The overall human impact, damage and spatial sectoral data were collected for the 2010 and 2022 floods. This data were compiled from official Post-Disaster Needs Assessments (PDNA), federal and provincial government records and non-governmental reports and data repositories. The details on data sources and data collection methodology for the PDNA process can be viewed in Supplementary Appendix 2. This process collected statistics for overall impacts, including populations and areas affected by floods, major economic indicators, and sectoral damage across various domains, such as housing, health, education, water supply and sanitation, agriculture, irrigation and flood management, roads and bridges, industry and commerce, governance infrastructure, and other infrastructure at the national and provincial levels. The data encompassed the primary monsoon flooding periods from mid-June to mid-September in 2010 and 2022, corresponding to each disaster. For this study, monetary values are mainly expressed in Pakistani Rupees (PKR) but also in US Dollars (US $) whenever required by the methodology for comparison reported for each respective event year. The prevalent exchange rates for US $ in the year were used, mentioned in the PDNA reports.

The data were meticulously organized and harmonized by ensuring uniform sector definitions from 2010 to 2022, where sub-sectoral items were aggregated to the nearest consistent sectoral category. The administrative units were aligned, such as adjusting 2010 data to match the district boundaries of 2022, where necessary. The details are given in Supplementary Appendix 1. The assembled data were then utilized as inputs for general comparison and for the two analytical frameworks outlined below. Figure 3 presents a flow chart of the study design, depicting the data collection and analysis process.

Utilizing secondary data from post-disaster reports presents certain limitations. Firstly, the damage figures from PDNAs are derived from rapid assessments and may lack field validation or ground truthing, potentially impacting their accuracy. Data collection methods and definitions might have varied between 2010 and 2022, leading to standardization challenges. This is mitigated by aligning sector categories and, where possible, cross-referencing figures with multiple sources (e.g., comparing NDMA and PDMA reports) to ensure consistency. No primary field surveys were conducted; instead, it relies on the best available official data, which, while comprehensive, may contain inconsistencies or reporting biases. Secondly, monetary damage estimates are nominal and were not uniformly adjusted for inflation or changes in asset values over the 12-year span. This lack of normalization can distort direct financial comparisons across years. For instance, the value of infrastructure in 2022 was higher due to inflation and development, so a rupee of damage in 2010 represents a larger real loss than a rupee in 2022. In this study, monetary figures are reported in nominal terms as documented in official assessments; however, predominantly cross-year analytical comparisons rely on normalized patterns and human-centric metrics rather than absolute monetary values, to minimize inflation-related bias. In this way, the limitation is addressed. Thirdly, differences in the number of districts and the spatial extent of flooding between the two events introduce a comparison bias. The 2010 flood affected a province of 24 districts, reported to be ten by PDMA in 2012, whereas KP had 35 districts by 2022, of which only 17 were declared calamity-hit. The analysis is restricted to the 11 undivided districts present in 2010, common to both events, to ensure a like-for-like comparison and discuss the implications of the differing flood footprints in our analysis. Lastly, intangible and long-term impacts (e.g., mental health issues, environmental degradation) are not captured in these datasets. The lifeyears framework partially addresses human impacts, but intangible losses remain outside the quantitative scope and are acknowledged qualitatively. Being transparent about these limitations enhances the credibility of the findings and clarifies that results should be interpreted with these considerations in mind. These limitations also highlight areas for future research, such as incorporating primary data collection and broadening impact metrics.

To conduct a general comparison of the impacts in terms of monetary losses and quantity, this study employed JI and the Noy-lifeyears framework. These frameworks are elaborated in detail in the following subsections.

In nominal terms, the total financial impact of the 2022 floods amounted to Rs. 201.00 billion, representing a 100% increase compared to the 2010 floods, which incurred damages of Rs. 99.63 billion. In 2010, the floods affected the national GDP by 5.8%, whereas in 2022, the impact was 4.8%. This suggests that the floods in 2010 were more damaging in nature.

At the national level, the GDP growth rate remained constant in the year following the 2010 floods, showing no change in growth rate. However, after the 2022 floods, the GDP growth rate from +6.2% became −0.2% in the subsequent year, indicating a decline in the growth rate of −6.4%. This also led to a significant increase in the consumer price index (CPI) inflation. The inflation rate in the subsequent year surged by +11.9% after the 2022 floods, in contrast to a decrease of −2.7% after the 2010 floods. This indicates that the 2022 floods impacted an already vulnerable economy characterized by high inflation, thereby exacerbating macroeconomic pressures. An increased proportion of the population was pushed below the poverty line in both floods. It was estimated that in 2022, 5.9% more people were pushed below the poverty line compared to 2.2% in 2010 (MoPDSI, 2010; MoPDSI, 2022). In the case of KP, the 2022 flood was anticipated to increase multidimensional poverty by up to 13 percentage points in the affected areas, suggesting that a significant portion of the rise in poverty can be directly attributed to the impacts of the disaster. The more pronounced impact of poverty in 2022 corresponds with the floods’ extensive geographic reach and the pre-existing economic vulnerabilities, such as high inflation, prevalent during that period.

The 2022 floods affected 4.35 million individuals, in contrast to the 3.8 million affected in 2010, indicating a 14% increase in absolute number. The proportion of the displaced population relative to the total affected population decreased by 26% in 2022. The inundated area decreased from 5034.76 sq. km in 2010 to 3982.46 sq. km in 2022, reflecting a 21% reduction, with only 31% of houses damaged in 2022 when compared to the number of houses damaged in 2010 (91,767/2022 → 295,684/2010). Although the effect on the population in 2022 floods is more than that in floods 2010, the reduction in displacement can be attributed to lesser inundations, indicating that many affected individuals did not relocate. The total cost of damage to the housing sector was 28% higher in 2022, although comparatively lesser numbers were damaged, indicating an escalating disaster cost in this sector. This observation aligns with the global trend of increasing costs per unit of damage.

The fatalities and injuries were less severe in the 2022 floods, with 419 injuries compared to 1,196, and 1,156 fatalities compared to 1,739 in 2010. The 2022 floods had a substantial financial impact on the KP; however, the severity was comparatively less than that in other provinces of the country. In 2022, other provinces like Sindh and Balochistan received more damage. The damage cost for KP constituted only 4.70% of the national damage cost in 2022, whereas it was 12% of the national cost in 2010. Comparison of major damages in terms of proportional percentage is provided in Table 1 and Figure 4.

The province-wide nominal sectoral damage costs of the 2022 floods were compared with those of the 2010 floods, and comparative percentages were computed. This comparison reveals that not all sectors exhibit a comprehensive similar improvement. This is despite the lessons learned from the catastrophic 2010 floods and the implementation of a proactive regime during that year. Sectors, such as irrigation and flood management, roads and bridges, agriculture and livestock, the private sector and industries, governance, and environmental sectors, demonstrated lower damage costs and thus improved resilience. The most improved sectors were agriculture and livestock (−82%), roads and bridges (−69%), and irrigation and flood management (−57%). Other critical sectors, such as Housing, health, education, and water supply and sanitation, exhibited increased damage costs in 2022. Water supply and sanitation systems experienced significant breakdowns during the 2022 floods (Islamic ReliefP, 2024), and the most significant increase in damage costs was recorded in this sector, with a rise of 143%. Floods in 2022 caused massive disruption to the education sector (Sujaya et al., 2023), and it experienced a 71% increase in 2022 compared to the 2010 flood period. The health sector, too, exhibited a concerning trend. It rose from 43% to 56% in 2022 floods. Overall, the sectors that demonstrated poorer outcomes in 2022, namely, housing, health, education, and water supply and sanitation, are those intrinsically linked to social wellbeing. These outcomes highlight deficiencies in the integration of hazard-resistant standards within all social infrastructure. This further suggests that while there may have been slight improvements in the resilience of physical infrastructure, societal exposure and vulnerability remain pronounced in essential human services. These findings indicate an uneven implementation of “proactive” disaster risk reduction across sectors.

This is also indicative of the provincial government’s initiation of extensive restoration and strengthening of roads, bridges, and flood defenses and some other sectors. These investments likely paid dividends by 2022. Monetary comparison of sectoral damage costs is provided in Table 2 and Figure 5.

When sectoral damage proportions are compared within each flood year and across corresponding sectors, the results indicate that sectors such as irrigation and flood management, roads and bridges, agriculture and livestock, private sector and industry, governance, and the environment accounted for a larger share of total damages during the 2010 floods. In contrast, these same sectors contributed a comparatively smaller proportion of overall damage during the 2022 flood event. Conversely, the housing, health, education, water supply, and sanitation sectors accounted for more damage during the 2022 flood (Figure 6).

Overall, the results indicate that while the 2022 floods in Khyber Pakhtunkhwa were financially larger in nominal terms, they were structurally less severe yet socially more consequential than the 2010 event. Reduced inundation extent, lower fatalities, injuries, and displacement, and a smaller provincial share of national damages point to improvements in preparedness and resilience of certain physical infrastructure sectors, particularly irrigation, roads, bridges, and agriculture. However, the increased damage burden in housing, health, education, and water and sanitation sectors, alongside heightened macroeconomic stress, inflation, and poverty impacts, highlights a growing vulnerability in social infrastructure and essential services. This divergence suggests that post-2010 risk reduction efforts have yielded uneven outcomes: gains in structural protection have not been matched by commensurate improvements in people-centered resilience. Consequently, the 2022 floods underscore the need to rebalance disaster risk management in KP toward safeguarding social services and livelihoods, rather than relying predominantly on reductions in physical damage as a proxy for resilience. It also indicates that the transition to a “proactive” disaster management regime post-2010 has not consistently led to a reduction in vulnerability at the grassroots level.

In 2010, subsequent to the recent implementation of the disaster management framework and the onset of flooding, the KP government declared all 24 districts as calamity affected. However, the actual number of impacted districts was fewer, primarily those situated along riverbanks, due to excessive rainfall that resulted in riverine flooding (Rahman & Khan, 2013). This fact was also substantiated when damaged districts were noted as 10 districts in the monsoon contingency report published in 2012 by the PDMA (PDMA, 2012), retrospectively citing lessons learned and as a reference for vulnerable regions for upcoming monsoon. In contrast, during the 2022 floods, the KP government declared 17 out of 35 districts, representing 50% of the province, as calamity-affected, with the damage being more widespread across the province (Mahmood, 2024). The declaration was carried out by PDMA which has well established reporting and coordination mechanisms and has matured during the past 12 years. This spatial redistribution suggests a shift in flood manifestation from large-scale riverine flooding toward a combination of localized riverine, pluvial, and flash-flood processes.

The predominant event in 2010 was a four-day extreme rainfall that set new rainfall records, causing flash floods in the mountainous north and riverine floods downstream, primarily affecting the riverbank districts such as Swat, Charsadda, Nowshera, and DI Khan. The floods persisted for several weeks in KP (late July to mid-August) in 2010. Conversely, the 2022 floods were characterized by a series of excessive rainfall events distributed across the province. Districts in the north, center, and south were affected, including Swat, Shangla, Dir, Peshawar, Charsadda, Swabi, DI Khan, Tank, and Lakki Marwat, among others. The established disaster management reporting regime recorded twenty-four flood events that occurred over a three-month period from June to August (Gul et al., 2024; P&DD, Planning and Development Department, 2022; PDMA, 2023). This temporal complexity contributed to repeated stress on settlements, infrastructure, and services, even in areas where inundation depths and extents were comparatively lower.

The varied geographic and temporal distribution of the 2022 floods, characterized by their prolonged, repetitive, and non-simultaneous nature, posed different challenges to the provincial response capacity compared to 2010. These shifts highlight the dynamic nature of flood risk in KP and caution against assuming static vulnerability based solely on historical flood footprints.

The affected population increased by 14.5% in 2022 compared to 2010, while a smaller proportion of the population was displaced in 2022 compared to 2010, as stated in the earlier section. The difference in damage patterns were also direct results of the specificity of the floods type. For example, the water supply and sanitation systems are effected differently in 2010 compared to 2022 mainly because the former were predominately flash and riverine floods affecting some of the mountainous and riverbank districts, while the latter is episodic multiple events of varying nature of floods including GLOF covering increased number of districts not necessarily along rivers. Overall, the comparison of temporal and geographic variability reveals that while structural flood extent was reduced in 2022, exposure became more dispersed and recurrent over time. This finding helps explain why the 2022 floods generated significant social and economic disruption despite lower overall inundation. This underscores the importance of hazard-specific resilience measures rather than generic ones. Therefore, resilience planning should account for all flood types and differently. Rather than investing in strengthening embankments and levees against riverine floods, other types should be comprehensively included for preparedness and mitigation strategies.

Jaccard analysis was conducted across 11 districts and nine sectors, with a focus on the quantitative assessment of infrastructure damage, categorized as complete, partial, or by type. The Jaccard analysis revealed varying degrees of similarity in terms of damage across corresponding sectors in similar geographic units (districts) between the two floods. Among the nine sectors analyzed, two exhibited high similarity, one demonstrated moderate similarity, five showed less similarity, and one displayed no similarity for the two floods. With an arithmetic mean of JIs for all nine sectors of 0.3, it can be cautiously inferred that the floods exhibit limited similarity to one another.

The JI results indicate that highly similar damage occurred in both the irrigation and flood management and governance sectors across the corresponding selected districts. While the cost of damages in flood management and irrigation in 2022 were minimal compared to those in 2010, the districts exhibited similar characteristics in both events showing related damage extents, normalized on a scale from 1 to 10. This suggests a comparable persistent vulnerability of the structures across the districts despite interventions and despite less damage.

The housing sector experienced moderately similar levels of damage across the corresponding districts during both flood events on the JI, indicating that certain districts were affected in both instances, while others were not. For instance, the districts of Swat and Nowshera suffered significant losses in both floods, whereas the district of DI Khan was the most severely affected in 2022, in contrast to the districts of Dir and Charsadda in 2010.

The five sectors, health, education, industry and commerce, agriculture, and roads and bridges, exhibited less similarity in damage across the 11 districts. This implies that spatial distribution is different in both floods in these sectors. Importantly, this may also imply that the pattern of vulnerability may have changed with improvements in some respects while exposing new weaknesses. This shift could have also been influenced by the nature of floods, 2010’s singular large wave versus the 2022’s multiple events.

The water supply and sanitation sectors showed no similarity in damage across the 11 districts when comparing flood events. This indicates that the damage patterns were different in both floods across similar districts. Using the general metrics results, it can be inferred that water supply and sanitation became a major victim in the floods 2022 and were impacted where they were not previously. The Jaccard index results are summarized in Figure 7.

The overall depiction of limited similarity, with an average Jaccard Index (JI) of 0.3, indicates that the districts were predominantly affected in diverse ways. Furthermore, a lower JI does not necessarily signify improvement in a sector; rather, it may denote change. In other words, while it is possible to identify the usual factors, it is imperative to remain vigilant regarding emerging risks in various locations.

The Noy equation is presented once more below, consisting of three components: L i f e y e a r s = L   M , A d e a t h , A exp + I   N + DAM  Y , P = M   *     A exp   –   A med   + e * T * N + 1 − c   * Y /   P C G D P

The inputs for the various variables and sources are delineated as follows:

The mortality rate, M, from this research is 1,156 lives lost in 2010, and 419 lives lost in 2022. Noy used the World Health Organization (WHO) anticipated estimates of 92 years for life expectancy (A ^exp ) at birth. This figure is derived from WHO projections, which is generic across nations and theoretically upper bound. However, we utilize the corresponding annual data from the Economic Survey of Pakistan, which is actual national value and reflects Pakistan’s demography. According to the Economic Survey of Pakistan 2010–11, life expectancy, A ^exp , at birth in 2010 was 67.2 years, whereas the Economic Survey of Pakistan 2023–24 indicates it was 67.3 years in 2022. United Nations population datasheets for Pakistan corresponding to flood years report that the median age, A ^med , was 19 years in 2010 and 20 years in 2022 (Finance Division, 2010; Finance Division, 2023). The coefficient e, referred to as the 'welfare-reduction weight,’ is taken as 0.054, and variable T, denoting the time required for the population to return to normalcy after disaster, is considered 3, as used in relevant studies (Doan and Noy, 2021; Noy, 2015). Variable N, representing the number of individuals affected and from this research it is 3,800,000 for 2010 and 4,350,000 for 2022. Noy utilizes a discounting rate, c, of 75 percent, regarded as a conservative estimate, resulting in (1-c) being 0.25. The variable Y represents financial damage, encompassing the value of destroyed or damaged infrastructure, but not the replacement costs. In this study, Y is taken as Rs. 105,896 million (US$1,246,000,000) for 2010 and Rs. 201,000 million (US$935,000,000) for 2022 as per corresponding PDNAs. The per capita GDP (PCGDP) was US$1,013 in 2010, with an exchange rate of Rs 80 to a dollar (MoPDSI, 2010) and US$1,766 in 2022, with an exchange rate of Rs 204.85 per dollar (MoPDSI, 2022). A summary of the inputs for the various variables is presented in Table 3.

By inputting all the relevant variable values into the Noy-lifeyears equation, the total life years lost were calculated for both flood events. The results indicate that the 2010 floods resulted in an estimated 7,134,226.60 life years lost, which is approximately 7.134 million years. In comparison, the 2022 floods accounted for a slightly higher loss, totaling 7,463,902.51 life years, or approximately 7.463 million years.

The effect of each of the three constituent components in the Noy Equation is analyzed further to assess the mathematical contribution of each component to the final results for the corresponding flood year. The L-Component (Mortality) accounted for 55,719 (0.055 million) life years in 2010, whereas in 2022, it accounted for 19,818 (0.019 million) life years, reflecting a reduction of 0.036 million years. This reduction in mortality resulted in the preservation of 35,900 life years in 2022 compared to 2010. These values represent approximately 1% of the total life years for both years, indicating a relatively minor impact on the overall disaster effect. The I-Component (Injury/Displacement) recorded 6.156 million life years in 2010, increasing to 7.047 million life years in 2022, an increment of 891,000 life years attributable to a greater number of individuals affected. In 2010, this component accounted for 86% of the total lifeyears, increasing to 94% in 2022, highlighting the significant role of the number of individuals affected by the disaster as a key factor in human impact. The DAM-Component (Economic Damage) was 0.922 million life years in 2010, decreasing to 0.397 million life years in 2022, a reduction of 525,423 life years. The DAM component was nearly double in 2010 compared to 2022, indicating greater actual damage. Despite the higher nominal values of economic damage in 2022, the lower DAM value suggests that when adjusted for a larger economy, the human efforts and hence the lifeyears required for normalization and rebuilding were reduced. The DAM value contributed 13% in 2010 and 5% in 2022 to the total Noy Lifeyears losses.

The increase in the total lifeyears value for 2022 was primarily attributed to the elevated I component, which dominated the reductions observed in the L and DAM components. This suggests that, despite experiencing reduced mortality and lesser true damage, the resilience efforts were overshadowed by the disruption to the lives of a significantly greater number of population. Furthermore, the findings indicate that although the 2010 flood events exhibited a clear pattern of increased damage, the Noy-lifeyears index presented a contrasting narrative. This discrepancy creates a paradox concerning the linearity and proportionality of damage figures in relation to life costs and vulnerability. It also distinguishes structural resilience from human coping capacity and disconnects from enhanced social protection measures. The resilience gap becomes evident where traditional metrics may suggest improvement, yet in reality, there is a setback or stagnation. The Noy-lifeyears index is graphically presented in Figure 8.

This also poses a challenge to the improvements brought about by the implementation of the proactive disaster management regime post-2010. Although structural resilience has improved in certain areas, the overall gains and benefits to the primary stakeholders, namely, the people for whom the entire process and mechanism were established, are in fact diminished, considering all external and internal parameters.

In conclusion the floods 2022 were more costly in terms of human development than floods 2010. The total lifeyears lost is 4.6% higher indicating an inclusion of human centric metrics in disaster impact assessments.

Disasters should be understood as cumulative social phenomena rather than isolated disruptive incidents that destroy infrastructure. Metrics that focus solely on tangible damage are inadequate for capturing the profound disruptions to social structures and human fabric. This study critically examines the linear narrative of development and resilience prevalent in international policy discourse by employing the Noy index, which elucidates the impact dichotomy of flood events when assessed only with the direct perspective of damage figures. The Noy-lifeyears index, showing an increase of 4.6% in lifeyears lost in 2022, clearly augments the argument. The comparative findings in the study revealed that disaster resilience progress may be uneven and, in fact, illusory in KP. With the implementation of the new disaster management paradigm, there may be modest improvement when viewed from the damage perspective; however, beyond metrics, it reveals continuity and even worsening of the effects.

Another aspect that becomes evident is that while some sectors perform well, others may have been subject to neglect. For example, investment in roads, bridges, and irrigation infrastructure has resulted in resilience because it was easier to prioritize due to its visibility for political gain and ease of funding. The case is not the same for housing and health, which apart from immediate steps require long-term strategies like improving building codes, devising and implementing land use planning, and above all the will to enforce non-visible long-term resilience strategies which may not be politically popular. Extending the insight, this hints at the bureaucratic and political capacity and will to take the required action. A perception is created through plans, meetings, reports, and politically bankable projects of roads, bridges, and infrastructure or immediate reconstruction in these sectors, creating a false sense of security and progress, while vulnerabilities remain on the ground. This can be contextualized with theoretical concepts of necropolitics (Mbembe, 2003) and bureaucratic illusion (Peters and Nagel, 2025), which can be deduced from present study where social vulnerability in 2022 was evident from the results of this study. These theoretical lenses help explain the observed disparity: visible, politically rewarding infrastructure (roads, bridges) received investment that reduced physical damage (high JI similarity), while less visible, long-term social infrastructure (housing, health, water supply and sanitation) were neglected, leading to increased human suffering (higher Noy index). This pattern reflects a form of bureaucratic prioritization that overlooks human security.

Alongside other reasons, floods do have a role in the increase in poverty (5.9% more poverty in 2022 than 2.2% after floods in 2010) and the plunging of the economy (GDP) into negative growth (−6.4%) after floods in 2022. This results in driving people to lesser expenditure on food and health, movement to cheaper marginal lands for lodging, sustained trauma and mental health issues, interruption or cessation of education for children, and loss of livelihoods. Moreover, vulnerable communities and segments, including rural populations, ethnic minorities, individuals living below the poverty line, and those who are physically vulnerable, such as the elderly, disabled, women, and children, are often the primary victims. Although this study does not explicitly quantify these, they are associated intangible losses with disasters. Yet, it hints at these deficiencies in the system and the need to look beyond the metrics of monetary losses. This may again be attributed to necropolitics, where authorities exercise control over mortality through action or inaction. Thus, not striving to review the paradigm can be seen as inaction on the part of the authorities.

Beyond metrics, this study offers insights into shifting the emphasis of future research and practices from observable impacts and quantitative metrics to a broader system that cultivates resilience, encompassing governance, development, and human security. Consequently, disaster recovery extends beyond economic rehabilitation to encompass the intricate restoration of individual and collective temporality. The concept of resilience remains superficial without addressing ontological vulnerabilities.