstanding PDF written by Phil Bull

The Pakistan Flood, 2010 - click on the page below to open the PDF

River Management
Hard Engineering

Multi-Purpose Dams
Multi-Purpose Dams are often built across the channel in order to store water and regulate the discharge of the river. Water is stored in a resevoir upstream of the dam and released in a controlled way. This protects the catchment downstream from potential flooding. Large dam projects are used for clean hydro-electric power, for large scale irrigation schemes and for strategies that open up the interior of countries for transport and trade. The downside of large-scale dams are the expensive costs; 3-Gorges was estimated to cost over $25 billion and despite its ability to produce 10% of the country's power requirements the environmental impacts on its ecology and river system are expected to be immense. Typical problems concern the disruption of migratory fish, sedimentaion and increased rates of erosion downstream. All this said, in the case of 3-Gorges, the number of people now free from the risk of flooding is the equivalent to the entire population of Belgium; 10 million. Multi-purpose dams therefore offer a country a means of regulating rivers at a vast scale and in doing so they protect the lives of many many people. Some disputed concerns with large scale-dams question their safety and sustainability, especially in regard to weight induced earthquakes and the acceleration of sedimentation due to landslides. Read this 3-Gorges Article from the BBC.

Channelisation

Channelisation is a diliberate attempt to alter the natural geometry of the river. Channelisation can be achieved in many ways. The river can be deepened and widened to increase the capacity of the channel. This increases its hydraulic efficiency and allows a larger discharge to be contained within the channel. This wil hlep prevent flooding. The channel can be made straighter, through the use of artificial cut-offs. The channel can be realligned to artificially increase the long profile gradient so that there is an increase in velocity and flood waters can be removed more quickly, which speeds up the flow and also aid navigation. Channelisation is often achieved through concrete lining the banks and bed. This prevents bed or bank erosion.

Alongside channelisation, engineering and drianage of the floodplian aids reclamation of the wetland. Through drainage and a contained river the water tablef alls. This makes farm land more manageable, creates space for urban developments such as housing and industry. At the local scale channelisation allows for bridges and other transport infrastructure to be built more easily.

In dense urban environments containment can be used. This is an extreme form of channelisation, which is bulit underground and contains rivers in sub-surface tunnels. They are extremely useful enabling dual use of urban environments, such ring roads, businesses and houses. They do have to be carefully managed to ensure blockages don't exacerbate flood events.In general channelisation is seen as an efffective means of increasing the capacity, hydraulic efficiency and discharge of rivers. It has been successfully used to prevent erosion and flooding for large river systems like the Mississippi, but there remains some significant questions in regard to cost, ecological impact and their effectiveness to cope with high magnitude low frequency flood events.


Dredging

Dredging the river is the process by which bedload is removed from the channel of the river. This is achieved through either heavy industrial pumps and diggers or through dislodging sediment that then encourages the natural flow of the river to transport it. The purpose of the river is to increase the cross-sectional area to reduce channel roughness and increase capacity and hydraulic efficiency. The benefit of using dredging is that it maintains the natural aesthetics of the river channel. However, it is costly and time consuming process that is only suitable for small section of the river, for example close to or within urban environments. In addition, the prosess has high ecological impact on natural ecology.


Soft Engineering

Bank Protection, groynes and gabions
Bank Protection
The purpose of bank protection is to help prevent bank erosion. This can be achieved through providing bank support. There are a number of different methods. Gabions, which are wire cages containing stones allow for a softer appearance. If used in the right way they can also encourage deposition close to the banks, which over time, will build up deposition and riparian vegetation. Groynes or spurs  achieve the same affect through deflecting the fastest flow away from the bank; deposition builds up on the headward side of the groyne. Over time, riparian vegetation will develop. An even softer approach is the planting of bank vegetation or by allowing vegetation to develop. This is achieved by simply not cutting it back. This will have different consequences depending on the aim. Tree planting along the banks provides greater stabilty and may reduce erosion. However. allowing vegetation to recover in a unregulated way may in fact encourage greater channel roughness and meander migration. When combined with the dismantling of floodpalin drainage the effect would be a rise in the watertable and an increase in the liklihood of floods. In contrast vegetation clearance reduces channel roughness, discourages bank deposition and increases the hydraulic efficiency of the river.


Floodplain Restoration

Increasingly drainage basin managers are realising the importance of the floodplain for its capacity to store water, reduce discharge and recycle harmful agricultural run-off. Floodplain restoration is the process of engineering the river to restore its natural patterns of meander migration and flooding. This can be either achieved over a vast area of river floodplain, like in the case of the River Kissimmee Restoration Project in Florida or localised to protect towns, such as in the case of the River Dijle, in Belgium. Floodplain restoration can be created in two main ways. Firstly, through either the removal of river regulation, as explained earlier through dismantling floodplain drainage and encouraging bank vegetation. Secondly, and more commonly it can be achieved through engineering. In this way the river may be held back at certain points by sluice gates and landscaping, which then create a washland upstream. Alternatively, river flow can be diverted into designated washland space alongside the river. This appraoch was first pioneered in Germany alng the Rhine, with mixed success. There are several important considerations. Washlands only replicate natural floodplains if their scale is large enough to encourage shallow flooding. In Germany, the first flooded washlands were to small and the water level to deep. As a result large animals like deer drowned. Shallow flooding helps encourage greater wetland biodiversity in terms of both flora and fauna, deep flooding detsroys it. A second consideration is the frequency of inundation. Natural floodplains in unregulated rivers flood reguarly and so the ecology is unique and diverse and adapted to that frequency. Artificial washlands are often used as a back-up for already existing river managemet plans. As a result the discharge in the river is seldom at a level that requires their use. As a consequence, very little wetland ecology and biodiversity struggles to establish itself.


Enhancing River Ecology

In other river management schemes the emphasis is not placed on flood management but on restoring and enhancing the natural river ecology. The aim of these projects is to creae features such as greater sinuosity and flow diversity, riparian pond habitats and emergent marginal habitats. The historical context of how British natural river systems were lost and the restoration project in the River Hampshire Avon, in Wiltshire is superbly explained in the following video:

Flood Management Case Study: Leuven and the River Dijle Valley

Other useful case studies...

Atmospheric Science

Were the 2010 Pakistan floods predictable?

Authors

  • P. J. Webster,

    1. School of Earth and Atmospheric Science, Georgia Institute of Technology, Atlanta, Georgia, USA
    Search for more papers by this author
  • V. E. Toma,

    1. School of Earth and Atmospheric Science, Georgia Institute of Technology, Atlanta, Georgia, USA
    Search for more papers by this author
  • H.-M. Kim

    1. School of Earth and Atmospheric Science, Georgia Institute of Technology, Atlanta, Georgia, USA
    Search for more papers by this author

Abstract

During July 2010, a series of monsoonal deluges over northern Pakistan resulted in catastrophic flooding, loss of life and property and an agricultural crisis that may last for years. Was the rainfall abnormal compared to previous years? Furthermore, could a high probability of flooding have been predicted? To address these questions, regional precipitation is analyzed using three dataset sets covering the 1981–2010 time period. It is concluded that the 2010 average May to August (MJJA) rainfall for year 2010 is somewhat greater in magnitude than previous years. However, the rainfall rate of the July deluges, especially in North Pakistan was exceptionally rare as deduced from limited data. The location of the deluges over the mountainous northern part of the country lead to the devastating floods. The European Centre for Medium Range Weather Forecasts (ECMWF) 15-day Ensemble Prediction System (EPS) is used to assess whether the rainfall over the flood affected region was predictable. A multi-year analysis shows that Pakistan rainfall is highly predictable out to 6–8 days including rainfall in the summer of 2010. We conclude that if these extended quantitative precipitation forecasts had been available in Pakistan, the high risk of flooding could have been foreseen. If these rainfall forecasts had been coupled to a hydrological model then the high risk of extensive and prolonged flooding could have anticipated and actions taken to mitigate their impact.

1. Introduction

Two main factors control South Asian rainfall. On 2–5 year time scales, the El Niño-Southern Oscillation (ENSO) phenomena is associated with above average summer precipitation during a La Niña and deficits during an El Niño [Shukla and Paolina, 1983; Kumar et al., 2006]. Far more dramatic and higher amplitude modulations occur on subseasonal time scales. Over much of Asia the summer monsoon is divided into a series of “active” (rainy) and “break” (dry) periods following a roughly 20–40 days cycle [Lawrence and Webster, 2001; Webster and Hoyos, 2004; Hoyos and Webster, 2007] associated with the boreal summer Madden-Julian Oscillation [Madden and Julian, 1972] that produce a northeasterly excursion of large-scale convective anomalies under the action of a strong cross-equatorial pressure gradient [Stephens et al., 2004; Wang et al., 2005, 2006]. The arrival of convection over the Indian subcontinent heralds an active pluvial period. Summer rainfall in Pakistan is also monsoonal and, as such, has active and break periods. However, the total summer rainfall is far less than in the east (Figure 1a) decreasing from the Bay of Bengal (16 mm/day) across the plains of northern India (8–10 mm/day) to values of about 6–8 mm/day in northern Pakistan. Pakistan is at the western edge of the pluvial region of the monsoon.

During the late boreal spring of 2010, the tropical Pacific Ocean entered a La Niña phase and during July 2010 the monsoon over the northern part of the Indian subcontinent was “active” with rainfall extending across the Gangetic Plains between the Bay of Bengal in the east to northern Pakistan in the west (Figure S1 of the auxiliary material). Embedded in this active period were the deluges that caused the devastating floods in Pakistan. In late July, some Pakistan stations recorded rainfall amounts exceeding 300 mm over a four-day period (http://www.pakmet.com.pk/FFD/index_files/rainfalljuly10.htm). During the following days and weeks, flooding extended through the entire Indus Valley eventually reaching the Arabian Sea leaving behind a wake of devastation and destruction. In the end, the death toll was close to 2000 and over 20 million people were affected. An estimated 20,000 cattle were drowned. Power stations and transmission towers were destroyed along with other major infrastructure such as barrages, bridges and roads. Irrigation systems were destroyed and planting of subsequent crops delayed or abandoned with agricultural costs exceeding $US500M. Overall, estimates of damage exceed $US40B. In general, it was the poor that suffered the most and many of these will face the prospect of intergenerational poverty as a result of the floods [Webster and Jian, 2011]. Most assessments of the 2010 Pakistan floods have appeared on the internet and in relief organization reports (http://www.pakistanfloods.pk/; http://en.wikipedia.org/wiki/2010_Pakistan_floods). Eventually, scholarly articles on the flooding will be forthcoming discussing, in more detail, the climate and meteorological conditions that produced the flooding (e.g., R. A. Houze Jr. et al., Anomalous atmospheric events leading to the summer 2010 floods in Pakistan, submitted to Bulletin of the American Meteorological Society, 2010). However, to date there has been an absence of any comment about the predictability of the deluges or the associated risk of floods. Eventually, skill in predicting floods reduces to the predictability of precipitation and the use of an adequately sophisticated hydrological model. Thus, an immediate and critical question is the degree to which rainfall at the western edge of the South Asian monsoon system is predictable on time scales of 1–2 weeks. Is the predictability of precipitation in the western edge of the monsoon comparable to that seen over the Ganges and Brahmaputra basins [Hopson and Webster, 2010; Webster et al., 2010]?

In this study we focus on the predictability of 1–15-day ECMWF EPS forecasts [Buizza et al., 2007] over Pakistan. In the next section details of the observation and numerical model data are introduced. Section 3 discusses the uniqueness of the July-August flooding events and examines the prediction skill of 15-days rainfall forecast followed by conclusions related to the predictability of floods in Pakistan.

2. Data and Analysis

Three precipitation data sets are used to assess the variability of the precipitation over the Pakistan region: the Global Precipitation Climatology Project (GPCP) data [Adler et al., 2003] for the 1981–2009 period, the Tropical Rainfall Measuring Mission (TRMM) [Huffman et al., 2005, 2007] TRMM_3B42 product for 1998–2010, and the NOAA CPC Morphing Technique (CMORPH) Precipitation Product for the 2003–2010 period [Joyce et al., 2004]. GPCP (a merging of rain gauge data with satellite geostationary and low-orbit infrared and passive microwave information) and TRMM data sets (specifically the TRMM_3B42 set) were chosen for their temporal extension (29 and 13 years, respectively). All of these precipitation products had a 0.25° × 0.25° horizontal resolution facilitating a comparison with model output. Figure S4 shows time series of monthly rainfall anomalies for each of the data sets.

A comparison of the CMORPH and TRMM data sets (Figures S3) reveals considerable differences in the magnitude of estimates of precipitation during the third precipitation pulse of July 2010 that occurred over the higher terrain of northern Pakistan (Figure S2d). The TRMM rainfall estimate was considerably higher than CMORPH by about a factor of two consistent with the discussion of Gopalan et al. [2010] who suggested that TRMM may overestimate precipitation rates over substantial terrain. Comparisons during earlier periods, when the precipitation maxima occurred over the plains of southern Pakistan and northwestern India are more comparable (Figure S2). Consequently, we use CMORPH as the principal data set for determining the sequence of events during 2010 and also as the principal agent for the statistical rendering of the quantified precipitation forecasts.

The ECMWF EPS forecasts consist of 51 ensemble members initialized twice per day (00 and 12 UTC), each ensemble member having a 15-day forecast horizon. The horizontal resolution of the model is 50 × 50 km from 0 to 10 days and then 80 km × 80 km from day 10–15 [Buizza et al., 2007]. For this initial study, model forecast precipitation for the months of July and August from 2007 to 2010 was converted into daily cumulative amounts. To minimize systematic model bias differences between the distributions of the ECMWF forecasts and the observed rainfall, a quantile-to-quantile (q-to-q) mapping technique was implemented following Hopson and Webster [2010] and Webster et al. [2010] (see method description in the auxiliary material). All rainfall forecasts presented here are adjusted using the q-to-q technique.

3. Results

Beginning in early July 2010, there were six major pulses of torrential rainfall occurring over Pakistan, each separated by about a week (Figure 1b). One of the most intense periods occurred between July 27–30 over the mountainous regions of the north. Figure S2 shows the distribution of rainfall for the major pulses of monsoon rain. The earlier rainfall events caused flooding in Balochistan in central Pakistan. Flooding followed across northern Pakistan in the Khyber Pakhtunkhwa province with the later July rains extending to the Punjab in late July/early August (http://www.unitar.org/unosat/node/44/1469). Here we address the uniqueness and predictability of the floods.

3.1. Uniqueness

There have been 67 reported flooding events in Pakistan occurring since 1900 with a clustering of 52 events of various severity in the last 30–40 years (International Disaster Data Base, http://www.emdat.be). Some of these events (e.g., 1950, 1973, 1976, 1977, 1992, 2001, 2007 and 2008) were also accompanied with large loss of life and property. This recent increase is consistent with the increase in intensity of the global monsoon accompanying the last three decades of general global warming (B. Wang et al., Recent intensification of global monsoon and precipitation, submitted to Nature, 2011) or perhaps changes in water management strategies, increases of damage due to a rapidly growing population or improved reporting through advances in communication.

Figure S4 shows the temporal variability of seasonal (MJJA) precipitation averaged in Pakistan (62°–74°E, 24°–36°N, blue rectangle in Figure 1a) and northern Pakistan (70°E–74°E, 30°N–36°N, red rectangle in Figure 1a) relative to the seasonal climatology for each of the data sets: GPCP and CMORPH. While there are amplitude differences between datasets, each shows substantial variability, with seasons of excessive rainfall and drought occurring irregularly over the past 30 years (Figures S3 and S4).

Were the rainfall events of 2010 worse than previous extreme events? Using a 13-year TRMM precipitation record, extreme events can be counted. An extreme event is defined here to occur when the two-days accumulated rainfall exceeds over 10 mm over all Pakistan and 20 mm over the northern Pakistan (Figures 1c and 1d). Note that the chosen thresholds for this analysis are much smaller than maximum daily rainfall measurements at specific stations (see http://www.pakmet.com.pk/FFD/index_files/rainfalljuly10.htm) due to a broader averaging area. Although there is considerable interannual variability, the number of extreme events over entire Pakistan, so defined, is larger in 2010 than in previous years, greater, for example than in 2008. In summary, 2010 stands out as a period of above average rainfall events over northern Pakistan. The number of extreme events over northern Pakistan is far more unique which, based on the very limited TRMM data set would have return periods of >30 years. Long-term variability for extreme events is calculated with GPCP pentad data set from 1981 (Figure S5) to 2007 overlapped with CMORPH pentad from 2003 to 2010. Although, there are differences between data sets, the high occurrence of Northern Pakistan extreme events in 2010 is relatively rare. Rainfall data is not sufficiently reliable prior to 1987 when GPCP data was generated on a daily basis. However, we do have CMORPH and TRMM data for 2008. As shown in Figure S6, the cumulative July-August rainfall for northern Pakistan is larger in 2010 than 2008, with values larger than 0.5 m in several areas.

3.2. Predictability

The next step is to examine the predictability of the rainfall events depicted in Figure 1b. Figure 2a shows the total average precipitation [mm/day] for July 28–29, based on the CMORPH observational dataset and the ECMWF forecast ensemble mean initialized 4 days before the event (Figure 2b). The q-to-q correction was applied to the precipitation forecast data. The forecasts compare well with the observed rainfall with ECMWF slightly underestimating the rainfall intensity in the northern part of the region. The ECMWF forecast showed average precipitation larger than 40 mm/day in some areas which is over 3 times larger than the CMORPH climatological average for the region.

Figures 2c and 2d show the temporal evolution of the ECMWF forecast commencing on 22nd and 24th July, 2010 through August 9, 2010 for the Khyber-Pakhtunkhwa province, located in the north west of the country (red rectangle in Figure 1a). The diagram shows the probability distribution of precipitation based on the 51 ensemble members with the ensemble mean plotted as the black dotted line. The blue line represents the CMORPH observed rainfall. Good predictive skill of the July 28–29 event is found up to 6 days in advance. The same analysis done for various other monsoon pulses have resulted in similar conclusions (Figure 3).

Figure S7 shows an assessment of precipitation predictability in northern Pakistan using all available hindcast data. Predictability is shown as correlations between predicted and observed CMORPH rainfall values as a function of lead time for July based on 2007–2010 period. Note that for 2007, the model prediction extends only up to 10 days but up to 15 days for the 2008–2010 period. Correlations ≥0.7 were found for predictions 5 days in advance indicating useful predictive skill. Thus, the quantitative rainfall forecasts could be used as a robust variable in a flood forecasting scheme for Pakistan region.

In order evaluate whether the model can provide useful information with regards to the actual severity of the major rainfall events of July-early August 2010, all ECMWF forecasts made during the period were extracted and bias corrected. Then, the probability that the predicted rainfall would exceed the observed climatological average plus 1 standard deviation was computed. In other words, for each forecast, at each lead time, the percentage of ensemble members exceeding the threshold was computed. The exceedance threshold is calculated using 2003–2010 CMORPH data, with mean and standard deviation based on July-August daily average data. Results are shown in Figure 3 as shaded contours. The blue line represents the observed CMORPH rainfall averaged for the same region and the same time period. For example, the July 28 event was predicted almost 8 days in advance with a probability >60% over the climatological average plus 1 standard deviation (Figure 3). All the other events appear to have similar skill at the 8 to 10 day horizon.

4. Conclusions

From a climatological perspective, July and August precipitation rates were above average in Pakistan although not exceptionally so. However, in terms of rainfall rate, the monsoon pulses were extreme events compared to other years in the period 1998–2010. The devastating flooding occurred from a conspiracy of events. The summer of 2009 was a severe drought period with rainfall well below average (http://www.pakmet.com.pk/monsoon2009ver.pdf) so that vegetation may have been sparser during 2010. The region is mountainous with steep valleys and ridges. Furthermore, deforestation in northern Pakistan has been severe [e.g., Ali et al., 2006]. Deforestation and sparse undergrowth would exacerbate runoff through the steep valleys of the heavy rains that occurred during the month of July and early August.

The major result of the study is that the heavy rainfall pulses throughout July and early August were predictable with a high probability 6–8 days in advance. If these forecasts had been available to the regions of northern Pakistan, government institutions and water resource managers could have anticipated rapid filling of dams, releasing water ahead of the deluges. A high probability of flooding could have been anticipated.

Finally, it appears that Pakistan would benefit from a hydrological forecasting scheme similar to that developed for Bangladesh [Hopson and Webster, 2010; Webster et al., 2010]. The Bangladesh system incorporates the same form of statistically rendered ensemble precipitation forecasts as discussed above but coupled to a hybrid hydrological model. Working with Government of Bangladesh authorities, these 10-day river forecasts were communicated to the union (county) and village level allowing time to prepare for the floods for three major Brahmaputra floods during 2007/8 allowing the saving of household and agricultural effects and the successful evacuation of those in peril [Webster et al., 2010; Webster and Jian, 2010] (ADPC, Flood forecasts application for disaster preparedness: Post flood forecasts assessment 2008: Community response to CFAN forecasts, 2009, available at http://www.adpc.net/v2007/ and http://pacific.eas.gatech.edu/∼pjw/FLOODS).

Acknowledgments

This research has been supported by the Climate Dynamics Division of the National Sciences Foundation under Award NSF-ATM 0965610. Once again, we are indebted to ECMWF for providing data to make this analysis possible. We wish to thank J. A. Curry for interesting discussions as summarized at http://judithcurry.com/2010/09/20/pakistan-on-my-mind/and comments on the paper.

Noah Diffenbaugh thanks the two anonymous reviewers.

Ancillary

Article Information

Format Available

Full text: HTML | PDF

Copyright 2011 by the American Geophysical Union.

Request Permissions

Publication History

  • Issue online:
  • Version of record online:
  • Manuscript Accepted:
  • Manuscript Revised:
  • Manuscript Received:

Keywords

  • flood forecast;
  • 2010 Pakistan flooding

Supporting Information

Auxiliary material for this article contains a text file and seven figures.

Auxiliary material files may require downloading to a local drive depending on platform, browser, configuration, and size. To open auxiliary materials in a browser, click on the label. To download, Right-click and select “Save Target As…” (PC) or CTRL-click and select “Download Link to Disk” (Mac).

Additional file information is provided in the readme.txt.

FilenameDescription
grl27810-sup-0001-readme.txtplain text document, 3KReadme.txt
grl27810-sup-0002-txts01.pdfPDF document, 588KText S1. Method description.
grl27810-sup-0003-fs01.epsPS document, 804KFigure S1. Latitude-time longitude-time cross section for observed outgoing longwave radiation during June-July 2010.
grl27810-sup-0004-fs02.epsPS document, 2276KFigure S2. Observed CMORPH precipitation for the 6 monsoon pulses during July-August 2010.
grl27810-sup-0005-fs03.epsPS document, 91KFigure S3. Scatter diagram of monthly precipitation anomaly between CMORPH and TRMM over the period from 2003 to 2010.
grl27810-sup-0006-fs04.epsPS document, 217KFigure S4. Seasonal mean precipitation for GPCP and CMORPH averaged over Pakistan and northern Pakistan.
grl27810-sup-0007-fs05.epsPS document, 187KFigure S5. Number of heavy rainfall events over the summer in GPCP and CMORPH.
grl27810-sup-0008-fs06.epsPS document, 637KFigure S6. Observed July-August cumulative CMORPH precipitation for years 2008 and 2010.
grl27810-sup-0009-fs07.epsPS document, 63KFigure S7. Overall estimates of the predictability of precipitation in the Pakistan region versus lead-time for July based on 15-day forecasts from 2007–2010.

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

References

  • Adler, R. F., et al. (2003), The version 2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present), J. Hydrometeorol., 4, 1147–1167

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *