Yin Tsarin Ambaliyar Ruwa ga Biranen ta Amfani da Girgizar Wutar Lantarki: Tsari da Nazarin Misali

1. Gabatarwa & Bayyani

Wannan takarda tana magance wata matsala mai mahimmanci a kimiyyar muhalli da tsara birane: ƙarfin lissafi na ingantaccen tsarin haɗarin ambaliyar ruwa. Ƙungiyoyi kamar gwamnatocin gida, kamfanonin injiniya, da masu inshora suna fuskantar buƙatu na doka da ƙwararru don ingantaccen hasashen ambaliyar ruwa amma galibi suna rasa albarkatun lissafi masu ƙarfi da dorewa. Marubutan sun ba da shawara kuma sun nuna mafita mai amfani: amfani da Girgizar Wutar Lantarki a matsayin Sabis (IaaS) don aiwatar da binciken "CityCat" na software na ambaliyar ruwa na birane. Wannan hanyar tana ba da damar samun babban ƙarfin lissafi bisa tsarin biyan kuɗi, yana ba da damar yin siminti a ma'aunin birane wanda ba zai yiwu ba tare da na'urar gida, don ayyuka na lokaci-lokaci.

2. Tsarin Tsakiya & Hanyoyin Aiki

2.1. Kalubalen Bincike Mai Zurfi

Yin tsarin ambaliyar ruwa a ƙarƙashin rashin tabbas yana buƙatar gudanar da siminti da yawa tare da ma'auni daban-daban na shigarwa (misali, ƙarfin ruwan sama, tsawon lokaci, shigar ƙasa). Wannan "bincike mai zurfi" aiki ne mai sauƙi amma ya zama mai tsada a ma'aunin birane. Shinge na al'ada sun haɗa da babban kashe kuɗi don tarin HPC da ƙwarewar fasaha da ake buƙata don rarraba lissafi.

2.2. Tsarin Aiwatarwa na Girgizar Wutar Lantarki

Marubutan sun ƙirƙira wani tsari don rage rikitarwar tura Girgizar Wutar Lantarki. Muhimman abubuwan sun haɗa da:

• Mai Samar da Ayyuka: Yana ƙirƙira ayyukan siminti masu zaman kansu ga kowane saitin ma'auni.
• Mai Samar da Albarkatu: Yana sarrafa haifuwar na'urori masu ƙwaƙwalwa (VMs) akan Girgizar Wutar Lantarki (misali, Amazon EC2, OpenStack).
• Mai Shirya Ayyuka & Mai Tura Ayyuka: Yana sarrafa rarraba ayyuka a cikin tafkin VM.
• Mai Tattara Bayanai: Yana tattarawa da haɗa sakamako daga duk simintin da aka kammala.

Wannan bututun yana canza matsalar siminti guda ɗaya zuwa aiki mai sarrafawa, mai faɗaɗawa.

3. Aiwatar Fasaha & Cikakkun Bayanai

3.1. Tsarin Lissafi: CityCat

Babban injin siminti, CityCat, yana warware daidaitattun ruwa maras zurfi (SWEs), saitin daidaitattun bambance-bambancen sassa waɗanda ke sarrafa kwararar saman kyauta:

$\frac{\partial \mathbf{U}}{\partial t} + \frac{\partial \mathbf{F}(\mathbf{U})}{\partial x} + \frac{\partial \mathbf{G}(\mathbf{U})}{\partial y} = \mathbf{S}(\mathbf{U})$

inda $\mathbf{U} = [h, hu, hv]^T$ shine vector na ma'auni da aka kiyaye (zurfin ruwa $h$, da fitar da raka'a $hu$, $hv$). $\mathbf{F}$ da $\mathbf{G}$ sune vectors na juzu'i, kuma $\mathbf{S}$ yana wakiltar sharuɗɗan tushe/ɓarna kamar gogayyar gado da ruwan sama. Binciken yana bambanta abubuwan shigarwa zuwa $\mathbf{S}$ da yanayin farko/iyaka.

3.2. Shirye-shiryen Aiki

Binciken mai yiwuwa ya yi amfani da kayan aikin aiki kamar Apache Airflow ko HTCondor, waɗanda aka daidaita don yanayin Girgizar Wutar Lantarki. Tsarin shine: 1) Ayyana sararin ma'auni; 2) Kunna CityCat da abubuwan da ya dogara da shi cikin hoton VM ko akwati; 3) Samar da tarin VMs; 4) Ai watar ayyuka; 5) Dakatar da albarkatu bayan kammalawa don rage farashi.

4. Sakamakon Gwaji & Aiki

Aiwatar girgizar wutar lantarki ta sami matsi mai yawa na lokacin "bango-agogo". Takardar ta ba da rahoton kammala kusan watanni 21 na daidaitaccen sarrafa jeri a cikin wata ɗaya na kalanda ta hanyar amfani da albarkatun Girgizar Wutar Lantarki. Wannan ya ba da damar yin nazarin haɗari na dukan birane wanda ba zai yiwu ba a baya. Muhimman ma'auni na aiki za su haɗa da:

• Haɓakawa: Kusan ma'auni mai layi tare da adadin misalan VM don bincike mai sauƙi.
• Ingancin Farashi: An kwatanta jimillar farashin haya na girgiza da babban kashe kuɗi (CapEx) na sayan daidaitaccen kayan aikin gida, musamman idan aka yi la'akari da yanayin amfani na lokaci-lokaci.
• Fitowa: Samar da taswirorin haɗarin ambaliyar ruwa mai ƙima, yana nuna zurfi da sauri a cikin birane don yanayin guguwa da yawa.

Bayanin Chati (An nuna): Chati na sandar zai nuna "Lokacin Siminti" akan y-axis (a cikin watanni) da "Hanyar Lissafi" akan x-axis. Babban sandar da aka yiwa lakabin "Aiwatar da Jerin Gida" zai kai ~watanni 21. Ƙaramin sandar da aka yiwa lakabin "Aiwatar da Girgizar Wutar Lantarki" zai kai ~wata ɗaya, yana nuna matsawar lokaci sosai.

5. Tsarin Nazari & Misalin Misali

Tsarin: Matrix na yanke shawara na Farashi-Amfani na Girgizar Wutar Lantarki don Lissafin Kimiyya

Yanayi: Sashen tsara birane yana buƙatar gudanar da siminti na ambaliyar ruwa 10,000 don sabon tsarin yanki a cikin makonni 4.

1. Siffanta Aiki: Shin yana da sauƙi? (Ee). Menene buƙatun ƙwaƙwalwar ajiya/CPU na kowane aiki? (Matsakaici). Canja wurin bayanai matsala ce? (Mai yuwuwa ga sakamako).
2. Kimanta Zaɓuɓɓuka:
   • Zaɓi A (Tarin Gida): CapEx: $50,000. Lokacin jagora: watanni 3. Lokacin gudu: makonni 8. Hukunci: Ya kasa karewa.
   • Zaɓi B (Fashewar Girgizar Wutar Lantarki): OpEx: ~$5,000. Lokacin jagora: rana 1. Lokacin gudu: mako 1 (ma'auni zuwa VMs 500). Hukunci: Ya cika wa'adin, ƙananan farashin farko.
3. Direban yanke shawara: Ƙimar lokaci na sakamako. Idan yanke shawarin yanki yana da tasirin tattalin arziki na miliyoyin daloli, saurin girgizar wutar lantarki yana ba da hujjar farashinsa, ko da an maimaita shi kowace shekara. Idan binciken ilimi ne ɗaya, ƙarfin farashi ya fi girma.

Wannan tsarin ya wuce kwatankwacin farashi mai sauƙi don haɗawa da lokaci-zuwa-magani da farashin dama, yana daidaitawa da fifikon takarda akan ƙayyadaddun wa'adin.

6. Nazari Mai mahimmanci & Hikimar Kwararru

Hikimar Tsakiya: Wannan takarda ba game da sabon tsarin ambaliyar ruwa ba ce; darasi ne mai zurfi a cikin tattalin arzikin lissafi da aka yi amfani da shi. Ya gano daidai cewa ga ƙungiyoyi da yawa, babban abin da ke hana su ba algorithm ba ne, amma tsarin samun damar lissafi. Ainihin ƙirƙira shine kayan gini wanda ke rage shingen fasaha, yana sa IaaS ta zama mai amfani ga masana kimiyyar yanki.

Kwararar Ma'ana: Hujjar tana da ƙarfi: 1) Matsala: Ana buƙatar babban lissafi na ɗan gajeren lokaci. 2) Magani: Tsarin girgizar wutar lantarki mai sassauƙa, biyan kuɗi. 3) Shinge: Rikitarwar fasaha na tsarin rarraba. 4) Aiwatarwa: Gina Layer na abstraction (tsarin su). 5) Tabbatarwa: Nuna lokaci/arha akan matsala ta gaske, mai tasiri (ambaliyar ruwa ma'aunin birane). Kwararar daga sharuddan tattalin arziki zuwa maganin fasaha zuwa ƙididdigewa sakamako yana da kyau.

Ƙarfi & Kurakurai:
Ƙarfi: Takardar tana da amfani sosai. Tana magance gibin karɓuwa na duniya ta gaske. Matsi na 21:1 sakamako ne mai kashewa. Yana hasashen "babu haɗin gwiwa" zargi na amfani da girgizar wutar lantarki kuma ya karyata shi daidai ga ayyukan lokaci-lokaci—mahimmin fahimtar kuɗi wanda masana fasaha suka yi kuskure.
Kurakurai: Giwa a cikin daki shine nauyin bayanai. Takardar ta taɓa canja wurin bayanai kaɗan amma ta ƙi ƙima tasirinsa na dabaru da farashi don ma'aunin bayanan sararin samaniya na petabyte. Matsar da terabytes na bayanan LIDAR zuwa da daga girgizar wutar lantarki na iya soke arha lissafi. Na biyu, an gabatar da tsarin a matsayin mafita ta musamman. A yau, muna buƙatar kimanta da dandamalin marasa sabis (AWS Lambda, Google Cloud Run) don sarrafa farashi mai kyau, ko sabis na batch da aka sarrafa (AWS Batch, Azure Batch) waɗanda tun daga lokacin suka fito don warware wannan matsala da kyau.

Hanyoyin Aiki:
1. Ga Masu Bincike: Yi amfani da sarrafa farashin girgizar wutar lantarki a matsayin babban ƙwarewar bincike. Yi amfani da misalan tabo/VM masu fifiko; da sun kashe farashin da aka bayar da kashi 60-80%. Kayan aiki kamar Kubernetes don shirye-shiryen akwati yanzu shine daidaitaccen Layer, ba rubutun al'ada ba.
2. Ga Masana'antu: Samfurin a nan yana iya maimaitawa ga kowane bincike (CFD, gano magunguna, Monte Carlo kuɗi). Harkar kasuwanci dole ne ta juya daga CapEx vs. OpEx zuwa "ƙimar haɓakar fahimta." Nawa ne samun taswirorin ambaliyar ruwa watanni 20 da suka gabata ya kamata ga mai inshora? Biliyoyin a cikin daidaita haɗari.
3. Ga Masu Samar da Girgizar Wutar Lantarki: Wannan takarda tsari ne don tallan ku na "democratization HPC". Haɓaka ƙarin samfura na musamman na yanki ("Yin Tsarin Ambaliyar Ruwa akan AWS") waɗanda ke haɗa bayanai, samfuri, da aiki, yana rage lokacin saiti daga makonni zuwa sa'o'i.

Aikin marubutan ya riga ya yi hasashen tsarin zamani na "kimiyya a matsayin sabis". Duk da haka, kwatanta shi da wani sabon abu na zamani kamar takardar CycleGAN (Zhu et al., 2017) yana da koyarwa. Dukansu suna rage shinge: CycleGAN ta kawar da buƙatar haɗaɗɗun bayanan horo, yana ba da damar fassarar hoto zuwa hoto. Wannan tsarin yin tsarin ambaliyar ruwa yana kawar da buƙatar cibiyar HPC ta keɓaɓɓu, yana ba da damar yin siminti mai girma. Nan gaba yana cikin haɗa waɗannan abubuwan: yin amfani da AI mai samuwa, mai samuwa (kamar GANs) don rage girman bayanan yanayi ko samar da ƙasa na roba, wanda sai ya shiga cikin samfuran jiki na girgizar wutar lantarki kamar CityCat, yana haifar da zagaye mai kyau na hasashen muhalli mai inganci, mai samuwa.

7. Aikace-aikacen Gaba & Jagorori

Hanyar da aka fara a nan tana da fa'ida mai faɗi:

• Nazarin Haɗarin Yanayi: Gudanar da ƙungiyoyin samfuran yanayi na yanki (RCMs) a ƙarƙashin ɗaruruwan yanayin fitarwa ga bankuna da masu sarrafa kadarori, kamar yadda aka gani a cikin aikin ƙungiyoyi kamar ClimateAI ko EU's Copernicus Climate Change Service.
• Tagwayen Digital don Biranen: Ƙirƙirar kwafin abubuwan more rayuwa na birane masu rai, masu kwaikwayo. Dandamalin girgizar wutar lantarki suna da mahimmanci don ci gaba da gudanar da siminti don zirga-zirga, grid ɗin makamashi, kuma a'a, magudanar ruwa, a matsayin wani ɓangare na allon juriya da aka haɗa.
• Haɗin AI/Physics Modelling: Iyakar gaba. Yi amfani da albarkatun girgizar wutar lantarki don horar da mai kwaikwayon koyo mai zurfi (samfurin maye) na tsada CityCat siminti. Da zarar an horar da shi, mai kwaikwayon zai iya samar da hasashe nan take, kusan, tare da cikakken samfurin da ake kira kawai don yanayi mai mahimmanci. Wannan tsarin "maye-a-kan-girgiza, horo-a-kan-girgiza" yana fitowa a cikin ayyukan da aka ambata akan arXiv (misali, a cikin hanyoyin sadarwar jijiyoyi masu ilimin kimiyyar lissafi).
• Shugabanci: Nan gaba ba kawai IaaS ba ne, amma Dandamali a matsayin Sabis (PaaS) da marasa sabis don ayyukan aikin kimiyya. Manufar ita ce matsar daga sarrafa VMs zuwa kawai gabatar da akwatin Docker da fayil ɗin ma'auni, tare da sabis ɗin girgizar wutar lantarki yana sarrafa komai—ma'auni, shirye-shirye, da ingantaccen farashi. Wannan yana wakiltar mataki na ƙarshe na rage shingen fasaha da takardar ta gano.

8. Nassoshi

1. Glenis, V., McGough, A.S., Kutija, V., Kilsby, C., & Woodman, S. (2013). Flood modelling for cities using Cloud computing. Journal of Cloud Computing: Advances, Systems and Applications, 2(1), 7.
2. Mell, P., & Grance, T. (2011). The NIST Definition of Cloud Computing. National Institute of Standards and Technology, SP 800-145.
3. Zhu, J., Park, T., Isola, P., & Efros, A.A. (2017). Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks. IEEE International Conference on Computer Vision (ICCV).
4. Armbrust, M., Fox, A., Griffith, R., et al. (2010). A view of cloud computing. Communications of the ACM, 53(4), 50-58.
5. European Centre for Medium-Range Weather Forecasts (ECMWF). Copernicus Climate Change Service (C3S). Retrieved from https://climate.copernicus.eu
6. Raissi, M., Perdikaris, P., & Karniadakis, G.E. (2019). Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics, 378, 686-707.